GNU bug report logs - #33600
[PATCH 0/3] Defaulting to ci.guix.info (aka. berlin.guixsd.org)

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 33600 in the body.
You can then email your comments to 33600 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: guix-patches <at> gnu.org
Cc: guix-devel <at> gnu.org, Ludovic Courtès <ludo <at> gnu.org>
Subject: [PATCH 0/3] Defaulting to ci.guix.info (aka. berlin.guixsd.org)
Date: Mon,  3 Dec 2018 16:43:35 +0100

Hello Guix!

These patches (actually the last one) switch Guix to default to
<https://ci.guix.info> for substitutes, in preparation for the
upcoming 0.16.0 release (hopefully this week!).

Rationale:

  • berlin.guixsd.org generally performs better than hydra.gnu.org;

  • berlin supports x86, ARMv7, and AArch64 (hydra lacks AArch64);

  • berlin has much more storage space than hydra;

  • berlin runs Cuirass on GuixSD, so software-wise, it corresponds
    to what we want to push; Cuirass still has limitations in some
    ways compared to Hydra, but we’re getting there, and
    hydra.gnu.org is still up and running so we can still use it,
    e.g., to track build progress until Cuirass is as convenient.

For the domain name I initially wanted “ci.guix.gnu.org” but we failed
to set that up.  Oh well, I think that’s OK.

This change modifies the manual.  Translations will thus be stale;
Julien, do you think it’d be reasonable to push a new pre-release
to the Translation Project?  Or should be just live with the
slight inaccuracy?

Thanks,
Ludo’.

Ludovic Courtès (3):
  etc: Add "ci.guix.info.pub" public key file.
  Remove most references to hydra.gnu.org.
  build: Default to https://ci.guix.info for substitutes.

 Makefile.am                                   |  5 +-
 build-aux/check-available-binaries.scm        |  4 +-
 .../check-final-inputs-self-contained.scm     |  2 +-
 config-daemon.ac                              | 10 +--
 doc/contributing.texi                         |  2 +-
 doc/guix.texi                                 | 67 +++++++++----------
 etc/substitutes/ci.guix.info.pub              |  1 +
 gnu/services/base.scm                         |  4 +-
 gnu/system/install.scm                        |  2 +-
 guix/scripts/build.scm                        |  2 +-
 guix/scripts/size.scm                         |  2 +-
 guix/scripts/substitute.scm                   |  2 +-
 guix/self.scm                                 |  3 +
 guix/store.scm                                |  2 +-
 14 files changed, 51 insertions(+), 57 deletions(-)
 create mode 120000 etc/substitutes/ci.guix.info.pub

-- 
2.19.2

Message #8 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: 33600 <at> debbugs.gnu.org
Cc: guix-devel <at> gnu.org, Ludovic Courtès <ludo <at> gnu.org>
Subject: [PATCH 3/3] build: Default to https://ci.guix.info for substitutes.
Date: Mon,  3 Dec 2018 16:53:35 +0100

* config-daemon.ac (guix_substitute_urls): Always default to
"https://ci.guix.info".
* doc/guix.texi (SUBSTITUTE-SERVER): Switch to ci.guix.info.
* guix/scripts/build.scm (%default-log-urls): Likewise.
* guix/scripts/substitute.scm (%default-substitute-urls): Likewise.
* guix/store.scm (%default-substitute-urls): Likewise.
---
 config-daemon.ac            | 10 +---------
 doc/guix.texi               | 13 ++++---------
 guix/scripts/build.scm      |  2 +-
 guix/scripts/substitute.scm |  2 +-
 guix/store.scm              |  2 +-
 5 files changed, 8 insertions(+), 21 deletions(-)

diff --git a/config-daemon.ac b/config-daemon.ac
index aac46817f5..e5b0ee082c 100644
--- a/config-daemon.ac
+++ b/config-daemon.ac
@@ -120,15 +120,7 @@ if test "x$guix_build_daemon" = "xyes"; then
 
   dnl Determine the appropriate default list of substitute URLs (GnuTLS
   dnl is required so we can default to 'https'.)
-  case "x$host_cpu" in
-    xaarch64)
-      # Currently only berlin.guixsd.org provides aarch64 binaries.
-      guix_substitute_urls="https://berlin.guixsd.org"
-      ;;
-    *)
-      guix_substitute_urls="https://mirror.hydra.gnu.org"
-      ;;
-  esac
+  guix_substitute_urls="https://ci.guix.info"
 
   AC_MSG_CHECKING([for default substitute URLs])
   AC_MSG_RESULT([$guix_substitute_urls])
diff --git a/doc/guix.texi b/doc/guix.texi
index d96c35eace..a5ae1c65e6 100644
--- a/doc/guix.texi
+++ b/doc/guix.texi
@@ -14,7 +14,7 @@
 @set KEY-SERVER pool.sks-keyservers.net
 
 @c The official substitute server used by default.
-@set SUBSTITUTE-SERVER ci.guix.info
+@set SUBSTITUTE-SERVER mirror.hydra.gnu.org
 
 @copying
 Copyright @copyright{} 2012, 2013, 2014, 2015, 2016, 2017, 2018 Ludovic Courtès@*
@@ -2374,14 +2374,9 @@ Then, you can run something like this:
 @end example
 
 @quotation Note
-Similarly, the @file{berlin.guixsd.org.pub} file contains the public key
-for the project's new build farm, reachable at
-@indicateurl{https://berlin.guixsd.org}.
-
-As of this writing @code{berlin.guixsd.org} is being upgraded so it can
-better scale up, but you might want to give it a try.  It is backed by
-20 x86_64/i686 build nodes and may be able to provide substitutes more
-quickly than @code{mirror.hydra.gnu.org}.
+Similarly, the @file{hydra.gnu.org.pub} file contains the public key
+of an independent build farm also run by the project, reachable at
+@indicateurl{https://mirror.hydra.gnu.org}.
 @end quotation
 
 Once this is in place, the output of a command like @code{guix build}
diff --git a/guix/scripts/build.scm b/guix/scripts/build.scm
index 5532c65eb6..0b7da3189e 100644
--- a/guix/scripts/build.scm
+++ b/guix/scripts/build.scm
@@ -65,7 +65,7 @@
 
 (define %default-log-urls
   ;; Default base URLs for build logs.
-  '("http://hydra.gnu.org/log"))
+  '("http://ci.guix.info/log"))
 
 ;; XXX: The following procedure cannot be in (guix store) because of the
 ;; dependency on (guix derivations).
diff --git a/guix/scripts/substitute.scm b/guix/scripts/substitute.scm
index eb82224016..d6dc9b6448 100755
--- a/guix/scripts/substitute.scm
+++ b/guix/scripts/substitute.scm
@@ -1052,7 +1052,7 @@ found."
     (#f
      ;; This can only happen when this script is not invoked by the
      ;; daemon.
-     '("http://hydra.gnu.org"))))
+     '("http://ci.guix.info"))))
 
 (define substitute-urls
   ;; List of substitute URLs.
diff --git a/guix/store.scm b/guix/store.scm
index 9dc651b26c..509fd4def6 100644
--- a/guix/store.scm
+++ b/guix/store.scm
@@ -693,7 +693,7 @@ encoding conversion errors."
   (map (if (false-if-exception (resolve-interface '(gnutls)))
            (cut string-append "https://" <>)
            (cut string-append "http://" <>))
-       '("mirror.hydra.gnu.org")))
+       '("ci.guix.info")))
 
 (define* (set-build-options server
                             #:key keep-failed? keep-going? fallback?
-- 
2.19.2

Message #11 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: 33600 <at> debbugs.gnu.org
Cc: guix-devel <at> gnu.org, Ludovic Courtès <ludo <at> gnu.org>
Subject: [PATCH 1/3] etc: Add "ci.guix.info.pub" public key file.
Date: Mon,  3 Dec 2018 16:53:33 +0100

* etc/substitutes/ci.guix.info.pub: New file.
* Makefile.am (dist_pkgdata_DATA): Add it.
* guix/self.scm (miscellaneous-files): Add
"share/guix/ci.guix.info.pub".
---
 Makefile.am                      | 3 ++-
 etc/substitutes/ci.guix.info.pub | 1 +
 guix/self.scm                    | 3 +++
 3 files changed, 6 insertions(+), 1 deletion(-)
 create mode 120000 etc/substitutes/ci.guix.info.pub

diff --git a/Makefile.am b/Makefile.am
index 32cebd5917..6803666a57 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -459,7 +459,8 @@ check-system: $(GOBJECTS)
 # Public key used to sign substitutes from hydra.gnu.org & co.
 dist_pkgdata_DATA =				\
   etc/substitutes/hydra.gnu.org.pub		\
-  etc/substitutes/berlin.guixsd.org.pub
+  etc/substitutes/berlin.guixsd.org.pub		\
+  etc/substitutes/ci.guix.info.pub
 
 # Bash completion file.
 dist_bashcompletion_DATA = etc/completion/bash/guix	\
diff --git a/etc/substitutes/ci.guix.info.pub b/etc/substitutes/ci.guix.info.pub
new file mode 120000
index 0000000000..b1d1e4d763
--- /dev/null
+++ b/etc/substitutes/ci.guix.info.pub
@@ -0,0 +1 @@
+berlin.guixsd.org.pub
\ No newline at end of file
diff --git a/guix/self.scm b/guix/self.scm
index 8476c422ec..f2db3dbf52 100644
--- a/guix/self.scm
+++ b/guix/self.scm
@@ -428,6 +428,9 @@ load path."
                    ,(file-append* source
                                   "/etc/substitutes/hydra.gnu.org.pub"))
                   ("share/guix/berlin.guixsd.org.pub"
+                   ,(file-append* source
+                                  "/etc/substitutes/berlin.guixsd.org.pub"))
+                  ("share/guix/ci.guix.info.pub"  ;alias
                    ,(file-append* source "/etc/substitutes/berlin.guixsd.org.pub")))))
 
 (define* (whole-package name modules dependencies
-- 
2.19.2

Message #14 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: 33600 <at> debbugs.gnu.org
Cc: guix-devel <at> gnu.org, Ludovic Courtès <ludo <at> gnu.org>
Subject: [PATCH 2/3] Remove most references to hydra.gnu.org.
Date: Mon,  3 Dec 2018 16:53:34 +0100

* Makefile.am (assert-binaries-available): Don't mention hydra.gnu.org
in comment.
* build-aux/check-available-binaries.scm: Likewise.
* build-aux/check-final-inputs-self-contained.scm: Likewise.
* doc/guix.texi (SUBSTITUTE-SERVER): New variable.
Use it throughout instead of "mirror.hydra.gnu.org".
* doc/contributing.texi (Submitting Patches): Likewise.
* gnu/services/base.scm (hydra-key-authorization)
(guix-activation): Remove mentions of "hydra.gnu.org" in comments and
messages.
* gnu/system/install.scm (%installation-services): Likewise.
* guix/scripts/size.scm (guix-size): Likewise.
---
 Makefile.am                                   |  2 +-
 build-aux/check-available-binaries.scm        |  4 +-
 .../check-final-inputs-self-contained.scm     |  2 +-
 doc/contributing.texi                         |  2 +-
 doc/guix.texi                                 | 56 ++++++++++---------
 gnu/services/base.scm                         |  4 +-
 gnu/system/install.scm                        |  2 +-
 guix/scripts/size.scm                         |  2 +-
 8 files changed, 38 insertions(+), 36 deletions(-)

diff --git a/Makefile.am b/Makefile.am
index 6803666a57..a7a67e81cf 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -763,7 +763,7 @@ assert-no-store-file-names: $(distdir)/ChangeLog
 	  exit 1 ;								\
 	fi
 
-# Make sure hydra.gnu.org has the important binaries.
+# Make sure important substitutes are available.
 assert-binaries-available: $(GOBJECTS)
 	$(AM_V_at)$(top_builddir)/pre-inst-env "$(GUILE)"			\
 	  "$(top_srcdir)/build-aux/check-available-binaries.scm"
diff --git a/build-aux/check-available-binaries.scm b/build-aux/check-available-binaries.scm
index b832d99935..c80db1a37e 100644
--- a/build-aux/check-available-binaries.scm
+++ b/build-aux/check-available-binaries.scm
@@ -1,5 +1,5 @@
 ;;; GNU Guix --- Functional package management for GNU
-;;; Copyright © 2013, 2014, 2015, 2016, 2017 Ludovic Courtès <ludo <at> gnu.org>
+;;; Copyright © 2013, 2014, 2015, 2016, 2017, 2018 Ludovic Courtès <ludo <at> gnu.org>
 ;;;
 ;;; This file is part of GNU Guix.
 ;;;
@@ -17,7 +17,7 @@
 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
 
 ;;;
-;;; Check whether important binaries are available at hydra.gnu.org.
+;;; Check whether important binaries are available.
 ;;;
 
 (use-modules (guix store)
diff --git a/build-aux/check-final-inputs-self-contained.scm b/build-aux/check-final-inputs-self-contained.scm
index 37dc883d3c..c819086508 100644
--- a/build-aux/check-final-inputs-self-contained.scm
+++ b/build-aux/check-final-inputs-self-contained.scm
@@ -17,7 +17,7 @@
 ;;; along with GNU Guix.  If not, see <http://www.gnu.org/licenses/>.
 
 ;;;
-;;; Check whether important binaries are available at hydra.gnu.org.
+;;; Check whether important binaries are available.
 ;;;
 
 (use-modules (guix store)
diff --git a/doc/contributing.texi b/doc/contributing.texi
index 7417f28d14..c55eb63382 100644
--- a/doc/contributing.texi
+++ b/doc/contributing.texi
@@ -443,7 +443,7 @@ as timestamps or randomly-generated output in the build result.
 
 Another option is to use @command{guix challenge} (@pxref{Invoking guix
 challenge}).  You may run it once the package has been committed and
-built by @code{hydra.gnu.org} to check whether it obtains the same
+built by @code{@value{SUBSTITUTE-SERVER}} to check whether it obtains the same
 result as you did.  Better yet: Find another machine that can build it
 and run @command{guix publish}.  Since the remote build machine is
 likely different from yours, this can catch non-determinism issues
diff --git a/doc/guix.texi b/doc/guix.texi
index b0aa5aa872..d96c35eace 100644
--- a/doc/guix.texi
+++ b/doc/guix.texi
@@ -13,6 +13,9 @@
 @set OPENPGP-SIGNING-KEY-ID 3CE464558A84FDC69DB40CFB090B11993D9AEBB5
 @set KEY-SERVER pool.sks-keyservers.net
 
+@c The official substitute server used by default.
+@set SUBSTITUTE-SERVER ci.guix.info
+
 @copying
 Copyright @copyright{} 2012, 2013, 2014, 2015, 2016, 2017, 2018 Ludovic Courtès@*
 Copyright @copyright{} 2013, 2014, 2016 Andreas Enge@*
@@ -1274,8 +1277,7 @@ remote procedure call (@pxref{The Store}).
 @anchor{daemon-substitute-urls}
 Consider @var{urls} the default whitespace-separated list of substitute
 source URLs.  When this option is omitted,
-@indicateurl{https://mirror.hydra.gnu.org https://hydra.gnu.org} is used
-(@code{mirror.hydra.gnu.org} is a mirror of @code{hydra.gnu.org}).
+@indicateurl{https://@value{SUBSTITUTE-SERVER}} is used.
 
 This means that substitutes may be downloaded from @var{urls}, as long
 as they are signed by a trusted signature (@pxref{Substitutes}).
@@ -2321,7 +2323,7 @@ also result from derivation builds, can be available as substitutes.
 
 @cindex hydra
 @cindex build farm
-The @code{mirror.hydra.gnu.org} server is a front-end to an official build farm
+The @code{@value{SUBSTITUTE-SERVER}} server is a front-end to an official build farm
 that builds packages from Guix continuously for some
 architectures, and makes them available as substitutes.  This is the
 default source of substitutes; it can be overridden by passing the
@@ -2353,22 +2355,22 @@ other substitute server.
 @cindex substitutes, authorization thereof
 @cindex access control list (ACL), for substitutes
 @cindex ACL (access control list), for substitutes
-To allow Guix to download substitutes from @code{hydra.gnu.org} or a
+To allow Guix to download substitutes from @code{@value{SUBSTITUTE-SERVER}} or a
 mirror thereof, you
 must add its public key to the access control list (ACL) of archive
 imports, using the @command{guix archive} command (@pxref{Invoking guix
-archive}).  Doing so implies that you trust @code{hydra.gnu.org} to not
+archive}).  Doing so implies that you trust @code{@value{SUBSTITUTE-SERVER}} to not
 be compromised and to serve genuine substitutes.
 
