squid failed to restart after update to squid 2.7.STABLE7-1ubuntu12.2

[pieter]

Binary package hint: squid

Update crashes, and asks me to report it.

$ lsb_release -rd
Description: Ubuntu 10.04.2 LTS
Release: 10.04

$ apt-cache policy squid
squid:
  Installed: 2.7.STABLE7-1ubuntu12.2
  Candidate: 2.7.STABLE7-1ubuntu12.2
  Version table:
 *** 2.7.STABLE7-1ubuntu12.2 0
        500 http://nl.archive.ubuntu.com/ubuntu/ lucid-updates/main Packages
        100 /var/lib/dpkg/status
     2.7.STABLE7-1ubuntu12 0
        500 http://nl.archive.ubuntu.com/ubuntu/ lucid/main Packages

ProblemType: Package
DistroRelease: Ubuntu 10.04
Package: squid 2.7.STABLE7-1ubuntu12.2
ProcVersionSignature: Ubuntu 2.6.32-29.58-generic 2.6.32.28+drm33.13
Uname: Linux 2.6.32-29-generic x86_64
NonfreeKernelModules: nvidia
Architecture: amd64
Date: Fri Mar 4 20:04:13 2011
ErrorMessage: subprocess installed post-installation script returned error exit status 1
InstallationMedia: Ubuntu 9.10 "Karmic Koala" - Release amd64 (20091027)
SourcePackage: squid
Title: package squid 2.7.STABLE7-1ubuntu12.2 failed to install/upgrade: subprocess installed post-installation script returned error exit status 1

[Dave Walker (davewalker)]

Thanks for reporting this bug.

Package version comparison in, squid-2.7.STABLE7/debian/preinst:
  dpkg --compare-versions "$2" lt '2.7.STABLE7-1ubuntu12.1' \

Should have been, 2.7.STABLE7-1ubuntu12.2

Marking confirmed.

[Dave Walker (davewalker)]

Hmm, maybe not..

This actually ran,
$ dpkg --compare-versions "2.7.STABLE7-1ubuntu12" lt '2.7.STABLE7-1ubuntu12.1'

Which should have correctly killed squid before the upgrade, reverting back to incomplete.

[Dave Walker (davewalker)]

As others have reported this same situation, it seems like a bug which can be confirmed. I believe Clint is currently try to triage this.

Thanks.

[Clint Byrum (clint-fewbar)]

I've been unable to reproduce this, but there are a ton of duplicates so there is no doubt that there is something wrong with the lucid updates.

I actually think this is directly related to bug #736348 . By calling 'restart' directly, we are not re-loading the new job file. So until the system is rebooted, squid will still be run with 'expect fork' and without -N. I am suspicious that this is just exposing *more* issues with the way that squid forks, since the bug we were originally trying to fix involves how squid handles SIGHUP, its conceivable that it handles SIGTERM in a way that is incompatible w/ upstart's expect fork as well.

I'm not ready to mark this as a duplicate of that bug, but I think it may be. I'm hoping somebody can provide us with some more information regarding the state of their system *before* the upgrade, that may suggest what state causes the restart to fail. In the mean time, it makes sense to get that fix into natty for testing, especially as people start to try natty out as it approaches beta.

It also seems at least one person has hit upstart bug #406397 , which also suggests further that this is caused by mixing expect fork with daemons that don't fork in the way upstart suggests. Unfortunately that is a particularly nasty bug whose only workaround is to make a program that forks until the pid that has gone missing can be recreated.

[Clint Byrum (clint-fewbar)]

For all affected parties, while we figure out what has caused this exactly, if you could possibly provide your /etc/squid/squid.conf , and the logs from /var/log/squid/cache.log and the output of 'grep init: /var/log/daemon.log', that would be extremely helpful in debugging the issue. Please be careful when uploading these, as this is a public bug report, and those files all may have sensitive information.

[Clint Byrum (clint-fewbar)]

Oops, I meant bug #726348 , not bug #736348

[jdiasn (jdiasn)]

I have same problem, In cache.log I found it

FATAL: Bungled squid.conf line 176: http_reply_access allow all
Squid Cache (Version 2.7.STABLE7): Terminated abnormally.

https://bugs.launchpad.net/bugs/730041

What do I now?

[jdiasn (jdiasn)]

opps, What do I do?

[Clint Byrum (clint-fewbar)]

jdiasn, in that case you actually have an error in your squid.conf.. you need to fix that or squid won't be able to start.

Could all those affected check their /var/log/squid/cache.log and make sure its not just a config problem?

[jdiasn (jdiasn)]

Clint Byrum, how fix I this problem?
Can you help me?

[Clint Byrum (clint-fewbar)]

jdiasn, This is an issue with your configuration file. You'll need to read squid's manual (there is a lot of help in the squid.conf itself) to know how its supposed to work. The error message says that on line 176 there's an error, so try fixing that line.

