Re: [squid-users] Re: cache.log Warnings

2014-02-17 Thread Carlos Defoe
http://wiki.squid-cache.org/KnowledgeBase/QueueCongestion

You're probably using aufs, those messages are normal.

On the restart log, I never saw the AuthUserHashPointer ones, but
since squid exits and starts normally, I don't think that is a
problem.

I was going to ask how are you checking your free memory, but I think
you just figured out that you have plenty of memory.

I looked quickly to your conf on the other message, and seems that
you're using a very small cache_mem. With 8GB, you can increase that
to at least 2 GB safely. The hot objects will be kept on RAM, thus
reducing the disk activity and speeding your proxy.

On Mon, Feb 17, 2014 at 9:26 PM, Scott Mayo scotgm...@gmail.com wrote:
 Nevermind on the memory usage.  Looks like it is fine by looking at my
 buffers/cache in free -m.

 I am curious if any of the messages in the log look like something I
 should worry about though.

 Thanks.
 Scott

 On Mon, Feb 17, 2014 at 5:20 PM, Scott Mayo scotgm...@gmail.com wrote:
 On Mon, Feb 17, 2014 at 2:31 PM, Scott Mayo scotgm...@gmail.com wrote:
 Just curious if these are anything that I should really worry about,
 or just need to keep an eye on my log file?

 2014/02/16 03:08:01| helperOpenServers: Starting 40/40
 'squid_ldap_auth' processes
 2014/02/16 03:08:01| helperOpenServers: Starting 5/5
 'squid_ldap_group' processes
 2014/02/17 08:59:13| TunnelStateData::Connection::error: FD 747:
 read/write failure: (32) Broken pipe
 2014/02/17 09:46:05| squidaio_queue_request: WARNING - Queue congestion
 2014/02/17 10:30:41| squidaio_queue_request: WARNING - Queue congestion
 2014/02/17 11:38:04| squidaio_queue_request: WARNING - Queue congestion
 2014/02/17 12:57:40| TunnelStateData::Connection::error: FD 1298:
 read/write failure: (110) Connection timed out
 2014/02/17 13:08:00| squidaio_queue_request: WARNING - Queue congestion
 2014/02/17 14:03:09| TunnelStateData::Connection::error: FD 1000:
 read/write failure: (32) Broken pipe
 2014/02/17 14:07:12| squidaio_queue_request: WARNING - Queue congestion

 I am assuming that I may just need a faster drive?  My network has
 been busing right along today and I have not seen any slowness at all.

 Thanks for any suggestions.


 On top of those few errors, I noticed at the end of school, my free
 memory was down to about 2.5GB out of 8GB.  I restarted squid just to
 see if that would affect anything.  I have a lot more things in my
 cache.log now.  They may not be anything, but I just wanted to ask.  I
 excluded a bit to make it not so long.

 I'll be curious if my memory frees up a bit later. I thought that
 maybe a lot was being used since I changed my auth_param basic
 credentialsttl to 9 hours.  Not sure if that would cause it to hold
 that much info in memory or not as far as logins go.

 Anyways, below is what was in my cache.log when restarting.  Is this normal?

 2014/02/17 16:15:03| Preparing for shutdown after 490522 requests
 2014/02/17 16:15:03| Waiting 30 seconds for active connections to finish
 2014/02/17 16:15:03| FD 105 Closing HTTP connection
 2014/02/17 16:15:35| Shutting down...
 2014/02/17 16:15:35| AuthUserHashPointer::removeFromCache: entry in
 use - not freeing
 2014/02/17 16:15:35| AuthUserHashPointer::removeFromCache: entry in
 use - not freeing
 2014/02/17 16:15:35| AuthUserHashPointer::removeFromCache: entry in
 use - not freeing
 .
 .  ((EXCLUDED ALL THE SAME LINES HERE THAT WERE IN BETWEEN FOR READABILITY))
 .
 2014/02/17 16:15:35| AuthUserHashPointer::removeFromCache: entry in
 use - not freeing
 2014/02/17 16:15:35| AuthUserHashPointer::removeFromCache: entry in
 use - not freeing