-The public key for @code{hydra.gnu.org} is installed along with Guix, in
-@code{@var{prefix}/share/guix/hydra.gnu.org.pub}, where @var{prefix} is
+The public key for @code{@value{SUBSTITUTE-SERVER}} is installed along with Guix, in
+@code{@var{prefix}/share/guix/@value{SUBSTITUTE-SERVER}.pub}, where @var{prefix} is
 the installation prefix of Guix.  If you installed Guix from source,
 make sure you checked the GPG signature of
 @file{guix-@value{VERSION}.tar.gz}, which contains this public key file.
 Then, you can run something like this:
 
 @example
-# guix archive --authorize < @var{prefix}/share/guix/hydra.gnu.org.pub
+# guix archive --authorize < @var{prefix}/share/guix/@value{SUBSTITUTE-SERVER}.pub
 @end example
 
 @quotation Note
@@ -2409,7 +2411,7 @@ $ guix build emacs --dry-run
 @end example
 
 @noindent
-This indicates that substitutes from @code{hydra.gnu.org} are usable and
+This indicates that substitutes from @code{@value{SUBSTITUTE-SERVER}} are usable and
 will be downloaded, when possible, for future builds.
 
 @cindex substitutes, how to disable
@@ -2501,9 +2503,9 @@ by a server.
 Today, each individual's control over their own computing is at the
 mercy of institutions, corporations, and groups with enough power and
 determination to subvert the computing infrastructure and exploit its
-weaknesses.  While using @code{hydra.gnu.org} substitutes can be
+weaknesses.  While using @code{@value{SUBSTITUTE-SERVER}} substitutes can be
 convenient, we encourage users to also build on their own, or even run
-their own build farm, such that @code{hydra.gnu.org} is less of an
+their own build farm, such that @code{@value{SUBSTITUTE-SERVER}} is less of an
 interesting target.  One way to help is by publishing the software you
 build using @command{guix publish} so that others have one more choice
 of server to download substitutes from (@pxref{Invoking guix publish}).
@@ -3658,11 +3660,11 @@ Read a single-item archive as served by substitute servers
 low-level operation needed in only very narrow use cases; see below.
 
 For example, the following command extracts the substitute for Emacs
-served by @code{hydra.gnu.org} to @file{/tmp/emacs}:
+served by @code{@value{SUBSTITUTE-SERVER}} to @file{/tmp/emacs}:
 
 @example
 $ wget -O - \
-  https://hydra.gnu.org/nar/@dots{}-emacs-24.5 \
+  https://@value{SUBSTITUTE-SERVER}/nar/@dots{}-emacs-24.5 \
   | bunzip2 | guix archive -x /tmp/emacs
 @end example
 
@@ -6683,7 +6685,7 @@ but you are actually on an @code{x86_64} machine:
 
 @example
 $ guix build --log-file gdb -s mips64el-linux
-https://hydra.gnu.org/log/@dots{}-gdb-7.10
+https://@value{SUBSTITUTE-SERVER}/log/@dots{}-gdb-7.10
 @end example
 
 You can freely access a huge library of build logs!
@@ -8342,7 +8344,7 @@ When @command{guix publish} runs, it spawns an HTTP server which allows
 anyone with network access to obtain substitutes from it.  This means
 that any machine running Guix can also act as if it were a build farm,
 since the HTTP interface is compatible with Hydra, the software behind
-the @code{hydra.gnu.org} build farm.
+the @code{@value{SUBSTITUTE-SERVER}} build farm.
 
 For security, each substitute is signed, allowing recipients to check
 their authenticity and integrity (@pxref{Substitutes}).  Because
@@ -8579,20 +8581,20 @@ any given store item.
 The command output looks like this:
 
 @smallexample
-$ guix challenge --substitute-urls="https://hydra.gnu.org https://guix.example.org"
-updating list of substitutes from 'https://hydra.gnu.org'... 100.0%
+$ guix challenge --substitute-urls="https://@value{SUBSTITUTE-SERVER} https://guix.example.org"
+updating list of substitutes from 'https://@value{SUBSTITUTE-SERVER}'... 100.0%
 updating list of substitutes from 'https://guix.example.org'... 100.0%
 /gnu/store/@dots{}-openssl-1.0.2d contents differ:
   local hash: 0725l22r5jnzazaacncwsvp9kgf42266ayyp814v7djxs7nk963q
-  https://hydra.gnu.org/nar/@dots{}-openssl-1.0.2d: 0725l22r5jnzazaacncwsvp9kgf42266ayyp814v7djxs7nk963q
+  https://@value{SUBSTITUTE-SERVER}/nar/@dots{}-openssl-1.0.2d: 0725l22r5jnzazaacncwsvp9kgf42266ayyp814v7djxs7nk963q
   https://guix.example.org/nar/@dots{}-openssl-1.0.2d: 1zy4fmaaqcnjrzzajkdn3f5gmjk754b43qkq47llbyak9z0qjyim
 /gnu/store/@dots{}-git-2.5.0 contents differ:
   local hash: 00p3bmryhjxrhpn2gxs2fy0a15lnip05l97205pgbk5ra395hyha
-  https://hydra.gnu.org/nar/@dots{}-git-2.5.0: 069nb85bv4d4a6slrwjdy8v1cn4cwspm3kdbmyb81d6zckj3nq9f
+  https://@value{SUBSTITUTE-SERVER}/nar/@dots{}-git-2.5.0: 069nb85bv4d4a6slrwjdy8v1cn4cwspm3kdbmyb81d6zckj3nq9f
   https://guix.example.org/nar/@dots{}-git-2.5.0: 0mdqa9w1p6cmli6976v4wi0sw9r4p5prkj7lzfd1877wk11c9c73
 /gnu/store/@dots{}-pius-2.1.1 contents differ:
   local hash: 0k4v3m9z1zp8xzzizb7d8kjj72f9172xv078sq4wl73vnq9ig3ax
-  https://hydra.gnu.org/nar/@dots{}-pius-2.1.1: 0k4v3m9z1zp8xzzizb7d8kjj72f9172xv078sq4wl73vnq9ig3ax
+  https://@value{SUBSTITUTE-SERVER}/nar/@dots{}-pius-2.1.1: 0k4v3m9z1zp8xzzizb7d8kjj72f9172xv078sq4wl73vnq9ig3ax
   https://guix.example.org/nar/@dots{}-pius-2.1.1: 1cy25x1a4fzq5rk0pmvc8xhwyffnqz95h2bpvqsz2mpvlbccy0gs
 
 @dots{}
@@ -8612,7 +8614,7 @@ the servers obtained a result different from the local build.
 
 @cindex non-determinism, in package builds
 As an example, @code{guix.example.org} always gets a different answer.
-Conversely, @code{hydra.gnu.org} agrees with local builds, except in the
+Conversely, @code{@value{SUBSTITUTE-SERVER}} agrees with local builds, except in the
 case of Git.  This might indicate that the build process of Git is
 non-deterministic, meaning that its output varies as a function of
 various things that Guix does not fully control, in spite of building
@@ -8626,14 +8628,14 @@ To find out what is wrong with this Git binary, we can do something along
 these lines (@pxref{Invoking guix archive}):
 
 @example
-$ wget -q -O - https://hydra.gnu.org/nar/@dots{}-git-2.5.0 \
+$ wget -q -O - https://@value{SUBSTITUTE-SERVER}/nar/@dots{}-git-2.5.0 \
    | guix archive -x /tmp/git
 $ diff -ur --no-dereference /gnu/store/@dots{}-git.2.5.0 /tmp/git
 @end example
 
 This command shows the difference between the files resulting from the
 local build, and the files resulting from the build on
-@code{hydra.gnu.org} (@pxref{Overview, Comparing and Merging Files,,
+@code{@value{SUBSTITUTE-SERVER}} (@pxref{Overview, Comparing and Merging Files,,
 diffutils, Comparing and Merging Files}).  The @command{diff} command
 works great for text files.  When binary files differ, a better option
 is @uref{https://diffoscope.org/, Diffoscope}, a tool that helps
@@ -8648,7 +8650,7 @@ In the meantime, @command{guix challenge} is one tool to help address
 the problem.
 
 If you are writing packages for Guix, you are encouraged to check
-whether @code{hydra.gnu.org} and other substitute servers obtain the
+whether @code{@value{SUBSTITUTE-SERVER}} and other substitute servers obtain the
 same build result as you did with:
 
 @example
@@ -11238,14 +11240,14 @@ Number of build user accounts to create.
 @item @code{authorize-key?} (default: @code{#t})
 @cindex substitutes, authorization thereof
 Whether to authorize the substitute keys listed in
-@code{authorized-keys}---by default that of @code{hydra.gnu.org}
+@code{authorized-keys}---by default that of @code{@value{SUBSTITUTE-SERVER}}
 (@pxref{Substitutes}).
 
 @vindex %default-authorized-guix-keys
 @item @code{authorized-keys} (default: @var{%default-authorized-guix-keys})
 The list of authorized key files for archive imports, as a list of
 string-valued gexps (@pxref{Invoking guix archive}).  By default, it
-contains that of @code{hydra.gnu.org} (@pxref{Substitutes}).
+contains that of @code{@value{SUBSTITUTE-SERVER}} (@pxref{Substitutes}).
 
 @item @code{use-substitutes?} (default: @code{#t})
 Whether to use substitutes.
@@ -23747,7 +23749,7 @@ system}.
 @cindex substituter
 Users can obtain the new package definition simply by running
 @command{guix pull} (@pxref{Invoking guix pull}).  When
-@code{hydra.gnu.org} is done building the package, installing the
+@code{@value{SUBSTITUTE-SERVER}} is done building the package, installing the
 package automatically downloads binaries from there
 (@pxref{Substitutes}).  The only place where human intervention is
 needed is to review and apply the patch.
diff --git a/gnu/services/base.scm b/gnu/services/base.scm
index 228d3c5926..cee9898d79 100644
--- a/gnu/services/base.scm
+++ b/gnu/services/base.scm
@@ -1518,7 +1518,7 @@ public key, with GUIX."
            (let ((status (cdr (waitpid pid))))
              (unless (zero? status)
                (format (current-error-port) "warning: \
-failed to register hydra.gnu.org public key: ~a~%" status))))))))
+failed to register public key '~a': ~a~%" key status))))))))
 
 (define %default-authorized-guix-keys
   ;; List of authorized substitute keys.
@@ -1630,7 +1630,7 @@ failed to register hydra.gnu.org public key: ~a~%" status))))))))
      ;; otherwise call 'chown' here, but the problem is that on a COW overlayfs,
      ;; chown leads to an entire copy of the tree, which is a bad idea.
 
-     ;; Optionally authorize hydra.gnu.org's key.
+     ;; Optionally authorize substitute server keys.
      (if authorize-key?
          #~(begin
              #$@(map (cut hydra-key-authorization <> guix) keys))
diff --git a/gnu/system/install.scm b/gnu/system/install.scm
index 45b3a0c839..c345ba0626 100644
--- a/gnu/system/install.scm
+++ b/gnu/system/install.scm
@@ -250,7 +250,7 @@ You have been warned.  Thanks for being so brave.\x1b[0m
           ;; The usual services.
           (syslog-service)
 
-          ;; The build daemon.  Register the hydra.gnu.org key as trusted.
+          ;; The build daemon.  Register the official server keys as trusted.
           ;; This allows the installation process to use substitutes by
           ;; default.
           (guix-service (guix-configuration (authorize-key? #t)))
diff --git a/guix/scripts/size.scm b/guix/scripts/size.scm
index 344be40883..25218a2945 100644
--- a/guix/scripts/size.scm
+++ b/guix/scripts/size.scm
@@ -297,7 +297,7 @@ Report the size of PACKAGE and its dependencies.\n"))
          (leave (G_ "missing store item argument\n")))
         ((files ..1)
          (leave-on-EPIPE
-          ;; Turn off grafts because (1) hydra.gnu.org does not serve grafted
+          ;; Turn off grafts because (1) substitute servers do not serve grafted
           ;; packages, and (2) they do not make any difference on the
           ;; resulting size.
           (parameterize ((%graft? #f))
-- 
2.19.2

Message #17 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: ludo <at> gnu.org (Ludovic Courtès)
To: 33600 <at> debbugs.gnu.org
Cc: guix-devel <at> gnu.org, Chris Marusich <cmmarusich <at> gmail.com>
Subject: Using a CDN or some other mirror?
Date: Mon, 03 Dec 2018 17:12:10 +0100

Hello,

Ludovic Courtès <ludo <at> gnu.org> skribis:

> These patches (actually the last one) switch Guix to default to
> <https://ci.guix.info> for substitutes, in preparation for the
> upcoming 0.16.0 release (hopefully this week!).

Right now, ci.guix.info points to berlin.guixsd.org, the front-end of
the build farm hosted at the MDC.

The previous setup was that mirror.hydra.gnu.org mirrors hydra.gnu.org
(the actual build farm front-end) using an nginx proxy:

  https://git.savannah.gnu.org/cgit/guix/maintenance.git/tree/hydra/nginx/mirror.conf

This provides a bit of redundancy that we don’t have currently for
berlin.

Thus, I’m thinking about using a similar setup, but hosting the mirror
on some Big Corp CDN or similar.  Chris Marusich came up with a setup
along these lines a while back:

  https://lists.gnu.org/archive/html/guix-devel/2016-03/msg00312.html

Compared to Chris’s setup, given that ‘guix publish’ now provides
‘Cache-Control’ headers (that was not the case back then, see
<https://lists.gnu.org/archive/html/guix-devel/2016-03/msg00360.html>),
caching in the proxy should Just Work.

I would like us to set up such a mirror for berlin and then have
ci.guix.info point to that.  The project should be able to pay the
hosting fees.

Thoughts?

Would someone like to get started?  You’ll undoubtedly get all the
appreciation of each one of us and a beverage of your choice next time
we meet!  :-)

Thanks,
Ludo’.

Message #20 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ricardo Wurmus <rekado <at> elephly.net>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: 33600 <at> debbugs.gnu.org
Subject: Re: [PATCH 1/3] etc: Add "ci.guix.info.pub" public key file.
Date: Mon, 03 Dec 2018 21:44:45 +0100

Ludovic Courtès <ludo <at> gnu.org> writes:

> * etc/substitutes/ci.guix.info.pub: New file.
> * Makefile.am (dist_pkgdata_DATA): Add it.
> * guix/self.scm (miscellaneous-files): Add
> "share/guix/ci.guix.info.pub".

LGTM!

-- 
Ricardo

Message #23 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ricardo Wurmus <rekado <at> elephly.net>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: Using a CDN or some other mirror?
Date: Mon, 03 Dec 2018 21:47:20 +0100

Ludovic Courtès <ludo <at> gnu.org> writes:

> Hello,
>
> Ludovic Courtès <ludo <at> gnu.org> skribis:
>
>> These patches (actually the last one) switch Guix to default to
>> <https://ci.guix.info> for substitutes, in preparation for the
>> upcoming 0.16.0 release (hopefully this week!).
>
> Right now, ci.guix.info points to berlin.guixsd.org, the front-end of
> the build farm hosted at the MDC.
>
> The previous setup was that mirror.hydra.gnu.org mirrors hydra.gnu.org
> (the actual build farm front-end) using an nginx proxy:
>
>   https://git.savannah.gnu.org/cgit/guix/maintenance.git/tree/hydra/nginx/mirror.conf
>
> This provides a bit of redundancy that we don’t have currently for
> berlin.
>
> Thus, I’m thinking about using a similar setup, but hosting the mirror
> on some Big Corp CDN or similar.  Chris Marusich came up with a setup
> along these lines a while back:
>
>   https://lists.gnu.org/archive/html/guix-devel/2016-03/msg00312.html

Large ISPs also provide CDN services.  I already contacted Deutsche
Telekom so that we can compare their CDN offer with the Amazon Cloudfont
setup that Chris has configured.