[pieter (p-maljaars)]

My cache.log doesn't show errors. Only a few warnings (I think at shutdown
or hibernate):

2011/03/25 20:10:10| Beginning Validation Procedure
2011/03/25 20:10:10| Completed Validation Procedure
2011/03/25 20:10:10| Validated 1151 Entries
2011/03/25 20:10:10| store_swap_size = 90128k
2011/03/25 20:10:11| storeLateRelease: released 0 objects
2011/03/25 20:47:00| WARNING: url_rewriter #1 (FD 7) exited
2011/03/25 20:47:00| WARNING: url_rewriter #3 (FD 9) exited
2011/03/25 20:47:00| WARNING: url_rewriter #4 (FD 10) exited
2011/03/25 20:47:00| Too few url_rewriter processes are running
2011/03/25 20:47:00| Starting new helpers
2011/03/25 20:47:00| helperOpenServers: Starting 5 'squidGuard' processes
2011/03/25 20:47:00| WARNING: url_rewriter #5 (FD 12) exited
2011/03/25 20:47:00| WARNING: url_rewriter #2 (FD 8) exited

[Marius (mariusm)]

My cache.log surrounding the time when the update failed (2011/04/09 08:32:11):

2011/04/09 06:48:00| storeDirWriteCleanLogs: Starting...
2011/04/09 06:48:00| Finished. Wrote 4384 entries.
2011/04/09 06:48:00| Took 0.0 seconds (2124031.0 entries/sec).
2011/04/09 06:48:00| logfileRotate: /var/log/squid/store.log
2011/04/09 06:48:00| logfileRotate (stdio): /var/log/squid/store.log
2011/04/09 06:48:00| logfileRotate: /var/log/squid/access.log
2011/04/09 06:48:00| logfileRotate (stdio): /var/log/squid/access.log
2011/04/09 06:48:00| helperOpenServers: Starting 5 'url_rewriter.py' processes
2011/04/09 08:30:27| WARNING: All url_rewriter processes are busy.
2011/04/09 08:30:27| WARNING: up to 5 pending requests queued
2011/04/09 08:32:11| WARNING: url_rewriter #4 (FD 14) exited
2011/04/09 08:32:11| WARNING: url_rewriter #5 (FD 19) exited
2011/04/09 08:32:11| WARNING: url_rewriter #1 (FD 10) exited
2011/04/09 08:32:11| Too few url_rewriter processes are running
2011/04/09 08:32:11| Starting new helpers
2011/04/09 08:32:11| helperOpenServers: Starting 5 'url_rewriter.py' processes
2011/04/09 08:32:11| WARNING: url_rewriter #3 (FD 13) exited
2011/04/09 08:32:11| WARNING: url_rewriter #2 (FD 12) exited
2011/04/09 08:32:11| Preparing for shutdown after 1802806 requests
2011/04/09 08:32:11| Waiting 30 seconds for active connections to finish
2011/04/09 08:32:11| FD 16 Closing HTTP connection
2011/04/09 08:32:42| Shutting down...
2011/04/09 08:32:42| FD 17 Closing ICP connection
2011/04/09 08:32:42| Closing unlinkd pipe on FD 18
2011/04/09 08:32:42| storeDirWriteCleanLogs: Starting...
2011/04/09 08:32:42| Finished. Wrote 4579 entries.
2011/04/09 08:32:42| Took 0.1 seconds (55899.4 entries/sec).
CPU Usage: 2697.533 seconds = 1535.696 user + 1161.837 sys
Maximum Resident Size: 107360 KB
Page faults with physical i/o: 109
Memory usage for squid via mallinfo():
        total space in arena: 24712 KB
        Ordinary blocks: 18065 KB 1571 blks
        Small blocks: 0 KB 6 blks
        Holding blocks: 280 KB 1 blks
        Free Small blocks: 0 KB
        Free Ordinary blocks: 6646 KB
        Total in use: 18345 KB 73%
        Total free: 6647 KB 27%
2011/04/09 08:32:42| logfileClose: closing log /var/log/squid/store.log
2011/04/09 08:32:42| logfileClose: closing log /var/log/squid/access.log
2011/04/09 08:32:42| Squid Cache (Version 2.7.STABLE7): Exiting normally.

[Clint Byrum (clint-fewbar)]

Judging by the number of dupe reports, I'm raising this to High, as it seems to affect more than just a small subset of users.

[Majal (majalmirasol)]

I'm new to this forum. :) I have Ubuntu 11.04 and squid 2.7 was freshly installed after DansGuardian2. When I tried to start squid fresh, I get this error message:

maj@Emeth:~$ sudo squid start
2011/07/28 00:15:10| ACL name 'all' not defined!
FATAL: Bungled squid.conf line 176: http_reply_access allow all
Squid Cache (Version 2.7.STABLE9): Terminated abnormally.

[Ralph Corderoy (ralph-inputplus)]

cache.log, attached, shows nothing odd AFAICS, normal shutdown. dist-upgrade failed,

    Preparing to replace squid 2.7.STABLE9-2ubuntu5.1 (using .../squid_2.7.STABLE9-2ubuntu5.2_amd64.deb) ...
    Unpacking replacement squid ...
    Processing triggers for ureadahead ...
    ureadahead will be reprofiled on next reboot
    Processing triggers for ufw ...
    Processing triggers for man-db ...
    Setting up squid-common (2.7.STABLE9-2ubuntu5.2) ...
    Setting up squid (2.7.STABLE9-2ubuntu5.2) ...
    restart: Job failed to restart
    dpkg: error processing squid (--configure):
     subprocess installed post-installation script returned error exit status 1
    Errors were encountered while processing:
     squid
    E: Sub-process /usr/bin/dpkg returned an error code (1)
    $

After finding this bug report I tried

    $ sudo status squid
    squid stop/waiting
    $ sudo start squid
    squid start/running, process 11334
    $ sudo status squid
    squid start/running, process 11334

which is 12:43:30 in the log file. It appears to be working fine now.

[Rolf Leggewie (r0lf)]

lucid has seen the end of its life and is no longer receiving any updates. Marking the lucid task for this ticket as "Won't Fix".

[David Britton (dpb)]

this version is older than what was shipped in precise. Tested on xenial, and updating the package from main -> updates worked just fine. Re-open if you can repro with modern versions.