 2014/02/17 16:15:35| basic/auth_basic.cc(97) done: Basic
 authentication Shutdown.
 2014/02/17 16:15:35| Closing unlinkd pipe on FD 103
 2014/02/17 16:15:35| storeDirWriteCleanLogs: Starting...
 2014/02/17 16:15:35| 65536 entries written so far.
 2014/02/17 16:15:35|   Finished.  Wrote 95602 entries.
 2014/02/17 16:15:35|   Took 0.03 seconds (2791951.40 entries/sec).
 CPU Usage: 1335.548 seconds = 616.616 user + 718.932 sys
 Maximum Resident Size: 2354176 KB
 Page faults with physical i/o: 0
 Memory usage for squid via mallinfo():
 total space in arena:  572856 KB
 Ordinary blocks:   401955 KB  49716 blks
 Small blocks:   0 KB  7 blks
 Holding blocks:  3160 KB  6 blks
 Free Small blocks:  0 KB
 Free Ordinary blocks:  170900 KB
 Total in use:  405115 KB 71%
 Total free:170900 KB 30%
 2014/02/17 16:15:35| Open FD UNSTARTED 7 DNS Socket IPv6
 2014/02/17 16:15:35| Open FD READ/WRITE8 DNS Socket IPv4
 2014/02/17 16:15:35| Open FD READ/WRITE9 Reading next request
 2014/02/17 16:15:35| Open FD READ/WRITE   11 Reading next request
 2014/02/17 16:15:35| Open FD READ/WRITE   13 Reading next request
 2014/02/17 16:15:35| Open FD READ/WRITE   14 clients3.google.com:443
 2014/02/17 16:15:35| Open FD READ/WRITE   15 core.mochibot.com idle 
 connection
 2014/02/17 16:15:35| Open FD 

Re: [squid-users] Re: cache.log Warnings

2014-02-17 Thread Scott Mayo
On Mon, Feb 17, 2014 at 6:35 PM, Carlos Defoe carlosde...@gmail.com wrote:
 http://wiki.squid-cache.org/KnowledgeBase/QueueCongestion

 You're probably using aufs, those messages are normal.

Yes, just changed that yesterday.



 On the restart log, I never saw the AuthUserHashPointer ones, but
 since squid exits and starts normally, I don't think that is a
 problem.


Okay, thanks.


 I was going to ask how are you checking your free memory, but I think
 you just figured out that you have plenty of memory.

 I looked quickly to your conf on the other message, and seems that
 you're using a very small cache_mem. With 8GB, you can increase that
 to at least 2 GB safely. The hot objects will be kept on RAM, thus
 reducing the disk activity and speeding your proxy.


Thanks, I was wondering about making that larger and if it would be
okay.  No one suggested it on that other post, so I figured I best
leave it where it was.  I'll increase it some tomorrow.


-- 
Scott Mayo
Mayo's Pioneer Seeds


Re: [squid-users] Re: cache.log Warnings

2014-02-17 Thread Amos Jeffries
On 18/02/2014 1:35 p.m., Carlos Defoe wrote:
 http://wiki.squid-cache.org/KnowledgeBase/QueueCongestion
 
 You're probably using aufs, those messages are normal.
 
 On the restart log, I never saw the AuthUserHashPointer ones, but
 since squid exits and starts normally, I don't think that is a
 problem.

Those lines mean the credentials were still tied to an active
transaction when they were removed from the auth cache. Normal for a
shutdown with active clients at the time.

The long list of Open FD  lines is possibly a worry. That indicates
open connections.

However that said most of them are port-443 (HTTPS tunnels) or idle
persistent connections. So not a problem really just an artifact of the
not-so-nice shutdown process in Squid.

Amos