--
Ricardo

Message #26 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Mark H Weaver <mhw <at> netris.org>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: guix-devel <at> gnu.org, guix-patches <at> gnu.org
Subject: Re: [PATCH 0/3] Defaulting to ci.guix.info (aka. berlin.guixsd.org)
Date: Mon, 03 Dec 2018 18:44:43 -0500

Hi Ludovic,

Ludovic Courtès <ludo <at> gnu.org> writes:

> These patches (actually the last one) switch Guix to default to
> <https://ci.guix.info> for substitutes, in preparation for the
> upcoming 0.16.0 release (hopefully this week!).

Who owns the guix.info domain?  Also, who owns the guixsd.org domain?

     Thanks,
       Mark

Message #29 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ricardo Wurmus <rekado <at> elephly.net>
To: Mark H Weaver <mhw <at> netris.org>
Cc: guix-devel <at> gnu.org, Ludovic Courtès <ludo <at> gnu.org>,
 33600 <at> debbugs.gnu.org
Subject: Re: [PATCH 0/3] Defaulting to ci.guix.info (aka. berlin.guixsd.org)
Date: Tue, 04 Dec 2018 06:55:14 +0100

Hi Mark,

> Ludovic Courtès <ludo <at> gnu.org> writes:
>
>> These patches (actually the last one) switch Guix to default to
>> <https://ci.guix.info> for substitutes, in preparation for the
>> upcoming 0.16.0 release (hopefully this week!).
>
> Who owns the guix.info domain?

I registered it and offered it to Guix Europe (though we aren’t yet
sharing administration of the domain).

--
Ricardo

Message #32 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ricardo Wurmus <rekado <at> elephly.net>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: [PATCH 2/3] Remove most references to hydra.gnu.org.
Date: Mon, 03 Dec 2018 21:44:00 +0100

Ludovic Courtès <ludo <at> gnu.org> writes:

> * Makefile.am (assert-binaries-available): Don't mention hydra.gnu.org
> in comment.
> * build-aux/check-available-binaries.scm: Likewise.
> * build-aux/check-final-inputs-self-contained.scm: Likewise.
> * doc/guix.texi (SUBSTITUTE-SERVER): New variable.
> Use it throughout instead of "mirror.hydra.gnu.org".
> * doc/contributing.texi (Submitting Patches): Likewise.
> * gnu/services/base.scm (hydra-key-authorization)
> (guix-activation): Remove mentions of "hydra.gnu.org" in comments and
> messages.
> * gnu/system/install.scm (%installation-services): Likewise.
> * guix/scripts/size.scm (guix-size): Likewise.

This looks good to me.

-- 
Ricardo

Message #35 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Andreas Enge <andreas <at> enge.fr>
To: Mark H Weaver <mhw <at> netris.org>
Cc: guix-devel <at> gnu.org, Ludovic Courtès <ludo <at> gnu.org>,
 33600 <at> debbugs.gnu.org
Subject: Re: [bug#33600] [PATCH 0/3] Defaulting to ci.guix.info (aka.
 berlin.guixsd.org)
Date: Tue, 4 Dec 2018 09:59:48 +0100

On Mon, Dec 03, 2018 at 06:44:43PM -0500, Mark H Weaver wrote:
> Who owns the guix.info domain?  Also, who owns the guixsd.org domain?

guixsd.org is owned by the Guix Europe association. guix.info, as far as
I know, by Ricardo. Plans are to drop guixsd.org in the long run, and I
would suggest to transfer guix.info to Guix Europe at the next renewal.

Andreas

Message #38 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: ludo <at> gnu.org (Ludovic Courtès)
To: Ricardo Wurmus <rekado <at> elephly.net>
Cc: guix-devel <at> gnu.org, Mark H Weaver <mhw <at> netris.org>,
 Andreas Enge <andreas <at> enge.fr>, 33600 <at> debbugs.gnu.org
Subject: Re: [bug#33600] [PATCH 0/3] Defaulting to ci.guix.info (aka.
 berlin.guixsd.org)
Date: Tue, 04 Dec 2018 10:03:01 +0100

Hello,

Ricardo Wurmus <rekado <at> elephly.net> skribis:

>> Ludovic Courtès <ludo <at> gnu.org> writes:
>>
>>> These patches (actually the last one) switch Guix to default to
>>> <https://ci.guix.info> for substitutes, in preparation for the
>>> upcoming 0.16.0 release (hopefully this week!).
>>
>> Who owns the guix.info domain?
>
> I registered it and offered it to Guix Europe (though we aren’t yet
> sharing administration of the domain).

Could you make sure several people have access to the administration
interface?

As for guixsd.org, it was initially owned by Andreas Enge who
transferred it to Guix Europe and also provided credentials to me and I
think a couple of other people, though I forgot who.  Andreas?

We should write all this information down in guix-maintenance.git like
we did for build machines, so people know who to contact in case
something goes wrong or what to do when someone leaves the project.

Thanks,
Ludo’.

Message #41 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Andreas Enge <andreas <at> enge.fr>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: Ricardo Wurmus <rekado <at> elephly.net>, Mark H Weaver <mhw <at> netris.org>,
 guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: [bug#33600] [PATCH 0/3] Defaulting to ci.guix.info (aka.
 berlin.guixsd.org)
Date: Tue, 4 Dec 2018 11:08:07 +0100

On Tue, Dec 04, 2018 at 10:03:01AM +0100, Ludovic Courtès wrote:
> As for guixsd.org, it was initially owned by Andreas Enge who
> transferred it to Guix Europe and also provided credentials to me and I
> think a couple of other people, though I forgot who.  Andreas?

Apart from you, to Ricardo Wurmus, Christopher Baines and Manolis Ragkousis.
I am of course happy to share the credentials with more people from the
Guix sysadmin group.

Andreas

Message #46 received at 33600-done <at> debbugs.gnu.org (full text, mbox):

From: ludo <at> gnu.org (Ludovic Courtès)
To: Ricardo Wurmus <rekado <at> elephly.net>
Cc: guix-devel <at> gnu.org, 33600-done <at> debbugs.gnu.org
Subject: Re: [bug#33600] [PATCH 2/3] Remove most references to hydra.gnu.org.
Date: Tue, 04 Dec 2018 11:21:59 +0100

Ricardo Wurmus <rekado <at> elephly.net> skribis:

> Ludovic Courtès <ludo <at> gnu.org> writes:
>
>> * Makefile.am (assert-binaries-available): Don't mention hydra.gnu.org
>> in comment.
>> * build-aux/check-available-binaries.scm: Likewise.
>> * build-aux/check-final-inputs-self-contained.scm: Likewise.
>> * doc/guix.texi (SUBSTITUTE-SERVER): New variable.
>> Use it throughout instead of "mirror.hydra.gnu.org".
>> * doc/contributing.texi (Submitting Patches): Likewise.
>> * gnu/services/base.scm (hydra-key-authorization)
>> (guix-activation): Remove mentions of "hydra.gnu.org" in comments and
>> messages.
>> * gnu/system/install.scm (%installation-services): Likewise.
>> * guix/scripts/size.scm (guix-size): Likewise.
>
> This looks good to me.

Great, pushed!

Ludo’.

Message #49 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: ludo <at> gnu.org (Ludovic Courtès)
To: Andreas Enge <andreas <at> enge.fr>
Cc: guix-devel <at> gnu.org, Mark H Weaver <mhw <at> netris.org>, 33600 <at> debbugs.gnu.org
Subject: Re: [bug#33600] [PATCH 0/3] Defaulting to ci.guix.info (aka.
 berlin.guixsd.org)
Date: Tue, 04 Dec 2018 11:28:51 +0100

Hello,

Andreas Enge <andreas <at> enge.fr> skribis:

> On Mon, Dec 03, 2018 at 06:44:43PM -0500, Mark H Weaver wrote:
>> Who owns the guix.info domain?  Also, who owns the guixsd.org domain?
>
> guixsd.org is owned by the Guix Europe association. guix.info, as far as
> I know, by Ricardo. Plans are to drop guixsd.org in the long run, and I
> would suggest to transfer guix.info to Guix Europe at the next renewal.

Awesome.  I’ve added a file to record that info in
guix-maintenance.git.  Please feel free to complete it; for instance
it’d be good to know the expiration date of each domain.

Thanks!

Ludo’.

Message #52 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Andreas Enge <andreas <at> enge.fr>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: guix-devel <at> gnu.org, Mark H Weaver <mhw <at> netris.org>, 33600 <at> debbugs.gnu.org
Subject: Re: [bug#33600] [PATCH 0/3] Defaulting to ci.guix.info (aka.
 berlin.guixsd.org)
Date: Tue, 4 Dec 2018 11:46:59 +0100

On Tue, Dec 04, 2018 at 11:28:51AM +0100, Ludovic Courtès wrote:
> Awesome.  I’ve added a file to record that info in
> guix-maintenance.git.  Please feel free to complete it; for instance
> it’d be good to know the expiration date of each domain.

Done for guixsd.org. It expires on January 27. I would be okay to let it
run out if guix.info is sufficiently marketed in the meantime.

Andreas

Message #55 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: ludo <at> gnu.org (Ludovic Courtès)
To: Andreas Enge <andreas <at> enge.fr>
Cc: guix-devel <at> gnu.org, Mark H Weaver <mhw <at> netris.org>, 33600 <at> debbugs.gnu.org
Subject: Re: [bug#33600] [PATCH 0/3] Defaulting to ci.guix.info (aka.
 berlin.guixsd.org)
Date: Tue, 04 Dec 2018 15:12:20 +0100

Andreas Enge <andreas <at> enge.fr> skribis:

> On Tue, Dec 04, 2018 at 11:28:51AM +0100, Ludovic Courtès wrote:
>> Awesome.  I’ve added a file to record that info in
>> guix-maintenance.git.  Please feel free to complete it; for instance
>> it’d be good to know the expiration date of each domain.
>
> Done for guixsd.org. It expires on January 27. I would be okay to let it
> run out if guix.info is sufficiently marketed in the meantime.

Thanks.  I think it’s safer to keep it for a while because there may be
machine referring to berlin.guixsd.org here and there.

Ludo’.

Message #58 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: ludo <at> gnu.org (Ludovic Courtès)
To: Ricardo Wurmus <rekado <at> elephly.net>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: [PATCH 3/3] build: Default to https://ci.guix.info for
 substitutes.
Date: Tue, 04 Dec 2018 15:14:31 +0100

Ricardo Wurmus <rekado <at> elephly.net> skribis:

> Ludovic Courtès <ludo <at> gnu.org> writes:
>
>> * doc/guix.texi (SUBSTITUTE-SERVER): Switch to ci.guix.info.
> […]
>> diff --git a/doc/guix.texi b/doc/guix.texi
>> index d96c35eace..a5ae1c65e6 100644
>> --- a/doc/guix.texi
>> +++ b/doc/guix.texi
>> @@ -14,7 +14,7 @@
>>  @set KEY-SERVER pool.sks-keyservers.net
>>  
>>  @c The official substitute server used by default.
>> -@set SUBSTITUTE-SERVER ci.guix.info
>> +@set SUBSTITUTE-SERVER mirror.hydra.gnu.org
>
> Is this really correct?

Oops, I must have messed up while preparing the patches.
Thanks for the heads-up!

Ludo’.

Message #61 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ricardo Wurmus <rekado <at> elephly.net>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: [PATCH 3/3] build: Default to https://ci.guix.info for
 substitutes.
Date: Mon, 03 Dec 2018 21:38:36 +0100

Ludovic Courtès <ludo <at> gnu.org> writes:

> * doc/guix.texi (SUBSTITUTE-SERVER): Switch to ci.guix.info.
[…]
> diff --git a/doc/guix.texi b/doc/guix.texi
> index d96c35eace..a5ae1c65e6 100644
> --- a/doc/guix.texi
> +++ b/doc/guix.texi
> @@ -14,7 +14,7 @@
>  @set KEY-SERVER pool.sks-keyservers.net
>  
>  @c The official substitute server used by default.
> -@set SUBSTITUTE-SERVER ci.guix.info
> +@set SUBSTITUTE-SERVER mirror.hydra.gnu.org

Is this really correct?

-- 
Ricardo

Message #64 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Chris Marusich <cmmarusich <at> gmail.com>
To: ludo <at> gnu.org (Ludovic Courtès), Ricardo Wurmus
 <rekado <at> elephly.net>, Hartmut Goebel <hartmut <at> goebel-consult.de>,
 "Thompson\, David" <dthompson2 <at> worcester.edu>, Meiyo Peng
 <meiyo.peng <at> gmail.com>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: Using a CDN or some other mirror?
Date: Sat, 08 Dec 2018 19:33:17 -0800

[Message part 1 (text/plain, inline)]

Hi everyone,

ludo <at> gnu.org (Ludovic Courtès) writes:

> Ludovic Courtès <ludo <at> gnu.org> skribis:
>
> [...] I’m thinking about using a similar setup, but hosting the mirror
> on some Big Corp CDN or similar.  Chris Marusich came up with a setup
> along these lines a while back:
>
>   https://lists.gnu.org/archive/html/guix-devel/2016-03/msg00312.html
>
> Compared to Chris’s setup, given that ‘guix publish’ now provides
> ‘Cache-Control’ headers (that was not the case back then, see
> <https://lists.gnu.org/archive/html/guix-devel/2016-03/msg00360.html>),
> caching in the proxy should Just Work.
>
> I would like us to set up such a mirror for berlin and then have
> ci.guix.info point to that.  The project should be able to pay the
> hosting fees.
>
> Thoughts?

Regarding DNS, it would be nice if we could use an official GNU
subdomain.  If we can't use a GNU subdomain, we should at least make
sure we have some kind of DNS auto-renewal set up so that nobody can
poach our domain names.  And the operators should take appropriate
precautions when sharing any credentials used for managing it all.

Regarding CDNs, I definitely think it's worth a try!  Even Debian is
using CloudFront (cloudfront.debian.net).  In fact, email correspondence
suggests that as of 2013, Amazon may even have been paying for it!

https://lists.debian.org/debian-cloud/2013/05/msg00071.html

I wonder if Amazon would be willing to pay for our CloudFront
distribution if we asked them nicely?

In any case, before deciding to use Amazon CloudFront for ci.guix.info,
it would be prudent to estimate the cost.  CloudFront, like most Amazon
AWS services, is a "pay for what you use" model.  The pricing is here:

https://aws.amazon.com/cloudfront/pricing

To accurately estimate the cost, we need to know how many requests we
expect to receive, and how many bytes we expect to transfer out, during
a single month.  Do we have information like this for berlin today?

Although I don't doubt that a CDN will perform better than what we have
now, I do think it would be good to measure the performance so that we
know for sure the money spent is actually providing a benefit.  It would
be nice to have some data before and after to measure how availability
and performance have changed.  Apart from anecdotes, what data do we
have to determine whether performance has improved after introducing a
CDN?  For example, the following information could be useful:

  * Network load on the origin server(s)
  * Clients' latency to (the addresses pointed to by) ci.guix.info
  * Clients' throughput while downloading substitutes from ci.guix.info

We don't log or collect client metrics, and that's fine.  It could be
useful to add code to Guix to measure things like this when the user
asks to do so, but perhaps it isn't necessary.  It may be good enough if
people just volunteer to manually gather some information and share it.
For example, you can define a shell function like this:

--8<---------------cut here---------------start------------->8---
measure_get () {
curl -L \
     -o /dev/null \
     -w "url_effective: %{url_effective}\\n\
http_code: %{http_code}\\n\
num_connects: %{num_connects}\\n\
num_redirects: %{num_redirects}\\n\
remote_ip: %{remote_ip}\\n\
remote_port: %{remote_port}\\n\
size_download: %{size_download} B\\n\
speed_download: %{speed_download} B/s\\n\
time_appconnect: %{time_appconnect} s\\n\
time_connect: %{time_connect} s\\n\
time_namelookup: %{time_namelookup} s\\n\
time_pretransfer: %{time_pretransfer} s\\n\
time_redirect: %{time_redirect} s\\n\
time_starttransfer: %{time_starttransfer} s\\n\
time_total: %{time_total} s\\n" \
"$1"
}
--8<---------------cut here---------------end--------------->8---

See "man curl" for the meaning of each metric.

