GNU bug report logs - #38555
shepherd not working out of the box in Debian

Previous Next

To reply to this bug, email your comments to 38555 AT debbugs.gnu.org.

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

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

From: burban <at> fdn.fr
To: bug-guix <at> gnu.org
Subject: shepherd not working out of the box in Debian
Date: Mon, 09 Dec 2019 21:51:41 +0100

I run shepherd 0.6.1 (with guile 2.2.4) as the init system for Debian
buster. It's an installation in the standard place /usr/local.

When /usr/local/var/run/shepherd/socket is left dangling due to a crash,
shepherd won't start on the next reboot, as error SOCADDRINUSE is
raised.

Obviously this won't happen if --prefix is / and /var/run points to a
volatile file system, but chosing that prefix is probably not what
someone trying to test shepherd will do.

Below is my patch:

*** shepherd.scm	2019-12-08 18:11:19.820000000 +0000
--- shepherd.scm-ORI	2019-04-18 10:39:12.000000000 +0100
***************
*** 46,51 ****
        (fcntl sock F_SETFL (logior O_NONBLOCK
                                    (fcntl sock F_GETFL)))
-       ;; if file-name is still there, SOCADDRINUSE error is triggered
-       (when (file-exists? file-name) (delete-file file-name))
        (bind sock address)
        (listen sock 10)
--- 46,49 ----
***************
*** 186,191 ****
                     (%current-logfile-date-format
                      (if (and (not logfile) (zero? (getuid)))
! 			;; for syslog, first 2 characters are truncated
!                         (format #f "  shepherd[~d]: " (getpid))
                          default-logfile-date-format))
                     (current-output-port
--- 184,188 ----
                     (%current-logfile-date-format
                      (if (and (not logfile) (zero? (getuid)))
!                         (format #f "shepherd[~d]: " (getpid))
                          default-logfile-date-format))
                     (current-output-port


Notice the second patch, which corrects these kind of output in
/var/syslog:

Dec  9 20:18:18 buster cron[567]: (CRON) STARTUP (fork ok)
Dec  9 20:18:18 buster cron[567]: (CRON) INFO (Running @reboot jobs)
Dec  9 20:18:18 buster epherd[1]: Service sshd has been started.
Dec  9 20:18:18 buster epherd[1]: Service hwclock is already running.
Dec  9 20:18:18 buster epherd[1]: Service sendsigs is already running.
Dec  9 20:18:18 buster epherd[1]: Service hostname is already running.
Dec  9 20:18:18 buster epherd[1]: Service mountall-bootclean is already running.
Dec  9 20:18:19 buster rsyslogd:  [origin software="rsyslogd" swVersion="8.1901.0" x-pid="556" x-info="https://www.rsyslog.com"] start
Dec  9 20:23:18 buster anacron[562]: Job `cron.daily' started
Dec  9 20:23:18 buster anacron[625]: Updated timestamp for job `cron.daily' to 2019-12-09

-- 

Bernard

Previous Next

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