[squid-users] Squid 3.0 Crashing every few minutes

From: Sam Reynolds <sreynolds@dont-contact.us>
Date: Mon, 13 Jun 2005 12:18:06 -0400

We upgraded to Squid-3.0-20050609, although this may be unrelated (the
fact that we upgraded). We upgraded trying to resolve the fact that the
server is crashing every few minutes. The only thing in the logs when
it crashes is:

Jun 13 11:59:38 xxxxxxxx squid[3608]: storeDirWriteCleanLogs:
Starting...
Jun 13 11:59:38 xxxxxxxx squid[3608]: WARNING: Closing open FD 16
Jun 13 11:59:38 xxxxxxxx squid[3608]: Finished. Wrote 0 entries.
Jun 13 11:59:38 xxxxxxxx squid[3608]: Took 0.0 seconds ( 0.0
entries/sec).

I have attempted to get a core dump to be able to send to analyze, but
so far unsuccessful. I did however set debug to All,9 in the squid.conf
to see what we could see. The following is the output during one of the
times it crashes (a few milliseconds before and after).

2005/06/13 11:54:20.727| storeSwapOut:
http://www.usfood.com/jsp/images/default/ad_09.gif
2005/06/13 11:54:20.727| storeSwapOut: store_status = STORE_PENDING
2005/06/13 11:54:20.727| storeSwapOut: mem->inmem_lo = 0
2005/06/13 11:54:20.727| storeSwapOut: mem->endOffset() = 164
2005/06/13 11:54:20.727| storeSwapOut: swapout.queue_offset = 0
2005/06/13 11:54:20.727| storeSwapOut: lowest_offset = 0
2005/06/13 11:54:20.727| InvokeHandlers:
ACA8796FDDB1AA18E3EC432D69117B60
2005/06/13 11:54:20.727| InvokeHandlers: checking client #0
2005/06/13 11:54:20.727| storeClientCopy2: returning because
ENTRY_FWD_HDR_WAIT set
2005/06/13 11:54:20.727| httpProcessReplyHeader: HTTP CODE: 304
2005/06/13 11:54:20.727| 0x8fe4920 lookup for 49
2005/06/13 11:54:20.727| 0x8fe4920 lookup for 33
2005/06/13 11:54:20.727| storeExpireNow:
'ACA8796FDDB1AA18E3EC432D69117B60'
2005/06/13 11:54:20.727| ctx: exit level 0
2005/06/13 11:54:20.727| httpPconnTransferDone: FD 75
2005/06/13 11:54:20.727| 0x8fe4920 lookup for 37
2005/06/13 11:54:20.727| 0x8fe4920 lookup for 9
2005/06/13 11:54:20.727| 0x8fe4920: joining for id 9
2005/06/13 11:54:20.727| 0x8fe4920: joined for id 9: keep-alive
2005/06/13 11:54:20.727| httpPconnTransferDone: content_length=0
2005/06/13 11:54:20.727| commSetTimeout: FD 75 timeout -1
2005/06/13 11:54:20.727| cbdataUnlock: 0xbf2ef018=2
2005/06/13 11:54:20.728| comm_remove_close_handler: FD 75,
handler=0x8095f78, data=0xbf2ef018
2005/06/13 11:54:20.728| cbdataUnlock: 0xbf2ef018=1
2005/06/13 11:54:20.728| fwdUnregister:
http://www.usfood.com/jsp/images/default/ad_09.gif
2005/06/13 11:54:20.728| comm_remove_close_handler: FD 75,
handler=0x80899dc, data=0x90c6548
2005/06/13 11:54:20.728| cbdataUnlock: 0x90c6548=0
2005/06/13 11:54:20.728| storeDirWriteCleanLogs: Starting...
2005/06/13 11:54:20.728| file_open: FD 16
2005/06/13 11:54:20.728| WARNING: Closing open FD 16
2005/06/13 11:54:20.728| fd_close FD 16 HTTP Socket
2005/06/13 11:54:20.728| commSetSelect: FD 16 type 1
2005/06/13 11:54:20.728| commSetSelect: FD 16 type 2
2005/06/13 11:54:20.728| fd_open FD 16 /var/squid/cache/swap.state.clean
2005/06/13 11:54:20.728| storeDirWriteCleanLogs: opened
/var/squid/cache/swap.state.clean, FD 16

Our proxy is currently set to authenticate users from a Windows 2003
Radius server, before letting them out to the web. The only other thing
that we notice is in the access.log during those times that it crashes
is the following:

"172.18.255.107":;"64.219.172.3":;"172.18.10.200":;1118677549:;2005-06-1
3
11:45:49:224:;245:;"-":;"-":;"200":;"GET":;"http://www.aarc.org/members_
area/help_line/thread.asp?":;"1.0":;2311:;"text/html":;"TCP_MISS"
or
"172.20.9.11":;"-":;"172.18.10.200":;1118678377:;2005-06-13
11:59:37:293:;0:;"-":;"-":;"407":;"GET":;"http://download1.aveo-attune.c
om/downloader/atmdlusr.exe":;"1.0":;2412:;"text/html":;"TCP_DENIED"

A normal one that doesn't crash the proxy looks like this where
"username" is the users login ID:

"172.18.253.9":;"206.24.222.125":;"172.18.10.200":;1118676923:;2005-06-1
3
11:35:23:193:;2448:;"username":;"username":;"200":;"GET":;"http://msn.me
tahost.digisle.tv/msaverizon/broadband_15_mbr.asx":;"1.0":;679:;"video/x
-ms-asf":;"TCP_MISS"

Any thoughts on this?

Thanks in advance.
Received on Mon Jun 13 2005 - 10:19:21 MDT

This archive was generated by hypermail pre-2.1.9 : Fri Jul 01 2005 - 12:00:02 MDT