You can then use this function to measure a substitute download.  Here's
an example in which I download a large substitute (linux-libre) from one
of my machines in Seattle:

--8<---------------cut here---------------start------------->8---
$ measure_get https://berlin.guixsd.org/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19 2>/dev/null
url_effective: https://berlin.guixsd.org/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
http_code: 200
num_connects: 1
num_redirects: 0
remote_ip: 141.80.181.40
remote_port: 443
size_download: 69899433 B
speed_download: 4945831.000 B/s
time_appconnect: 0.885277 s
time_connect: 0.459667 s
time_namelookup: 0.254210 s
time_pretransfer: 0.885478 s
time_redirect: 0.000000 s
time_starttransfer: 1.273994 s
time_total: 14.133584 s
$ 
--8<---------------cut here---------------end--------------->8---

Here, it took 0.459667 - 0.254210 = 0.205457 seconds (about 205 ms) to
establish the TCP connection after the DNS lookup.  The average
throughput was 1924285 bytes per second (about 40 megabits per second,
where 1 megabit = 10^6 bits).  It seems my connection to berlin is
already pretty good!

We can get more information about latency by using a tool like mtr:

--8<---------------cut here---------------start------------->8---
$ sudo mtr -c 10 --report-wide --tcp -P 443 berlin.guixsd.org
Start: 2018-12-08T16:57:40-0800
HOST: localhost.localdomain                        Loss%   Snt   Last   Avg  Best  Wrst StDev
[... I've omitted the intermediate hops because they aren't relevant ...]
 13.|-- 141.80.181.40                                 0.0%    10  205.0 201.9 194.9 212.8   5.6
--8<---------------cut here---------------end--------------->8---

My machine's latency to berlin is about 202 ms, which matches what we
calculated above.

For experimentation, I've set up a CloudFront distribution at
berlin-mirror.marusich.info that uses berlin.guixsd.org as its origin
server.  Let's repeat these steps to measure the performance of the
distribution from my machine's perspective (before I did this, I made
sure the GET would result in a cache hit by downloading the substitute
once before and verifying that the same remote IP address was used):

--8<---------------cut here---------------start------------->8---
$ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19 2>/dev/null
url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
http_code: 200
num_connects: 1
num_redirects: 0
remote_ip: 13.32.254.57
remote_port: 443
size_download: 69899433 B
speed_download: 9821474.000 B/s
time_appconnect: 0.607593 s
time_connect: 0.532417 s
time_namelookup: 0.511086 s
time_pretransfer: 0.608029 s
time_redirect: 0.000000 s
time_starttransfer: 0.663578 s
time_total: 7.117266 s
$ sudo mtr -c 10 --report-wide --tcp -P 443 berlin-mirror.marusich.info
Start: 2018-12-08T17:04:48-0800
HOST: localhost.localdomain                        Loss%   Snt   Last   Avg  Best  Wrst StDev
[... I've omitted the intermediate hops because they aren't relevant ...]
 14.|-- server-52-84-21-199.sea32.r.cloudfront.net    0.0%    10   19.8  20.3  14.3  28.9   4.9
--8<---------------cut here---------------end--------------->8---

Establishing the TCP connection took about 21 ms (which matches the mtr
output), and the throughput was about 79 megabits per second.  (On this
machine, 100 Mbps is the current link speed, according to dmesg output.)
This means that in my case, when using CloudFront the latency is 10x
lower, and the throughput (for a cache hit) is 2x higher, than using
berlin.guixsd.org directly!

It would be interesting to see what the performance is for others.

Ricardo Wurmus <rekado <at> elephly.net> writes:

> Large ISPs also provide CDN services.  I already contacted Deutsche
> Telekom so that we can compare their CDN offer with the Amazon Cloudfont
> setup that Chris has configured.

That's great!  There are many CDN services out there.  I am unfamiliar
with most of them.  It will be good to see how Deutsche Telekom's
offering compares to CloudFront.

FYI, CloudFront has edge locations in the following parts of the world:

https://aws.amazon.com/cloudfront/features/

Hartmut Goebel <hartmut <at> goebel-consult.de> writes:

> Am 03.12.2018 um 17:12 schrieb Ludovic Courtès:
>> Thus, I’m thinking about using a similar setup, but hosting the mirror
>> on some Big Corp CDN or similar.
>
> Isn't this a contradiction: Building a free infrastructure relaying on
> servers from some Big Corporation? Let allow the privacy concerns
> raising when delivering data via some Big Corporation.
>
> If delivering "packages" works via static data without requiring any
> additional service, we could ask universities to host Guix, too. IMHO
> this is a much preferred solution since this is a decentralized publish
> infrastructure already in place for many GNU/Linux distributions.

I understand your concern about using a third-party service.  However,
we wouldn't be using a CDN as a "software substitute", which is one of
the primary risks of using a web service today:

https://www.gnu.org/philosophy/who-does-that-server-really-serve.html

Instead, we would be using a CDN as a performance optimization that is
transparent to a Guix user.  You seem unsettled by the idea of
entrusting any part of substitute delivery to a third party, but
concretely what risks do you foresee?

Regarding your suggestion to ask universities to host mirrors (really,
caching proxies), I think it could be a good idea.  As Leo mentioned,
the configuration to set up an NGINX caching proxy of Hydra (or berlin)
is freely available in maintenance.git.  Do you think we could convince
some universities to host caching proxies that just run an NGINX web
server using those configurations?

If we can accomplish that, it may still be helpful.  If there is
interest in going down this path, I can explore some possibilities in
the Seattle area.  If the university-owned caching proxies are easily
discoverable (i.e., we list them on the website), then users might
manually set their substitute URL to point to one that's close by.

Going further, if our DNS provider supports something like "geolocation
routing" for DNS queries, we might even be able to create DNS records
for ci.guix.info that point to those universities' caching proxies.  In
this way, when a user resolves ci.guix.info, they would get the address
of a university-owned caching proxy close by.  This could have the
benefits of requiring less money than a full-fledged CDN like Amazon
CloudFront, and also decentralizing the substitute delivery, while still
remaining transparent to Guix users.  However, it would still require us
to rely on a third-party DNS service.

For example, Amazon Route 53 provides this sort of geolocation routing:

https://docs.aws.amazon.com/Route53/latest/DeveloperGuide/routing-policy.html#routing-policy-geo

I wouldn't be surprised if there are other DNS providers out there who
offer something similar.  However, I also wouldn't be surprised if the
overall performance of CloudFront turns out to be better.

"Thompson, David" <dthompson2 <at> worcester.edu> writes:

> If AWS CloudFront is the path chosen, it may be worthwhile to follow
> the "infrastructure as code" practice and use CloudFormation to
> provision the CloudFront distribution and any other supporting
> resources. The benefit is that there would be a record of exactly
> *how* the project is using these commercial services and the setup
> could be easily reproduced.  The timing is interesting here because I
> just attended the annual AWS conference on behalf of my employer and
> while I was there I felt inspired to write a Guile API for building
> CloudFormation "stacks".  You can see a small sample of what it does
> here: https://gist.github.com/davexunit/db4b9d3e67902216fbdbc66cd9c6413e

Nice!  That seems useful.  I will have to play with it.  I created my
distributions manually using the AWS Management Console, since it's
relatively easy to do.  I agree it would be better to practice
"infrastructure as code."

On that topic, I've also heard good things about Terraform by HashiCorp,
which is available under the Mozilla Public License 2.0:

https://github.com/hashicorp/terraform

Here is a comparison of Terraform and CloudFormation:

https://www.terraform.io/intro/vs/cloudformation.html

I looked briefly into packaging Terraform for Guix.  It's written in Go.
It seems possible, but I haven't invested enough time yet.

As a final option, since the AWS CLI is already packaged in Guix, we
could just drive CloudFormation or CloudFront directly from the CLI.

Meiyo Peng <meiyo.peng <at> gmail.com> writes:

> I like the idea of IPFS. We should try it. It would be great if it works
> well.
>
> If at some point we need to setup traditional mirrors like other major
> Gnu/Linux distros, I can contact my friends in China to setup mirrors in
> several universities. I was a member of LUG <at> USTC, which provides the
> largest FLOSS mirror in China.

IPFS would be neat.  So would Gnunet.  Heck, even a publication
mechanism using good old BitTorrent would be nice.  All of these would
require changes to Guix, I suppose.

A CDN would require no changes to Guix, and that's part of why it's so
appealing.

-- 
Chris

[signature.asc (application/pgp-signature, inline)]

Message #67 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: Hartmut Goebel <hartmut <at> goebel-consult.de>
Cc: Chris Marusich <cmmarusich <at> gmail.com>, "Thompson,
 David" <dthompson2 <at> worcester.edu>, Ricardo Wurmus <rekado <at> elephly.net>,
 Meiyo Peng <meiyo.peng <at> gmail.com>, guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: Using a CDN or some other mirror?
Date: Sun, 09 Dec 2018 14:58:51 +0100

Hi Hartmut,

Hartmut Goebel <hartmut <at> goebel-consult.de> skribis:

> Am 09.12.2018 um 04:33 schrieb Chris Marusich:
>> Instead, we would be using a CDN as a performance optimization that is
>> transparent to a Guix user.  You seem unsettled by the idea of
>> entrusting any part of substitute delivery to a third party, but
>> concretely what risks do you foresee?
>
> I have serious privacy concerns.
>
> TL;DR: A CDN is a centralized infrastructure, allowing to collect
> information about valuable vulnerability information of almost all
> Guix-users and -systems. This is might become a thread to freedom of
> speech, human rights, democracy and economics. Guix should build on a
> decentralized infrastructure.

Heck it would be ironic to find myself arguing in favor of centralized
commercial services.  So I won’t do that.  :-)

Clearly, I do understand the concerns you list.  As a maintainer, I’m
looking for solutions that can address real problems (availability of
substitutes and bandwidth) while not being a threat to our user’s
privacy and security.

The operator of a substitute server (or caching proxy), in general,
knows which IPs downloaded vulnerable software.  This is the main
threat.

This can be mitigated by talking to nearby mirrors and not just
ci.guix.info, a feature we implemented a year ago (see
<https://gnu.org/s/guix/blog/2017/reproducible-builds-a-status-update/>),
or by using several substitute servers, or by not using (or not always
using) substitutes.  Few distros have all these options.

We might also be able to somehow balance requests between several CDNs
or mirrors.

But again, medium- to long-term, the goal is to move towards IPFS or
GNUnet/Bittorrent.  IPFS is attractive because it would probably require
no modifications to ‘guix substitutes’ and only minor changes to ‘guix
publish’ since the IPFS daemon has an HTTP interface.

>> Regarding your suggestion to ask universities to host mirrors (really,
>> caching proxies), I think it could be a good idea.  As Leo mentioned,
>> the configuration to set up an NGINX caching proxy of Hydra (or berlin)
>> is freely available in maintenance.git.  Do you think we could convince
>> some universities to host caching proxies that just run an NGINX web
>> server using those configurations?
>
> The difference is: For a traditional "ftp"-mirror, an organization just
> needs to add another source to its existing configuration and administer
> to the save way as all other mirrors. Whereas for a caching proxy they
> need to change the setup of the web-server and learn how to administer
> the cache. This difference might make it difficult to convince
> organizations to mirror.
>
> I could try and ask a few organizations in my area, but I would need
> figures for this.

What would you need to know?  ‘guix weather’ can provide info about
storage size.

Thanks,
Ludo’.

Message #70 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: Chris Marusich <cmmarusich <at> gmail.com>
Cc: Hartmut Goebel <hartmut <at> goebel-consult.de>, "Thompson,
 David" <dthompson2 <at> worcester.edu>, Ricardo Wurmus <rekado <at> elephly.net>,
 Meiyo Peng <meiyo.peng <at> gmail.com>, guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: CDN performance
Date: Sun, 09 Dec 2018 16:59:13 +0100

Hello Chris,

Chris Marusich <cmmarusich <at> gmail.com> skribis:

> Regarding DNS, it would be nice if we could use an official GNU
> subdomain.  If we can't use a GNU subdomain, we should at least make
> sure we have some kind of DNS auto-renewal set up so that nobody can
> poach our domain names.  And the operators should take appropriate
> precautions when sharing any credentials used for managing it all.

Agreed.

Regarding the GNU sub-domain, as I replied to Meiyo, I’m in favor of it,
all we need is someone to champion setting it up.

> Regarding CDNs, I definitely think it's worth a try!  Even Debian is
> using CloudFront (cloudfront.debian.net).  In fact, email correspondence
> suggests that as of 2013, Amazon may even have been paying for it!
>
> https://lists.debian.org/debian-cloud/2013/05/msg00071.html

(Note that debian.net is not Debian, and “there’s no cloud, only other
people’s computer” as the FSFE puts it.)

> Although I don't doubt that a CDN will perform better than what we have
> now, I do think it would be good to measure the performance so that we
> know for sure the money spent is actually providing a benefit.  It would
> be nice to have some data before and after to measure how availability
> and performance have changed.  Apart from anecdotes, what data do we
> have to determine whether performance has improved after introducing a
> CDN?  For example, the following information could be useful:
>
>   * Network load on the origin server(s)
>   * Clients' latency to (the addresses pointed to by) ci.guix.info
>   * Clients' throughput while downloading substitutes from ci.guix.info

Note that performance is one aspect; another one is availability (we’ve
seen that with the recent hydra.gnu.org outage!).  We know we’ll always
win in terms of availability by having a CDN in front of our servers.

That said, measuring performance is very useful and it’s great that we
can benefit from your expertise here!

> Here, it took 0.459667 - 0.254210 = 0.205457 seconds (about 205 ms) to
> establish the TCP connection after the DNS lookup.  The average
> throughput was 1924285 bytes per second (about 40 megabits per second,
> where 1 megabit = 10^6 bits).  It seems my connection to berlin is
> already pretty good!

Indeed.  The bandwidth problem on berlin is when you’re the first to
download a nar and it’s not been cached by nginx yet.  In that case, you
get very low bandwidth (like 10 times less than when the item is cached
by nginx.)  I’ve looked into it, went as far as strace’ing nginx, but
couldn’t find the reason of this.

Do you any idea?

> Establishing the TCP connection took about 21 ms (which matches the mtr
> output), and the throughput was about 79 megabits per second.  (On this
> machine, 100 Mbps is the current link speed, according to dmesg output.)
> This means that in my case, when using CloudFront the latency is 10x
> lower, and the throughput (for a cache hit) is 2x higher, than using
> berlin.guixsd.org directly!

Impressive.

> It would be interesting to see what the performance is for others.

I’ve tried this from home (in France, with FTTH):

--8<---------------cut here---------------start------------->8---
$ measure_get https://berlin.guixsd.org/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19  2>/dev/null
url_effective: https://berlin.guixsd.org/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
http_code: 200
num_connects: 1
num_redirects: 0
remote_ip: 141.80.181.40
remote_port: 443
size_download: 69899433 B
speed_download: 1522001.000 B/s
time_appconnect: 0.178892 s
time_connect: 0.049649 s
time_namelookup: 0.000422 s
time_pretransfer: 0.178934 s
time_redirect: 0.000000 s
time_starttransfer: 0.278312 s
time_total: 45.926021 s
$ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19 2>/dev/null
url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
http_code: 200
num_connects: 1
num_redirects: 0
remote_ip: 2600:9000:2116:6e00:c:49d4:5a80:93a1
remote_port: 443
size_download: 69899433 B
speed_download: 20803402.000 B/s
time_appconnect: 0.552008 s
time_connect: 0.482477 s
time_namelookup: 0.467598 s
time_pretransfer: 0.552157 s
time_redirect: 0.000000 s
time_starttransfer: 0.735758 s
time_total: 3.360500 s
--8<---------------cut here---------------end--------------->8---

Wall-clock time is less than a tenth; woow.

