Gonzalo PG wrote:
Hello,
I have running Squid Cache: Version 3.0.STABLE13 with ntlm auth, using
samba-3.2.10 and winbind, also SquidGuard 1.4
Since a few days I am detecting that squid is going down (then restart
again without any problem) with the error:
2009/05/06 12:59:33| assertion failed: comm.cc:572: "fdc_table[fd].active == 1"
[2009/05/06 12:59:34, 1] utils/ntlm_auth.c:manage_squid_request(2167)
fgets() failed! dying..... errno=1 (Operation not permitted)
I really don't know what`s going wrong. It usually occurs when I do a
squid reload. If you need more information please say to me. The
complete cache.log of the error is:
2009/05/06 12:59:16| storeDirWriteCleanLogs: Starting...
2009/05/06 12:59:16| 65536 entries written so far.
...
2009/05/06 12:59:17| 1245184 entries written so far.
2009/05/06 12:59:17| Finished. Wrote 1280543 entries.
2009/05/06 12:59:17| Took 0.86 seconds (1489072.75 entries/sec).
2009/05/06 12:59:17| logfileRotate: /var/log/squid/access.log
2009/05/06 12:59:17| Pinger socket opened on FD 11
2009/05/06 12:59:17| helperOpenServers: Starting 30 'squidGuard' processes
2009/05/06 12:59:19| helperStatefulOpenServers: Starting 50
'ntlm_auth' processes
2009/05/06 12:59:24| helperOpenServers: Starting 5 'ntlm_auth' processes
2009/05/06 12:59:24| helperOpenServers: Starting 10 'wbinfo_group.pl' processes
2009/05/06 12:59:25| Reconfiguring Squid Cache (version 3.0.STABLE13)...
2009/05/06 12:59:25| FD 121 Closing HTTP connection
2009/05/06 12:59:25| Closing Pinger socket on FD 11
2009/05/06 12:59:25| Processing Configuration File:
/etc/squid/squid.conf (depth 0)
2009/05/06 12:59:25| cache_cf.cc(346) squid.conf:122 unrecognized: ' #'
2009/05/06 12:59:25| Initializing https proxy context
2009/05/06 12:59:25| Store logging disabled
2009/05/06 12:59:25| User-Agent logging is disabled.
2009/05/06 12:59:25| Referer logging is disabled.
2009/05/06 12:59:25| DNS Socket created at 0.0.0.0, port 42477, FD 8
2009/05/06 12:59:25| Adding nameserver 172.28.1.90 from /etc/resolv.conf
2009/05/06 12:59:25| Adding nameserver 172.28.1.91 from /etc/resolv.conf
2009/05/06 12:59:25| Adding domain iipp.int from /etc/resolv.conf
2009/05/06 12:59:25| helperOpenServers: Starting 30 'squidGuard' processes
2009/05/06 12:59:28| helperStatefulOpenServers: Starting 50
'ntlm_auth' processes
2009/05/06 12:59:32| helperOpenServers: Starting 5 'ntlm_auth' processes
2009/05/06 12:59:32| helperOpenServers: Starting 10 'wbinfo_group.pl' processes
2009/05/06 12:59:33| Accepting HTTP connections at 0.0.0.0, port 3128, FD 181.
2009/05/06 12:59:33| HTCP Disabled.
2009/05/06 12:59:33| Pinger socket opened on FD 216
2009/05/06 12:59:33| Configuring Parent 192.168.113.4/8080/0
2009/05/06 12:59:33| Loaded Icons.
2009/05/06 12:59:33| Ready to serve requests.
2009/05/06 12:59:33| assertion failed: comm.cc:572: "fdc_table[fd].active == 1"
[2009/05/06 12:59:34, 1] utils/ntlm_auth.c:manage_squid_request(2167)
fgets() failed! dying..... errno=1 (Operation not permitted)
[2009/05/06 12:59:34, 1] utils/ntlm_auth.c:manage_squid_request(2167)
fgets() failed! dying..... errno=1 (Operation not permitted)
[2009/05/06 12:59:34, 1] utils/ntlm_auth.c:manage_squid_request(2167)
fgets() failed! dying..... errno=1 (Operation not permitted)
2009/05/06 12:59:37| Starting Squid Cache version 3.0.STABLE13 for
i686-suse-linux-gnu...
Thank you for this report.
Firstly, can you replicate this with 3.0.STABLE15? There have been a few
ntlm_auth handling fixes recently.
By 'usually occurs when I do a squid reload' do you have any actual
trace from it happening without a reload operation? And it there any
difference visible?
I suspect this is another side effect of an FD issue we are aware of
that occurs only during reconfigure and shutdown. The fix for that one
is taking a while and may need to be a part of 3.1.
Amos
--
Please be using
Current Stable Squid 2.7.STABLE6 or 3.0.STABLE15
Current Beta Squid 3.1.0.7