Thanks for sharing your insight and scripts!

Ludo’.

Message #73 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Hartmut Goebel <hartmut <at> goebel-consult.de>
To: Chris Marusich <cmmarusich <at> gmail.com>, Ludovic Courtès
 <ludo <at> gnu.org>, Ricardo Wurmus <rekado <at> elephly.net>,
 "Thompson, David" <dthompson2 <at> worcester.edu>,
 Meiyo Peng <meiyo.peng <at> gmail.com>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: Using a CDN or some other mirror?
Date: Sun, 9 Dec 2018 13:12:20 +0100

[Message part 1 (text/plain, inline)]

Am 09.12.2018 um 04:33 schrieb Chris Marusich:
> Instead, we would be using a CDN as a performance optimization that is
> transparent to a Guix user.  You seem unsettled by the idea of
> entrusting any part of substitute delivery to a third party, but
> concretely what risks do you foresee?

I have serious privacy concerns.

TL;DR: A CDN is a centralized infrastructure, allowing to collect
information about valuable vulnerability information of almost all
Guix-users and -systems. This is might become a thread to freedom of
speech, human rights, democracy and economics. Guix should build on a
decentralized infrastructure.

A distribution provider gets a notion which system is running which
software in which version. In case of guix, the provider even gets the
exact version of the software and all its dependencies. Combining this
with the rise of IPv6, which per default uses the MAC address as part of
the IP address, actually allows identifying a single system.

This information is extremely valuable for all kinds of attackers as it
eases attacking a system a lot. This becomes a thread to

  * to opposition members, dissidents and human rights activists as the
    intelligent agencies can target these persons much more precisely,
  * to companies all over the world as many countries do industrial
    espionage.

This becomes even worst when using a CDN, since the CDN is a centralized
system: A single CDN provider gains knowledge for almost all systems all
over the world. Which means: this valuable vulnerability information is
collected at a single place. Intelligence agencies might be keen on
getting access to this information and a centralized system makes it
easy for them. And there is evidence they actually collect this
information [*].

This gets even worse when the CDN belongs to one of these companies
compiling personal profiles, like Google, Facebook or Tencent. Amazon
belongs to this group.

I have the strong opinion that Guix should build on a decentralized
infrastructure to support keeping the freedom of speech, democracy and
human rights.

[*] Actually it is known the US-American intelligence agencies have
equipment placed at Verizon to collect all kind of data [1]. One can
reason the same is true for other big providers in the US. The USA has
the FISA act AFAIU enforcing US companies to collaborate in industrial
espionage. In Germany it is known that the BND is extracting high-volume
data at the central internet exchange (DE-CIX) [2]. One can reason such
also happens in other countries, esp. members of the five-eyes, France,
Russia, China, Israel, Saudi Arabia, Iran, Irak, etc.

> Regarding your suggestion to ask universities to host mirrors (really,
> caching proxies), I think it could be a good idea.  As Leo mentioned,
> the configuration to set up an NGINX caching proxy of Hydra (or berlin)
> is freely available in maintenance.git.  Do you think we could convince
> some universities to host caching proxies that just run an NGINX web
> server using those configurations?

The difference is: For a traditional "ftp"-mirror, an organization just
needs to add another source to its existing configuration and administer
to the save way as all other mirrors. Whereas for a caching proxy they
need to change the setup of the web-server and learn how to administer
the cache. This difference might make it difficult to convince
organizations to mirror.

I could try and ask a few organizations in my area, but I would need
figures for this.


[1] https://www.bbc.com/news/world-us-canada-23123964 or search the
internet for e.g. "cia verizon espionage"
[2]
https://www.heise.de/newsticker/meldung/Gerichtsurteil-BND-darf-weiterhin-Internet-Knoten-De-CIX-anzapfen-4061494.html
[3] https://en.wikipedia.org/wiki/Foreign_Intelligence_Surveillance_Act
[4]

-- 
+++hartmut

| Hartmut Goebel            |                       |
| hartmut <at> goebel-consult.de | www.goebel-consult.de |

[Message part 2 (text/html, inline)]

Message #76 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Meiyo Peng <meiyo.peng <at> gmail.com>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Tue, 11 Dec 2018 13:17:05 +0800

Hi,

I have changed my mind. I don't care about the domain name of the
default substitute server any more. I am more worried about the network
speed within China. While I am writing this email, my computer is
downloading substitutes from ci.guix.info at an average speed of about
50kB/s.

After careful thought, I realized the new CDN won't benefit China
residents as planned. Any popular CDN outside China is significantly
throttled by ISP/GFW and the situation is worse every year. A CDN will
be a great improvement for western countries but not for many asia
countries.

I have been running a nginx caching mirror at home since the first time
I met guix. My first attempt to install guix took me one afternoon and
one night. Most of the time was wasted on downloading substitutes. After
that, I searched online and found the nginx mirror configuration in
guix/maintenance.git. Then I set up my own caching mirror. It would be
extremely painful for me to play with guix during these months without
this caching mirror. Debian was my favorite GNU/Linux distro before I
discovered Guix. It usually costs me only 30 minutes to install Debian
from scratch. Other GNU/Linux distros are about the same. If guix takes
more than 5 hours to install, most new users will be scared away.

Although now there are only a few Chinese guix users, I still prefer to
set up a caching mirror within China. The user experience improvement
will be significant. Like Hartmut Goebel described, I need information
about "expected storage demand, expected transfer traffic from upstream,
how often updates are published, how long to hold copies" so I can write
a proposal to maintainers of mirrors.ustc.edu.cn to add a new mirror for
guix. I tried the `guix weather` command, but it failed with an error:

#+BEGIN_EXAMPLE
  computing 8,821 package derivations for x86_64-linux...
  guix weather: error: lstat: No such file or directory: "/src/config/packages/patches/libjxr-add-cmakelists-file.patch"
#+END_EXAMPLE

Can anyone provide the numbers for me? The numbers of
mirror.hydra.gnu.org would be a good reference.

In addition, I think we should put up a page on guix's website providing
information about available substitute servers including address,
backend build farm, geolocation, maintainer, so users (many asia
residents) can pick the best substitute server for themselves. This will
be necessary until alternative solutions like IPFS or GNUnet are
implemented.

This email only describes the situation in China. I don't intend to
discourage you on the CDN plan. There is no conflicts between my
proposal to set up a caching mirror in China and our plan to set up a
new CDN.

--
Meiyo Peng

Message #79 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Meiyo Peng <meiyo.peng <at> gmail.com>
To: Zephyr Waitzman <i <at> wi24rd.ml>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Tue, 11 Dec 2018 13:59:21 +0800

Hi,

Zephyr Waitzman <i <at> wi24rd.ml> writes:

> 大哥你的镜像速度怎么样？可以蹭网吗？（手动滑稽

My caching mirror is at home and it has an ip address of
192.168.x.x. You may connect to it if you have magic power.

BTW. Nice to see another Chinese here. What's your experience
with the default substitute server?

--
Meiyo Peng

Message #82 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Meiyo Peng <meiyo.peng <at> gmail.com>
To: Zephyr Waitzman <i <at> wi24rd.ml>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Tue, 11 Dec 2018 14:14:43 +0800

Hi,

Zephyr Waitzman <i <at> wi24rd.ml> writes:

> 以前官方的慢死人，最近的话安装靠玄学（找以前邮件列表里的几个 substitude 换着用也就那样吧），偶尔会出一些（感觉不像是网络的）问题。
> 以后有时间再深入学习下。

Thank you for your feedback. That means we do need to set up a caching
mirror within China.

--
Meiyo Peng

Message #85 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Giovanni Biscuolo <g <at> xelera.eu>
To: Ludovic Courtès <ludo <at> gnu.org>,
 Hartmut Goebel <hartmut <at> goebel-consult.de>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: Using a CDN or some other mirror?
Date: Tue, 11 Dec 2018 17:38:27 +0100

[Message part 1 (text/plain, inline)]

Hi all,

my two cents...

(I can't still help with a public cache, I hope soon...)

Ludovic Courtès <ludo <at> gnu.org> writes:

[...]

>> TL;DR: A CDN is a centralized infrastructure, allowing to collect
>> information about valuable vulnerability information of almost all
>> Guix-users and -systems. This is might become a thread to freedom of
>> speech, human rights, democracy and economics. Guix should build on a
>> decentralized infrastructure.

I completely agree with you, decentralization is the solution

unfortunately the **only functioning** way is to avoid current Internet,
since it's broken (https://youbroketheinternet.org/); I see GuixSD as an
integral part of The Project Map™ https://youbroketheinternet.org/map

...but to fix the situation we need a substantial GNUnet(work) effect
and for that we _need_ GuixSD substitutes to be easily and quickly
downloaded (can we avoid this asking potential adopters to be patient or
to build?)

maybe we should divide this task in two steps:

1. distributed substitutes: caching servers hosted by a network of
friendly institutions and companies donated to GNU/GuixSD, with a
haproxy frontend for geolocated load-balancing [1]

2. decentralized substitutes: caching servers on IPFS or better (since
it allows complete anonimity) on GNUnet

> Heck it would be ironic to find myself arguing in favor of centralized
> commercial services.  So I won’t do that.  :-)

I see no problems with commercial services, _unfortunately_ nowadays
this *almost* always means centralized silos, usually exploited for
global surveillance (since Internet is broken)

[...]

> The operator of a substitute server (or caching proxy), in general,
> knows which IPs downloaded vulnerable software.  This is the main
> threat.

on Internet, and on IPFS? (sorry for the ignorance)

on GNUNet filesharing can be completely anonymous, but the performace is
degraded (so we need a large network effect here)

> This can be mitigated by talking to nearby mirrors and not just
> ci.guix.info, a feature we implemented a year ago (see
> <https://gnu.org/s/guix/blog/2017/reproducible-builds-a-status-update/>),
> or by using several substitute servers, or by not using (or not always
> using) substitutes.  Few distros have all these options.
>
> We might also be able to somehow balance requests between several CDNs
> or mirrors.

did someone explored an haproxy (with geolocation) solution?

is there a wip-haproxy attempt?

[...]

HTH
Giovanni


[1] in the next few weeks I'm going to test an haproxy instance with
geolocated ACLs following this directions
https:/www.haproxy.com/blog/use-geoip-database-within-haproxy/

-- 
Giovanni Biscuolo

Xelera IT Infrastructures

[signature.asc (application/pgp-signature, inline)]

Message #88 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Chris Marusich <cmmarusich <at> gmail.com>
To: Meiyo Peng <meiyo.peng <at> gmail.com>
Cc: guix-devel <at> gnu.org, Ludovic Courtès <ludo <at> gnu.org>,
 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Wed, 12 Dec 2018 23:11:25 -0800

[Message part 1 (text/plain, inline)]

Meiyo Peng <meiyo.peng <at> gmail.com> writes:

> After careful thought, I realized the new CDN won't benefit China
> residents as planned. Any popular CDN outside China is significantly
> throttled by ISP/GFW and the situation is worse every year. A CDN will
> be a great improvement for western countries but not for many asia
> countries.

Could you try running the measure_get shell function I included in the
following email?

https://lists.gnu.org/archive/html/guix-devel/2018-12/msg00192.html

For convenience, here is the definition:

--8<---------------cut here---------------start------------->8---
measure_get () {
curl -L \
     -o /dev/null \
     -w "url_effective: %{url_effective}\\n\
http_code: %{http_code}\\n\
num_connects: %{num_connects}\\n\
num_redirects: %{num_redirects}\\n\
remote_ip: %{remote_ip}\\n\
remote_port: %{remote_port}\\n\
size_download: %{size_download} B\\n\
speed_download: %{speed_download} B/s\\n\
time_appconnect: %{time_appconnect} s\\n\
time_connect: %{time_connect} s\\n\
time_namelookup: %{time_namelookup} s\\n\
time_pretransfer: %{time_pretransfer} s\\n\
time_redirect: %{time_redirect} s\\n\
time_starttransfer: %{time_starttransfer} s\\n\
time_total: %{time_total} s\\n" \
"$1"
}
--8<---------------cut here---------------end--------------->8---

Specifically, I am curious to know what performance you get when you run

  measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19

from a computer in China.  Please be sure to run it two times in a row,
to ensure that CloudFront has cached the object.

CloudFront has edge locations in Hong Kong, so I am curious to know what
performance improvement, if any, you observe.

-- 
Chris

[signature.asc (application/pgp-signature, inline)]

Message #91 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Chris Marusich <cmmarusich <at> gmail.com>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: Hartmut Goebel <hartmut <at> goebel-consult.de>, "Thompson,
 David" <dthompson2 <at> worcester.edu>, Ricardo Wurmus <rekado <at> elephly.net>,
 Meiyo Peng <meiyo.peng <at> gmail.com>, guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Thu, 13 Dec 2018 00:05:06 -0800

[Message part 1 (text/plain, inline)]

Ludovic Courtès <ludo <at> gnu.org> writes:

> Regarding the GNU sub-domain, as I replied to Meiyo, I’m in favor of it,
> all we need is someone to champion setting it up.

I could help with this.  Whom should I contact?

>> Regarding CDNs, I definitely think it's worth a try!  Even Debian is
>> using CloudFront (cloudfront.debian.net).  In fact, email correspondence
>> suggests that as of 2013, Amazon may even have been paying for it!
>>
>> https://lists.debian.org/debian-cloud/2013/05/msg00071.html
>
> (Note that debian.net is not Debian, and “there’s no cloud, only other
> people’s computer” as the FSFE puts it.)

I do try to avoid the term "cloud" whenever possible.  It's hard to
avoid when it's in the product name, though!  A wise man once said, "A
cloud in the mind is an obstacle to clear thinking."  ;-)

You may be right about debian.net.  I don't know who owns that domain.
It's confusing, since debian.org is definitely owned by the Debian
project, and the following page says they're using Amazon CloudFront:

https://deb.debian.org/

Maybe Debian still uses Amazon CloudFront, or maybe they don't any more.
In any case, I've found the following email thread, which documents a
thoughtful discussion regarding whether or not Debian should use a CDN.
They discussed many of the same concerns we're discussing here.

https://lists.debian.org/debian-project/2013/10/msg00029.html

A summary, in the middle of the long thread, is here:

https://lists.debian.org/debian-project/2013/10/msg00074.html

Later, part of the thread broke off and continued here:

https://lists.debian.org/debian-project/2014/02/msg00001.html

That's as far as I've read.

Judging by that email thread, one of the reasons why Debian considered
using a CDN was because they felt that the cost, in terms of people
power, of maintaining their own "proto-CDN" infrastructure had grown too
great.  I believe it!  I think it would be ill-advised for the Guix
project to expend effort and capital on building and maintaining its own
CDN.  I think it would be wiser to focus on developing a decentralized
substitute solution (GNUnet, IPFS, etc.).

That said, I still think that today Guix should provide a third-party
CDN option.  For many Guix users, a CDN would improve performance and
availability of substitutes.  Contracting with a third party to provide
the CDN service would require much less effort and capital than building
and maintaining a CDN from scratch.  This would also enable the project
to focus more on building a decentralized substitute solution.  And once
that decentralized solution is ready, it will be easy to just "turn off"
the CDN.

I also understand Hartmut's concerns.  The risks he points out are
valid.  Because of those risks, even if we make a third-party CDN option
available, some people will choose not to use it.  For that reason, we
should not require Guix users to use a third-party CDN, just as we do
not require them to use substitutes from our build farm.

However, not everyone shares the same threat model.  For example,
although some people choose not to trust substitutes from our build
farm, still others do.  The choice is based on one's own individual
situation.  Similarly, if we make a third-party CDN option available and
explain the risks of using it, Guix users will be able to make an
educated decision for themselves about whether or not to use it.

>> Here, it took 0.459667 - 0.254210 = 0.205457 seconds (about 205 ms) to
>> establish the TCP connection after the DNS lookup.  The average
>> throughput was 1924285 bytes per second (about 40 megabits per second,
>> where 1 megabit = 10^6 bits).  It seems my connection to berlin is
>> already pretty good!
>
> Indeed.  The bandwidth problem on berlin is when you’re the first to
> download a nar and it’s not been cached by nginx yet.  In that case, you
> get very low bandwidth (like 10 times less than when the item is cached
> by nginx.)  I’ve looked into it, went as far as strace’ing nginx, but
> couldn’t find the reason of this.
>
> Do you any idea?

I made a typo here.  The value "1924285" should have been "4945831",
which is what measure_get printed.  However, the intended result (40
Mbps) is still correct.

Actually, I thought 40 megabits per second was pretty great for a
single-threaded file transfer that originated in Europe (I think?) and
terminated in Seattle (via my residential Comcast downlink).  I
requested that particular file many times before that final test run, so
it was probably already cached by nginx.

However, I believe you when you say that it's slow the first time you
download the substitute from berlin.  What path does the data take from
its origin through berlin?  If berlin needs to download the initial file
from another server, perhaps the connection between berlin and that
server is the bottleneck?  Maybe we should discuss that in a different
email thread, though.

> I’ve tried this from home (in France, with FTTH):
>
> [...]
>
> speed_download: 20803402.000 B/s

Wow, that's 166 megabits per second!  I'm jealous.  :-)

> Wall-clock time is less than a tenth; woow.

I expect others will see similar performance improvements, as long as
they are close to the edge locations provided by Amazon CloudFront and
their local ISP downlink is fast enough to see a benefit.  Again, the
edge locations are here:

https://aws.amazon.com/cloudfront/features/

Here's a direct link to their current map:

https://d1.awsstatic.com/global-infrastructure/maps/CloudFront%20Network%20Map%2010.12.18.59e838df2f373247d2efaeb548076e084fd8993e.png

-- 
Chris

[signature.asc (application/pgp-signature, inline)]

Message #94 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Giovanni Biscuolo <g <at> xelera.eu>
To: Chris Marusich <cmmarusich <at> gmail.com>,
 Ludovic Courtès <ludo <at> gnu.org>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Thu, 13 Dec 2018 11:41:06 +0100

[Message part 1 (text/plain, inline)]

Hi Chris,

nice to see this discussion, IMHO how GuixSD subsitutes are distributed
is a key issue in our ecosystem and is _all_ about privacy and metadata
*mass* collection

most "normal users" are not concerned about this so they are fine with
super-centralization since it's a convenience... not us :-)

personally I've come to GuixSD because I see this project as a key part
in liberating me from this class of problems

Chris Marusich <cmmarusich <at> gmail.com> writes:

[...]

> A summary, in the middle of the long thread, is here:
>
> https://lists.debian.org/debian-project/2013/10/msg00074.html

thank you for the reference, I've only read this summary

the key part of it IMHO is
  "Q: Do CDNs raise more security/privacy concerns than our mirrors?"

and the related subthread
 https://lists.debian.org/debian-project/2013/10/msg00033.html

the quick reply to the above question is: yes, CDNs raise more
secutiry/privacy concerns than "distributed mirrors"

obviuosly "distributed mirrors" _does_ rise some security/privacy
concerns but *centralization*... much more

[...]

> Judging by that email thread, one of the reasons why Debian considered
> using a CDN was because they felt that the cost, in terms of people
> power, of maintaining their own "proto-CDN" infrastructure had grown too
> great.

I'm still new to guixsd but understood enough to se we are much more
well equipped to maintain our distributed network of substitutes caching
servers... **transparently** configured :-)

[...]

> I also understand Hartmut's concerns.  The risks he points out are
> valid.  Because of those risks, even if we make a third-party CDN option
> available, some people will choose not to use it.

probably I'll be one of those, I'm considering to maintain a caching
substitute server in a "semi-trusted" colocated space and I'd be very
happy to share that server with the community

[...]

> However, not everyone shares the same threat model.  For example,
> although some people choose not to trust substitutes from our build
> farm, still others do.

for this very reason IMHO we should work towards a network of **very
trusted** build farms directly managed and controlled by the GuixSD
project sysadmins; if build farms will be able to quickly provide
substitutes, caching mirrors will be _much more_ effective than today

... and a network of "automated guix challenge" servers to spot
not-reproducible software in GuixSD

with a solid infrastructure of "scientifically" trustable build farms,
there are no reasons not to trust substitutes servers (this implies
working towards 100% reproducibility of GuixSD)

> The choice is based on one's own individual
> situation.  Similarly, if we make a third-party CDN option available and
> explain the risks of using it, Guix users will be able to make an
> educated decision for themselves about whether or not to use it.

that's an option... like a "last resort" in order to be able to use
guixSD :-)

we could also teach people how to setup their own caching servers and
possibly share them with the rest of the local community (possibly with
some coordination effort from the project sysadmins)

for Milan I've plans to setup such caching mirror in Jan 2019

[...]

happy hacking!
Gio

-- 
Giovanni Biscuolo

Xelera IT Infrastructures

[signature.asc (application/pgp-signature, inline)]

Message #97 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Hartmut Goebel <h.goebel <at> crazy-compilers.com>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: Chris Marusich <cmmarusich <at> gmail.com>, "Thompson,
 David" <dthompson2 <at> worcester.edu>, Ricardo Wurmus <rekado <at> elephly.net>,
 Meiyo Peng <meiyo.peng <at> gmail.com>, guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: Using a CDN or some other mirror?
Date: Fri, 14 Dec 2018 09:35:38 +0100

[Message part 1 (text/plain, inline)]

Am 09.12.2018 um 14:58 schrieb Ludovic Courtès:
>> I could try and ask a few organizations in my area, but I would need
>> figures for this.
> What would you need to know?  ‘guix weather’ can provide info about
> storage size.

I don't know yet, which info the admins need for a decision. FMPOV I'd
says: Disk-space and traffic to be expected.

`guix weather` only provides the disk-space, but even this is not
obvious for me:

  13912.1 MiB of nars (compressed)
  41176.6 MiB on disk (uncompressed)

From reading the manual, I assume 13.9 GB are required on the server
(which is quite a lot IMHO). Is this correct?

-- 
+++hartmut

| Hartmut Goebel            |                       |
| hartmut <at> goebel-consult.de | www.goebel-consult.de |

[Message part 2 (text/html, inline)]

Message #100 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Pierre Neidhardt <mail <at> ambrevar.xyz>
To: Hartmut Goebel <h.goebel <at> crazy-compilers.com>
Cc: guix-devel <at> gnu.org, Ludovic Courtès <ludo <at> gnu.org>,
 33600 <at> debbugs.gnu.org
Subject: Re: Using a CDN or some other mirror?
Date: Fri, 14 Dec 2018 10:02:33 +0100

[Message part 1 (text/plain, inline)]

Talking about this, I recently discussed with Ludovic the idea of compressing
nars in Lzip instead of gzip.

http://lzip.nongnu.org/lzip.html (benchmark included)

I can work on some Lzip guile-bindings, it should be quite easy, then we could
save some 10-50% (!!!) disk usage.

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[signature.asc (application/pgp-signature, inline)]

Message #103 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: Chris Marusich <cmmarusich <at> gmail.com>
Cc: Hartmut Goebel <hartmut <at> goebel-consult.de>,
 Julien Lepiller <julien <at> lepiller.eu>, "Thompson,
 David" <dthompson2 <at> worcester.edu>, Ricardo Wurmus <rekado <at> elephly.net>,
 Meiyo Peng <meiyo.peng <at> gmail.com>, guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: guix.gnu.org sub-domain
Date: Fri, 14 Dec 2018 11:26:25 +0100

Hi Chris,

Chris Marusich <cmmarusich <at> gmail.com> skribis:

> Ludovic Courtès <ludo <at> gnu.org> writes:
>
>> Regarding the GNU sub-domain, as I replied to Meiyo, I’m in favor of it,
>> all we need is someone to champion setting it up.
>
> I could help with this.  Whom should I contact?

We discussed this over the last few days in Paris and Julien (roptat on
IRC) volunteered to come up with a Knot service setup for bayfront.scm.
When that’s ready, we can contact the FSF sysadmins so they delegate to
bayfront.

I’m sure Julien wouldn’t mind getting some help or insight, so please do
get in touch!

Ludo’.

Message #106 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: Chris Marusich <cmmarusich <at> gmail.com>
Cc: Hartmut Goebel <hartmut <at> goebel-consult.de>, "Thompson,
 David" <dthompson2 <at> worcester.edu>, Ricardo Wurmus <rekado <at> elephly.net>,
 Meiyo Peng <meiyo.peng <at> gmail.com>, guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Fri, 14 Dec 2018 11:35:37 +0100

Hello!

Chris Marusich <cmmarusich <at> gmail.com> skribis:

> Ludovic Courtès <ludo <at> gnu.org> writes:


[...]

> Judging by that email thread, one of the reasons why Debian considered
> using a CDN was because they felt that the cost, in terms of people
> power, of maintaining their own "proto-CDN" infrastructure had grown too
> great.  I believe it!  I think it would be ill-advised for the Guix
> project to expend effort and capital on building and maintaining its own
> CDN.  I think it would be wiser to focus on developing a decentralized
> substitute solution (GNUnet, IPFS, etc.).
>
> That said, I still think that today Guix should provide a third-party
> CDN option.  For many Guix users, a CDN would improve performance and
> availability of substitutes.  Contracting with a third party to provide
> the CDN service would require much less effort and capital than building
> and maintaining a CDN from scratch.  This would also enable the project
> to focus more on building a decentralized substitute solution.  And once
> that decentralized solution is ready, it will be easy to just "turn off"
> the CDN.
>
> I also understand Hartmut's concerns.  The risks he points out are
> valid.  Because of those risks, even if we make a third-party CDN option
> available, some people will choose not to use it.  For that reason, we
> should not require Guix users to use a third-party CDN, just as we do
> not require them to use substitutes from our build farm.
>
> However, not everyone shares the same threat model.  For example,
> although some people choose not to trust substitutes from our build
> farm, still others do.  The choice is based on one's own individual
> situation.  Similarly, if we make a third-party CDN option available and
> explain the risks of using it, Guix users will be able to make an
> educated decision for themselves about whether or not to use it.

That summarizes the situation very well.

In Paris, there was consensus on two things: that a CDN would be helpful
for performance and availability (though availability could suddenly be
worse if we run… out of money), and that it raises privacy concerns.

Based on this, some suggested having a way to opt out of the CDN
distribution: for example, we might make ci.guix.info point to the CDN
and also advertise berlin.guixsd.org as a way to bypass it.

Of course in parallel we want to do our best to develop decentralized
solutions.

> However, I believe you when you say that it's slow the first time you
> download the substitute from berlin.  What path does the data take from
> its origin through berlin?  If berlin needs to download the initial file
> from another server, perhaps the connection between berlin and that
> server is the bottleneck?  Maybe we should discuss that in a different
> email thread, though.

On the first hit, nginx fetches the file from ‘guix publish’, which is
also on berlin.  ‘guix publish’ gets the file from local storage and
passes it with sendfile(2), so it’s as fast as can be.

Thanks for the info!

Ludo’.

Message #109 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: Hartmut Goebel <h.goebel <at> crazy-compilers.com>
Cc: Chris Marusich <cmmarusich <at> gmail.com>, "Thompson,
 David" <dthompson2 <at> worcester.edu>, Ricardo Wurmus <rekado <at> elephly.net>,
 Meiyo Peng <meiyo.peng <at> gmail.com>, guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: Using a CDN or some other mirror?
Date: Fri, 14 Dec 2018 15:45:11 +0100

Hello,

Hartmut Goebel <h.goebel <at> crazy-compilers.com> skribis:

> Am 09.12.2018 um 14:58 schrieb Ludovic Courtès:
>>> I could try and ask a few organizations in my area, but I would need
>>> figures for this.
>> What would you need to know?  ‘guix weather’ can provide info about
>> storage size.
>
> I don't know yet, which info the admins need for a decision. FMPOV I'd
> says: Disk-space and traffic to be expected.
>
> `guix weather` only provides the disk-space, but even this is not
> obvious for me:
>
>   13912.1 MiB of nars (compressed)
>   41176.6 MiB on disk (uncompressed)
>
> From reading the manual, I assume 13.9 GB are required on the server
> (which is quite a lot IMHO). Is this correct?

If you’re running a caching proxy, you’ll need 13G.  However note that
it’s only for one architecture and one revision of Guix.  The total
space needed is obviously a function of time (number of Guix revisions
served) and number of architectures.

You could choose an expiration time in your caching proxy that satisfies
your disk space constraints, though.

Our machines that run Cuirass + ‘guix publish --cache’ need roughly
41+13G since they contain both /gnu/store and /var/cache/guix/publish.

HTH!

Ludo’.

Message #112 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: Pierre Neidhardt <mail <at> ambrevar.xyz>
Cc: guix-devel <at> gnu.org, Hartmut Goebel <h.goebel <at> crazy-compilers.com>,
 33600 <at> debbugs.gnu.org
Subject: Compressing nars with lzip or similar
Date: Fri, 14 Dec 2018 15:48:24 +0100

Pierre Neidhardt <mail <at> ambrevar.xyz> skribis:

> Talking about this, I recently discussed with Ludovic the idea of compressing
> nars in Lzip instead of gzip.
>
> http://lzip.nongnu.org/lzip.html (benchmark included)
>
> I can work on some Lzip guile-bindings, it should be quite easy, then we could
> save some 10-50% (!!!) disk usage.

That would be sweet!

Though we have to keep in mind that today’s ‘guix substitute’ doesn’t
know about lzip at all (see (guix scripts substitute) and
‘call-with-decompressed-port’.)  So while lzip would be my preference,
we may not be able to use it until we can be sure our users can actually
unpack it.

Ludo’.

Message #115 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Pierre Neidhardt <mail <at> ambrevar.xyz>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: guix-devel <at> gnu.org, Hartmut Goebel <h.goebel <at> crazy-compilers.com>,
 33600 <at> debbugs.gnu.org
Subject: Re: Compressing nars with lzip or similar
Date: Fri, 14 Dec 2018 16:21:33 +0100

[Message part 1 (text/plain, inline)]

Absolutely, but once the bindings are up and running, it should be
straightforward to update (guix scripts substitute), no?

-- 
Pierre Neidhardt
https://ambrevar.xyz/

[signature.asc (application/pgp-signature, inline)]

Message #118 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Mark H Weaver <mhw <at> netris.org>
To: Giovanni Biscuolo <g <at> xelera.eu>
Cc: guix-devel <at> gnu.org, Ludovic Courtès <ludo <at> gnu.org>,
 Chris Marusich <cmmarusich <at> gmail.com>, 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Fri, 14 Dec 2018 20:40:55 -0500

Hi Giovanni,

Giovanni Biscuolo <g <at> xelera.eu> writes:
> with a solid infrastructure of "scientifically" trustable build farms,
> there are no reasons not to trust substitutes servers (this implies
> working towards 100% reproducibility of GuixSD)

What does "scientifically trustable" mean?

      Mark

Message #121 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Pierre Neidhardt <mail <at> ambrevar.xyz>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: guix-devel <at> gnu.org, Hartmut Goebel <h.goebel <at> crazy-compilers.com>,
 33600 <at> debbugs.gnu.org
Subject: Re: Compressing nars with lzip or similar
Date: Sat, 15 Dec 2018 13:17:05 +0100

[Message part 1 (text/plain, inline)]

I've done some quick research over the various options.

- lzip: better than gz and bzip2 for sure, possibly better than xz (at
least according to the author).

- plzip: for "parallel lzip".  With 4 threads I was able to compress
icecat 2.5x faster.  It used 5x more memory though.  The compression
ratio is 1-2% worse.

- lrzip: it would crash whenever I would change the compression level.
  Seems less stable.  It's as fast as plzip, while being 1-2% less
  compressed.  I don't think it's worth using.

All in all, lzip is a definite win over most options.  The main question
is: lzip or plzip?

In my opinion it makes more sense to use plzip, gotta put those cores to
some use (on the user side at least)!  Not sure what the build farm
would think about this though.

Finally, plzip can be quite memory-intensive.  When passed "--threads=1"
however, it gets closer to lzip (bit faster, bit more memory).  So maybe
the ideal would be to support plzip with a user-settable "threads" option.

Thoughts?

--
Pierre Neidhardt
https://ambrevar.xyz/

[signature.asc (application/pgp-signature, inline)]

Message #124 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: Pierre Neidhardt <mail <at> ambrevar.xyz>
Cc: guix-devel <at> gnu.org, Hartmut Goebel <h.goebel <at> crazy-compilers.com>,
 33600 <at> debbugs.gnu.org
Subject: Re: Compressing nars with lzip or similar
Date: Sat, 15 Dec 2018 19:04:39 +0100

Pierre Neidhardt <mail <at> ambrevar.xyz> skribis:

> Absolutely, but once the bindings are up and running, it should be
> straightforward to update (guix scripts substitute), no?

Sure!  But it will take time before we can assume all our users are
running it.

Ludo’.

Message #127 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: Pierre Neidhardt <mail <at> ambrevar.xyz>
Cc: guix-devel <at> gnu.org, Hartmut Goebel <h.goebel <at> crazy-compilers.com>,
 33600 <at> debbugs.gnu.org
Subject: Re: Compressing nars with lzip or similar
Date: Sat, 15 Dec 2018 19:06:21 +0100

Pierre Neidhardt <mail <at> ambrevar.xyz> skribis:

> I've done some quick research over the various options.
>
> - lzip: better than gz and bzip2 for sure, possibly better than xz (at
> least according to the author).
>
> - plzip: for "parallel lzip".  With 4 threads I was able to compress
> icecat 2.5x faster.  It used 5x more memory though.  The compression
> ratio is 1-2% worse.
>
> - lrzip: it would crash whenever I would change the compression level.
>   Seems less stable.  It's as fast as plzip, while being 1-2% less
>   compressed.  I don't think it's worth using.
>
> All in all, lzip is a definite win over most options.  The main question
> is: lzip or plzip?

‘guix publish’ has its own worker pool and handles parallelism
internally, so in that context plain sequential lzip would be more
appropriate IMO.

Ludo’.

Message #130 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Chris Marusich <cmmarusich <at> gmail.com>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: Hartmut Goebel <hartmut <at> goebel-consult.de>,
 Julien Lepiller <julien <at> lepiller.eu>, "Thompson,
 David" <dthompson2 <at> worcester.edu>, Ricardo Wurmus <rekado <at> elephly.net>,
 Meiyo Peng <meiyo.peng <at> gmail.com>, guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: guix.gnu.org sub-domain
Date: Sat, 15 Dec 2018 15:20:32 -0800

[Message part 1 (text/plain, inline)]

Hi Ludo,

Ludovic Courtès <ludo <at> gnu.org> writes:

> I’m sure Julien wouldn’t mind getting some help or insight, so please do
> get in touch!

OK, I'll speak privately with Julien about the DNS setup to avoid adding
noise to this email thread.

-- 
Chris

[signature.asc (application/pgp-signature, inline)]

Message #133 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Meiyo Peng <meiyo.peng <at> gmail.com>
To: Chris Marusich <cmmarusich <at> gmail.com>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Mon, 17 Dec 2018 14:48:42 +0800

Hi Chris,

Chris Marusich <cmmarusich <at> gmail.com> writes:

> Meiyo Peng <meiyo.peng <at> gmail.com> writes:
>
>> After careful thought, I realized the new CDN won't benefit China
>> residents as planned. Any popular CDN outside China is significantly
>> throttled by ISP/GFW and the situation is worse every year. A CDN will
>> be a great improvement for western countries but not for many asia
>> countries.
>
> Could you try running the measure_get shell function I included in the
> following email?
>
> https://lists.gnu.org/archive/html/guix-devel/2018-12/msg00192.html
>
> For convenience, here is the definition:
>
> measure_get () {
> curl -L \
>      -o /dev/null \
>      -w "url_effective: %{url_effective}\\n\
> http_code: %{http_code}\\n\
> num_connects: %{num_connects}\\n\
> num_redirects: %{num_redirects}\\n\
> remote_ip: %{remote_ip}\\n\
> remote_port: %{remote_port}\\n\
> size_download: %{size_download} B\\n\
> speed_download: %{speed_download} B/s\\n\
> time_appconnect: %{time_appconnect} s\\n\
> time_connect: %{time_connect} s\\n\
> time_namelookup: %{time_namelookup} s\\n\
> time_pretransfer: %{time_pretransfer} s\\n\
> time_redirect: %{time_redirect} s\\n\
> time_starttransfer: %{time_starttransfer} s\\n\
> time_total: %{time_total} s\\n" \
> "$1"
> }
>
> Specifically, I am curious to know what performance you get when you run
>
>   measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>
> from a computer in China.  Please be sure to run it two times in a row,
> to ensure that CloudFront has cached the object.
>
> CloudFront has edge locations in Hong Kong, so I am curious to know what
> performance improvement, if any, you observe.

Sorry for the delay. My computer was reinstalled with Windows and taken
away by my girlfriend. So I have been waiting for my new computer that I
bought online to arrive. Finally, it arrived yesterday and I
successfully installed Guix on it.

I tested your script several times.

1. Tested today at home. China Unicom home broadband. 50Mb/s.

The result is slow as usual. curl failed once.
berlin-mirror.marusich.info is resolved to Seattle, WA, US.

#+BEGIN_EXAMPLE
  ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
   55 66.6M   55 36.9M    0     0  17926      0  1:04:59  0:36:02  0:28:57 17733
  url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 52.85.158.151
  remote_port: 443
  size_download: 38764357 B
  speed_download: 17926.000 B/s
  time_appconnect: 6.078850 s
  time_connect: 3.006821 s
  time_namelookup: 2.659785 s
  time_pretransfer: 6.079097 s
  time_redirect: 0.000000 s
  time_starttransfer: 9.626001 s
  time_total: 2162.379211 s
  curl: (92) HTTP/2 stream 0 was not closed cleanly: INTERNAL_ERROR (err 2)

  ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
  100 66.6M  100 66.6M    0     0   109k      0  0:10:25  0:10:25 --:--:--  241k
  url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 52.85.158.22
  remote_port: 443
  size_download: 69899433 B
  speed_download: 111816.000 B/s
  time_appconnect: 3.507528 s
  time_connect: 2.650373 s
  time_namelookup: 2.261801 s
  time_pretransfer: 3.507637 s
  time_redirect: 0.000000 s
  time_starttransfer: 5.995298 s
  time_total: 625.129571 s

  ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
  100 66.6M  100 66.6M    0     0   109k      0  0:10:23  0:10:23 --:--:--  141k
  url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 52.85.158.22
  remote_port: 443
  size_download: 69899433 B
  speed_download: 112187.000 B/s
  time_appconnect: 2.280972 s
  time_connect: 1.407197 s
  time_namelookup: 1.056180 s
  time_pretransfer: 2.281234 s
  time_redirect: 0.000000 s
  time_starttransfer: 3.167703 s
  time_total: 623.061584 s
#+END_EXAMPLE


2. Tested 3 days ago at my office. China Telecom enterprise broadband. 50Mb/s.

Unusually fast! berlin-mirror.marusich.info is resolved to Seattle, WA,
US. I have no idea why it's so fast that day.

#+BEGIN_EXAMPLE
  ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
  100 66.6M  100 66.6M    0     0  1364k      0  0:00:50  0:00:50 --:--:-- 1352k
  url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 13.35.20.109
  remote_port: 443
  size_download: 69899433 B
  speed_download: 1397429.000 B/s
  time_appconnect: 2.432387 s
  time_connect: 0.200842 s
  time_namelookup: 0.000446 s
  time_pretransfer: 2.432659 s
  time_redirect: 0.000000 s
  time_starttransfer: 2.673045 s
  time_total: 50.020945 s

  ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
  100 66.6M  100 66.6M    0     0  1592k      0  0:00:42  0:00:42 --:--:-- 2506k
  url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 13.35.20.109
  remote_port: 443
  size_download: 69899433 B
  speed_download: 1630687.000 B/s
  time_appconnect: 0.653270 s
  time_connect: 0.209455 s
  time_namelookup: 0.001582 s
  time_pretransfer: 0.658399 s
  time_redirect: 0.000000 s
  time_starttransfer: 0.883126 s
  time_total: 42.865868 s
#+END_EXAMPLE


3. Tested today at my office. China Telecom enterprise broadband. 50Mb/s.

Slow as usual. berlin-mirror.marusich.info is still resolved to Seattle,
WA, US. I killed the program several times because it hung there with no
data transfer for a few minutes. The TCP connection was probably closed
by GFW. This is very common here.

#+BEGIN_EXAMPLE
  ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
  100 66.6M  100 66.6M    0     0  48110      0  0:24:12  0:24:12 --:--:-- 41808
  url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 52.85.158.151
  remote_port: 443
  size_download: 69899433 B
  speed_download: 48110.000 B/s
  time_appconnect: 0.872926 s
  time_connect: 0.282048 s
  time_namelookup: 0.000524 s
  time_pretransfer: 0.873099 s
  time_redirect: 0.000000 s
  time_starttransfer: 1.187467 s
  time_total: 1452.904154 s

  ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
    2 66.6M    2 1809k    0     0   5760      0  3:22:15  0:05:21  3:16:54     0^C%

  ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
   52 66.6M   52 34.9M    0     0  16772      0  1:09:27  0:36:26  0:33:01     0^C%

  ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
  100 66.6M  100 66.6M    0     0  58181      0  0:20:01  0:20:01 --:--:-- 87975
  url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 52.85.158.22
  remote_port: 443
  size_download: 69899433 B
  speed_download: 58181.000 B/s
  time_appconnect: 2.297713 s
  time_connect: 1.904176 s
  time_namelookup: 1.727602 s
  time_pretransfer: 2.297974 s
  time_redirect: 0.000000 s
  time_starttransfer: 2.503263 s
  time_total: 1201.408929 s
#+END_EXAMPLE


Well. As you see, the network in China is both slow and unstable.

Message #136 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Giovanni Biscuolo <g <at> xelera.eu>
To: Mark H Weaver <mhw <at> netris.org>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Wed, 19 Dec 2018 13:40:19 +0100

[Message part 1 (text/plain, inline)]

Hi Mark,

sorry for the late reply

Mark H Weaver <mhw <at> netris.org> writes:

> Giovanni Biscuolo <g <at> xelera.eu> writes:
>> with a solid infrastructure of "scientifically" trustable build farms,
>> there are no reasons not to trust substitutes servers (this implies
>> working towards 100% reproducibility of GuixSD)
>
> What does "scientifically trustable" mean?

I'm still not able to elaborate on that (working on it, a sort of
self-research-hack project) but I'm referencing to this message related
to reduced bootstrap tarballs:

  https://lists.gnu.org/archive/html/guix-devel/2018-11/msg00347.html

and the related reply by Jeremiah (unfortunately cannot find it in
archives, Message-ID: <877eh81tm4.fsf <at> ITSx01.pdp10.guru>)

in particular Jeremiah replied this:

--8<---------------cut here---------------start------------->8---
> so, if I don't get it wrong, every skilled engineer will be able to
> build an "almost analogic" (zero bit of software preloaded) computing
> machine ad use stage0/mes [1] as the "metre" [2] to calibrate all other
> computing machines (thanks to reproducible builds)?

well, I haven't thought of it in those terms but yes I guess that is one
of the properties of the plan.
--8<---------------cut here---------------end--------------->8---

and

--8<---------------cut here---------------start------------->8---
> so, having the scientific proof that binary conforms to source, there
> will be noo need to trust (the untrastable)

Well, that is what someone else could do with it but not a direct goal
of the work.
--8<---------------cut here---------------end--------------->8---

maybe a more correct definition of the above "scientific proof" should be
"mathematical proof"

I lack the theoretical basis to be more precise now, sorry :-S

a marketing-like campaign sould be "no more trusting trust"

best regards
Giovanni

-- 
Giovanni Biscuolo

Xelera IT Infrastructures

[signature.asc (application/pgp-signature, inline)]

Message #139 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Mark H Weaver <mhw <at> netris.org>
To: Giovanni Biscuolo <g <at> xelera.eu>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Trustworthiness of build farms (was Re: CDN performance)
Date: Thu, 20 Dec 2018 19:23:47 -0500

Hi Giovanni,

Giovanni Biscuolo <g <at> xelera.eu> writes:

> Mark H Weaver <mhw <at> netris.org> writes:
>
>> Giovanni Biscuolo <g <at> xelera.eu> writes:
>>> with a solid infrastructure of "scientifically" trustable build farms,
>>> there are no reasons not to trust substitutes servers (this implies
>>> working towards 100% reproducibility of GuixSD)
>>
>> What does "scientifically trustable" mean?
>
> I'm still not able to elaborate on that (working on it, a sort of
> self-research-hack project) but I'm referencing to this message related
> to reduced bootstrap tarballs:
>
>   https://lists.gnu.org/archive/html/guix-devel/2018-11/msg00347.html
>
> and the related reply by Jeremiah (unfortunately cannot find it in
> archives, Message-ID: <877eh81tm4.fsf <at> ITSx01.pdp10.guru>)
>
> in particular Jeremiah replied this:
>
>> so, if I don't get it wrong, every skilled engineer will be able to
>> build an "almost analogic" (zero bit of software preloaded) computing
>> machine ad use stage0/mes [1] as the "metre" [2] to calibrate all other
>> computing machines (thanks to reproducible builds)?
>
> well, I haven't thought of it in those terms but yes I guess that is one
> of the properties of the plan.
>
>
> and
>
>> so, having the scientific proof that binary conforms to source, there
>> will be noo need to trust (the untrastable)
>
> Well, that is what someone else could do with it but not a direct goal
> of the work.
>
> maybe a more correct definition of the above "scientific proof" should be
> "mathematical proof"

I agree that a mathematical proof is what we should be aiming for, and
the only kind of proof that I could trust in, in this scenario.

However, it's important to note several caveats:

* A mathematical proof showing that the binary conforms to the source
  requires a proof of correctness of the language implementation, which
  in turn requires formal semantics for both the source language and the
  underlying machine code.  As far as I know, the current
  bootstrappable.org effort does not include anything like this.
  Existing projects that provide this include CakeML and Jitawa.

* One must assume that the hardware behaves according to its formal
  specification.  The veracity of this assumption is not something we
  can currently verify, and even if we could, it would be invalidated if
  someone gained physical access to the machine and modified it.

* The hardware initialization program (e.g. coreboot) and the kernel
  used to perform the bootstrap must still be trusted, and unless I'm
  mistaken, the bootstrappable.org effort does not currently address
  these issues.

* The software running on the substitute servers could be compromised by
  stealing SSH keys from someone with root access.

* If the private signing key of the substitute server can be stolen,
  e.g. by gaining physical access to the machine for a short time, then
  a man-in-the-middle can deliver to users compromised binaries that
  appear to come from the substitute server itself.

* Not only the substitute server, but also all of its build slaves, must
  be trusted as well.

So, while I certainly agree that it will be a *major* improvement to
avoid the need to blindly trust precompiled bootstrap binaries, we
should be clear that the current efforts fall far short of a proof, and
there still remain several valid reasons not to place one's trust in
substitute servers.

Does that make sense?

    Regards,
      Mark

Message #142 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Chris Marusich <cmmarusich <at> gmail.com>
To: Meiyo Peng <meiyo.peng <at> gmail.com>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Fri, 21 Dec 2018 02:22:18 -0800

[Message part 1 (text/plain, inline)]

Hi Meiyo,

Thank you for sharing this information with us!

Can you also share what numbers you get when you run measure_get against
berlin.guixsd.org directly?  Clearly, the connection from you to
CloudFront is not as performant as it is for others in other parts of
the world, but I wonder if it's still better than accessing berlin
directly.  If you could run measure_get against berlin directly and
share the numbers, we can see if it represents any significant
improvement for you.

Meiyo Peng <meiyo.peng <at> gmail.com> writes:

> I tested your script several times.
>
> 1. Tested today at home. China Unicom home broadband. 50Mb/s.
>
> The result is slow as usual. curl failed once.
> berlin-mirror.marusich.info is resolved to Seattle, WA, US.

Well, that's not great.  Perhaps it's still better than it would be if
the DNS name resolved to a location in Europe, though.

> #+BEGIN_EXAMPLE
>   ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>     % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
>                                    Dload  Upload   Total   Spent    Left  Speed
>    55 66.6M   55 36.9M    0     0  17926      0  1:04:59  0:36:02  0:28:57 17733
>   url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>   http_code: 200
>   num_connects: 1
>   num_redirects: 0
>   remote_ip: 52.85.158.151
>   remote_port: 443
>   size_download: 38764357 B
>   speed_download: 17926.000 B/s
>   time_appconnect: 6.078850 s
>   time_connect: 3.006821 s
>   time_namelookup: 2.659785 s
>   time_pretransfer: 6.079097 s
>   time_redirect: 0.000000 s
>   time_starttransfer: 9.626001 s
>   time_total: 2162.379211 s
>   curl: (92) HTTP/2 stream 0 was not closed cleanly: INTERNAL_ERROR (err 2)

I see, this is about 143 Kbps (not Mbps) of throughput, and 347 ms after
DNS name resolution.

>   ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>     % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
>                                    Dload  Upload   Total   Spent    Left  Speed
>   100 66.6M  100 66.6M    0     0   109k      0  0:10:25  0:10:25 --:--:--  241k
>   url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>   http_code: 200
>   num_connects: 1
>   num_redirects: 0
>   remote_ip: 52.85.158.22
>   remote_port: 443
>   size_download: 69899433 B
>   speed_download: 111816.000 B/s
>   time_appconnect: 3.507528 s
>   time_connect: 2.650373 s
>   time_namelookup: 2.261801 s
>   time_pretransfer: 3.507637 s
>   time_redirect: 0.000000 s
>   time_starttransfer: 5.995298 s
>   time_total: 625.129571 s
>
>   ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>     % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
>                                    Dload  Upload   Total   Spent    Left  Speed
>   100 66.6M  100 66.6M    0     0   109k      0  0:10:23  0:10:23 --:--:--  141k
>   url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>   http_code: 200
>   num_connects: 1
>   num_redirects: 0
>   remote_ip: 52.85.158.22
>   remote_port: 443
>   size_download: 69899433 B
>   speed_download: 112187.000 B/s
>   time_appconnect: 2.280972 s
>   time_connect: 1.407197 s
>   time_namelookup: 1.056180 s
>   time_pretransfer: 2.281234 s
>   time_redirect: 0.000000 s
>   time_starttransfer: 3.167703 s
>   time_total: 623.061584 s
> #+END_EXAMPLE

897 Kbps, 351 ms after the name lookup.

> 2. Tested 3 days ago at my office. China Telecom enterprise broadband. 50Mb/s.
>
> Unusually fast! berlin-mirror.marusich.info is resolved to Seattle, WA,
> US. I have no idea why it's so fast that day.
>
> #+BEGIN_EXAMPLE
>   ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>     % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
>                                    Dload  Upload   Total   Spent    Left  Speed
>   100 66.6M  100 66.6M    0     0  1364k      0  0:00:50  0:00:50 --:--:-- 1352k
>   url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>   http_code: 200
>   num_connects: 1
>   num_redirects: 0
>   remote_ip: 13.35.20.109
>   remote_port: 443
>   size_download: 69899433 B
>   speed_download: 1397429.000 B/s
>   time_appconnect: 2.432387 s
>   time_connect: 0.200842 s
>   time_namelookup: 0.000446 s
>   time_pretransfer: 2.432659 s
>   time_redirect: 0.000000 s
>   time_starttransfer: 2.673045 s
>   time_total: 50.020945 s

11 Mbps, 200 ms after the name lookup.

>   ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>     % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
>                                    Dload  Upload   Total   Spent    Left  Speed
>   100 66.6M  100 66.6M    0     0  1592k      0  0:00:42  0:00:42 --:--:-- 2506k
>   url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>   http_code: 200
>   num_connects: 1
>   num_redirects: 0
>   remote_ip: 13.35.20.109
>   remote_port: 443
>   size_download: 69899433 B
>   speed_download: 1630687.000 B/s
>   time_appconnect: 0.653270 s
>   time_connect: 0.209455 s
>   time_namelookup: 0.001582 s
>   time_pretransfer: 0.658399 s
>   time_redirect: 0.000000 s
>   time_starttransfer: 0.883126 s
>   time_total: 42.865868 s
> #+END_EXAMPLE

13 Mbps, 208 ms after the name lookup.

> 3. Tested today at my office. China Telecom enterprise broadband. 50Mb/s.
>
> Slow as usual. berlin-mirror.marusich.info is still resolved to Seattle,
> WA, US. I killed the program several times because it hung there with no
> data transfer for a few minutes. The TCP connection was probably closed
> by GFW. This is very common here.
>
> #+BEGIN_EXAMPLE
>   ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>     % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
>                                    Dload  Upload   Total   Spent    Left  Speed
>   100 66.6M  100 66.6M    0     0  48110      0  0:24:12  0:24:12 --:--:-- 41808
>   url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>   http_code: 200
>   num_connects: 1
>   num_redirects: 0
>   remote_ip: 52.85.158.151
>   remote_port: 443
>   size_download: 69899433 B
>   speed_download: 48110.000 B/s
>   time_appconnect: 0.872926 s
>   time_connect: 0.282048 s
>   time_namelookup: 0.000524 s
>   time_pretransfer: 0.873099 s
>   time_redirect: 0.000000 s
>   time_starttransfer: 1.187467 s
>   time_total: 1452.904154 s

385 Kbps, 282 ms.

>   ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>     % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
>                                    Dload  Upload   Total   Spent    Left  Speed
>     2 66.6M    2 1809k    0     0   5760      0  3:22:15  0:05:21  3:16:54     0^C%
>
>   ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>     % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
>                                    Dload  Upload   Total   Spent    Left  Speed
>    52 66.6M   52 34.9M    0     0  16772      0  1:09:27  0:36:26  0:33:01     0^C%
>
>   ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>     % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
>                                    Dload  Upload   Total   Spent    Left  Speed
>   100 66.6M  100 66.6M    0     0  58181      0  0:20:01  0:20:01 --:--:-- 87975
>   url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
>   http_code: 200
>   num_connects: 1
>   num_redirects: 0
>   remote_ip: 52.85.158.22
>   remote_port: 443
>   size_download: 69899433 B
>   speed_download: 58181.000 B/s
>   time_appconnect: 2.297713 s
>   time_connect: 1.904176 s
>   time_namelookup: 1.727602 s
>   time_pretransfer: 2.297974 s
>   time_redirect: 0.000000 s
>   time_starttransfer: 2.503263 s
>   time_total: 1201.408929 s

465 Kbps, 177 ms.

-- 
Chris

[signature.asc (application/pgp-signature, inline)]

Message #145 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Meiyo Peng <meiyo.peng <at> gmail.com>
To: Chris Marusich <cmmarusich <at> gmail.com>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Sat, 22 Dec 2018 00:04:12 +0800

Hi Chris,

Thank you for your patience!

Chris Marusich <cmmarusich <at> gmail.com> writes:

> Can you also share what numbers you get when you run measure_get against
> berlin.guixsd.org directly?  Clearly, the connection from you to
> CloudFront is not as performant as it is for others in other parts of
> the world, but I wonder if it's still better than accessing berlin
> directly.  If you could run measure_get against berlin directly and
> share the numbers, we can see if it represents any significant
> improvement for you.

1. Tested today at home. China Unicom home broadband. 50Mb/s.

berlin.guixsd.org:

#+BEGIN_EXAMPLE
  ➜  ~ measure_get https://berlin.guixsd.org/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
   54 66.6M   54 36.3M    0     0  14981      0  1:17:45  0:42:25  0:35:20     0
  url_effective: https://berlin.guixsd.org/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 141.80.181.40
  remote_port: 443
  size_download: 38141765 B
  speed_download: 14981.000 B/s
  time_appconnect: 3.228601 s
  time_connect: 2.213136 s
  time_namelookup: 0.856194 s
  time_pretransfer: 3.228820 s
  time_redirect: 0.000000 s
  time_starttransfer: 3.851583 s
  time_total: 2545.889968 s
  curl: (56) GnuTLS recv error (-54): Error in the pull function.

  ➜  ~ measure_get https://berlin.guixsd.org/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
  100 66.6M  100 66.6M    0     0  20415      0  0:57:03  0:57:03 --:--:-- 25983
  url_effective: https://berlin.guixsd.org/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 141.80.181.40
  remote_port: 443
  size_download: 69899433 B
  speed_download: 20415.000 B/s
  time_appconnect: 2.005881 s
  time_connect: 0.785257 s
  time_namelookup: 0.000520 s
  time_pretransfer: 2.006124 s
  time_redirect: 0.000000 s
  time_starttransfer: 3.031582 s
  time_total: 3423.813489 s
#+END_EXAMPLE

berlin-mirror.marusich.info:

#+BEGIN_EXAMPLE
  ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
  100 66.6M  100 66.6M    0     0  1470k      0  0:00:46  0:00:46 --:--:-- 2368k
  url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 13.35.20.87
  remote_port: 443
  size_download: 69899433 B
  speed_download: 1505934.000 B/s
  time_appconnect: 3.343496 s
  time_connect: 3.164926 s
  time_namelookup: 3.060655 s
  time_pretransfer: 3.343581 s
  time_redirect: 0.000000 s
  time_starttransfer: 5.766543 s
  time_total: 46.416495 s

  ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
  100 66.6M  100 66.6M    0     0  3182k      0  0:00:21  0:00:21 --:--:-- 4612k
  url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 13.35.20.87
  remote_port: 443
  size_download: 69899433 B
  speed_download: 3259170.000 B/s
  time_appconnect: 0.225982 s
  time_connect: 0.070428 s
  time_namelookup: 0.000483 s
  time_pretransfer: 0.226055 s
  time_redirect: 0.000000 s
  time_starttransfer: 0.306621 s
  time_total: 21.447966 s
#+END_EXAMPLE


2. Tested today at my office. China Telecom enterprise broadband. 50Mb/s.

berlin.guixsd.org:

#+BEGIN_EXAMPLE
  ➜  ~ measure_get https://berlin.guixsd.org/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
  100 66.6M  100 66.6M    0     0  3091k      0  0:00:22  0:00:22 --:--:-- 3649k
  url_effective: https://berlin.guixsd.org/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 141.80.181.40
  remote_port: 443
  size_download: 69899433 B
  speed_download: 3166021.000 B/s
  time_appconnect: 3.288213 s
  time_connect: 2.733554 s
  time_namelookup: 2.486754 s
  time_pretransfer: 3.288320 s
  time_redirect: 0.000000 s
  time_starttransfer: 3.780341 s
  time_total: 22.078489 s

  ➜  ~ measure_get https://berlin.guixsd.org/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
  100 66.6M  100 66.6M    0     0  3499k      0  0:00:19  0:00:19 --:--:-- 4011k
  url_effective: https://berlin.guixsd.org/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 141.80.181.40
  remote_port: 443
  size_download: 69899433 B
  speed_download: 3583667.000 B/s
  time_appconnect: 0.761166 s
  time_connect: 0.244415 s
  time_namelookup: 0.000981 s
  time_pretransfer: 0.761275 s
  time_redirect: 0.000000 s
  time_starttransfer: 1.247935 s
  time_total: 19.505515 s
#+END_EXAMPLE

berlin-mirror.marusich.info:

#+BEGIN_EXAMPLE
  ➜  ~ measure_get https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
    % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                   Dload  Upload   Total   Spent    Left  Speed
    0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0
   49 66.6M   49 32.8M    0     0  19012      0  1:01:16  0:30:13  0:31:03 29265
  url_effective: https://berlin-mirror.marusich.info/nar/gzip/1bq783rbkzv9z9zdhivbvfzhsz2s5yac-linux-libre-4.19
  http_code: 200
  num_connects: 1
  num_redirects: 0
  remote_ip: 52.85.158.22
  remote_port: 443
  size_download: 34488133 B
  speed_download: 19012.000 B/s
  time_appconnect: 2.958899 s
  time_connect: 2.487483 s
  time_namelookup: 2.271520 s
  time_pretransfer: 2.959321 s
  time_redirect: 0.000000 s
  time_starttransfer: 5.447693 s
  time_total: 1813.938029 s
  curl: (92) HTTP/2 stream 0 was not closed cleanly: INTERNAL_ERROR (err 2)
#+END_EXAMPLE


Although both 13.35.20.0/24 and 52.85.158.0/24 IP ranges are located at Seattle,
the result shows that the connection to 13.35.20.0/24 is significantly faster.
This is pretty normal in China.  It's definitely caused by the GFW.  Giant
internet service providers (e.g. AWS) are the primary targets of the GFW.

Message #148 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Marius Bakke <mbakke <at> fastmail.com>
To: Giovanni Biscuolo <g <at> xelera.eu>, Mark H Weaver <mhw <at> netris.org>
Cc: guix-devel <at> gnu.org, 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Fri, 21 Dec 2018 21:47:02 +0100

[Message part 1 (text/plain, inline)]

Giovanni Biscuolo <g <at> xelera.eu> writes:

> Mark H Weaver <mhw <at> netris.org> writes:
>
>> Giovanni Biscuolo <g <at> xelera.eu> writes:
>>> with a solid infrastructure of "scientifically" trustable build farms,
>>> there are no reasons not to trust substitutes servers (this implies
>>> working towards 100% reproducibility of GuixSD)
>>
>> What does "scientifically trustable" mean?
>
> I'm still not able to elaborate on that (working on it, a sort of
> self-research-hack project) but I'm referencing to this message related
> to reduced bootstrap tarballs:
>
>   https://lists.gnu.org/archive/html/guix-devel/2018-11/msg00347.html
>
> and the related reply by Jeremiah (unfortunately cannot find it in
> archives, Message-ID: <877eh81tm4.fsf <at> ITSx01.pdp10.guru>)

FWIW the GNU list search can take message IDs:

https://lists.gnu.org/archive/cgi-bin/namazu.cgi?query=877eh81tm4.fsf%40ITSx01.pdp10.guru&submit=Search&idxname=guix-devel

[signature.asc (application/pgp-signature, inline)]

Message #151 received at 33600 <at> debbugs.gnu.org (full text, mbox):

From: Ricardo Wurmus <rekado <at> elephly.net>
To: Giovanni Biscuolo <g <at> xelera.eu>
Cc: guix-devel <at> gnu.org, Ludovic Courtès <ludo <at> gnu.org>,
 Chris Marusich <cmmarusich <at> gmail.com>, 33600 <at> debbugs.gnu.org
Subject: Re: CDN performance
Date: Mon, 24 Dec 2018 15:47:01 +0100

Hi Giovanni,

> for this very reason IMHO we should work towards a network of **very
> trusted** build farms directly managed and controlled by the GuixSD
> project sysadmins; if build farms will be able to quickly provide
> substitutes, caching mirrors will be _much more_ effective than today
>
> ... and a network of "automated guix challenge" servers to spot
> not-reproducible software in GuixSD
>
> with a solid infrastructure of "scientifically" trustable build farms,
> there are no reasons not to trust substitutes servers (this implies
> working towards 100% reproducibility of GuixSD)

This sets the bar very high.  I administer berlin.guix.info /
ci.guix.info (same server) and most of the associated build nodes, but I
don’t think people should trust these computers more than their own
computers or those managed by people they personally know.

The build servers do the same work that a user would do who builds
software locally without substitutes; the builders are supposed to
behave just like any other user.  (And we can challenge their authority
with “guix challenge”.)  I want us to keep in mind that build farms
don’t necessarily deserve any more trust than other computers you don’t
control.  Substitute servers are a convenience.

To improve distribution of binaries we are committed to working on
different strategies at the same time:

- improve our processes so that non-critical package changes only hit
  the master branch when we have already built the package and made it
  available for distribution.

- improve availability of the single server berlin.guix.info by hooking
  it up to a CDN via the name ci.guix.info (from which users can easily
  opt out by changing their default substitute server to
  berlin.guix.info).

- improve redundancy by setting up an off-site fail-over for the head of
  the build farm at berlin.guix.info (such as bayfront).

- distribute build artefacts over IPFS without requiring a central
  authority

All of these things can be done in parallel by different people.  This
increases the project’s resilience and the options that users can choose
from.

--
Ricardo

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.