Re: [squid-users] dead squid, Still looking,

While we're out there talking about cache size limits, I need a
refresher....

What is the general rule in terms of setting a cache size limit?
I, obviously, want the cachable space to be as large as possible, but seem
to experience a huge decrease in performance and increase in crashes as I
increase the cache size?

Is there some general rule that compares physical ram and cpu which would
help me set up the optimal cache size limitation?

I have 1 gig of ram 2 gig swap space and a 2.4 gb xeon processor.

Tim Rainier
Information Services, Kalsec, INC
trainier@kalsec.com

"John R. Van Lanen, Network Operations - TCCSA" <VANLANEN@tccsa.net>
08/29/2005 04:46 PM

To
trainier@kalsec.com
cc
squid-users@squid-cache.org, VANLANEN@tccsa.net
Subject
Re: [squid-users] dead squid, Still looking,

Start debug level 1 in squid.conf, The problem about strace, is squid will
start another process, that new process it the one that it can't find and
then
dies. This happens in milli seconds. No time to enter the command or know
which process will it be. Squid from what I see always starts two
processes, a
main and another. The other is the one it can't find therefore making
squid
die.

Example
squid dead but pid file exists
squid: ERROR Could not send signal 0 to process 5530: (3) No such
process,

This process was generated seconds before it attempts to send signal 0 to
the
process.

Could not possiblely do a trace before it happens.

Sorry if this is confusing but it's confusing me. I know other people run
squid
with more users than I do on more limited machines.

> The squid cache has reached it size limit and is being rebuilt to
> compensate...

> You could load squid in debug mode. There are a few debug mode options
> available for you...
> Just use squid -? to get a list of options.

> Alternatively, you can attach an strace session to squid to see exactly
> what it's doing.
> strace -p <pid of squid> > /var/log/squid.trace.out 2>&1

> Tim Rainier
> Information Services, Kalsec, INC

> "John R. Van Lanen, Network Operations - TCCSA" <VANLANEN@tccsa.net>
> 08/29/2005 03:29 PM

> To
> trainier@kalsec.com
> cc
> squid-users@squid-cache.org, VANLANEN@tccsa.net
> Subject
> Re: [squid-users] dead squid, Still looking,

> If I do this, its a matter of hours and the same occurs again. What
> triggers
> the process that squid feels it needs to start again and add a new
> process, why
> did the old process die to begin with? In other words, what triggers
the
> need
> to do this over and over again. When doing a squid status I only ever
see
> 2
> processes.

> Here is the last 1/8 of the File,

> > > 2005/08/19 17:28:21| Starting Squid Cache version 2.5.STABLE3 for
> > > i386-redhat-linux-gnu...
> > > 2005/08/19 17:28:21| Process ID 5486
> > > 2005/08/19 17:28:21| With 1024 file descriptors available
> > > 2005/08/19 17:28:21| DNS Socket created at 0.0.0.0, port 32785, FD 4
> > > 2005/08/19 17:28:21| Adding nameserver 10.1.1.101 from
> /etc/resolv.conf
> > > 2005/08/19 17:28:21| helperOpenServers: Starting 10 'ifp_redirector'
> > > processes
> > > 2005/08/19 17:28:21| User-Agent logging is disabled.
> > > 2005/08/19 17:28:21| Referer logging is disabled.
> > > 2005/08/19 17:28:21| Unlinkd pipe opened on FD 19
> > > 2005/08/19 17:28:21| Swap maxSize 2048000 KB, estimated 157538
objects
> > > 2005/08/19 17:28:21| Target number of buckets: 7876
> > > 2005/08/19 17:28:21| Using 8192 Store buckets
> > > 2005/08/19 17:28:21| Max Mem size: 8192 KB
> > > 2005/08/19 17:28:21| Max Swap size: 2048000 KB
> > > 2005/08/19 17:28:21| Rebuilding storage in /usr/squid (DIRTY)
> > > 2005/08/19 17:28:21| Using Least Load store dir selection
> > > 2005/08/19 17:28:21| chdir: /var/spool/squid: (2) No such file or
> > > directory
> > > 2005/08/19 17:28:21| Current Directory is /root
> > > 2005/08/19 17:28:21| Loaded Icons.
> > > 2005/08/19 17:28:22| Accepting HTTP connections at 0.0.0.0, port
7034,
> > > FD 21.
> > > 2005/08/19 17:28:22| Accepting ICP messages at 0.0.0.0, port 3130,
FD
> 22.
> > > 2005/08/19 17:28:22| WCCP Disabled.
> > > 2005/08/19 17:28:22| Ready to serve requests.
> > > 2005/08/19 17:28:22| Store rebuilding is 3.5% complete
> > > 2005/08/19 17:28:25| Starting Squid Cache version 2.5.STABLE3 for
> > > i386-redhat-linux-gnu...
> > > 2005/08/19 17:28:25| Process ID 5511
> > > 2005/08/19 17:28:25| With 1024 file descriptors available
> > > 2005/08/19 17:28:25| DNS Socket created at 0.0.0.0, port 32785, FD 4
> > > 2005/08/19 17:28:25| Adding nameserver 10.1.1.101 from
> /etc/resolv.conf
> > > 2005/08/19 17:28:25| helperOpenServers: Starting 10 'ifp_redirector'
> > > processes
> > > 2005/08/19 17:28:25| User-Agent logging is disabled.
> > > 2005/08/19 17:28:25| Referer logging is disabled.
> > > 2005/08/19 17:28:25| Unlinkd pipe opened on FD 19
> > > 2005/08/19 17:28:25| Swap maxSize 2048000 KB, estimated 157538
objects
> > > 2005/08/19 17:28:25| Target number of buckets: 7876
> > > 2005/08/19 17:28:25| Using 8192 Store buckets
> > > 2005/08/19 17:28:25| Max Mem size: 8192 KB
> > > 2005/08/19 17:28:25| Max Swap size: 2048000 KB
> > > 2005/08/19 17:28:25| Rebuilding storage in /usr/squid (DIRTY)
> > > 2005/08/19 17:28:25| Using Least Load store dir selection
> > > 2005/08/19 17:28:25| chdir: /var/spool/squid: (2) No such file or
> > > directory
> > > 2005/08/19 17:28:25| Current Directory is /root
> > > 2005/08/19 17:28:25| Loaded Icons.
> > > 2005/08/19 17:28:25| Accepting HTTP connections at 0.0.0.0, port
7034,
> > > FD 21.
> > > 2005/08/19 17:28:25| Accepting ICP messages at 0.0.0.0, port 3130,
FD
> 22.
> > > 2005/08/19 17:28:25| WCCP Disabled.
> > > 2005/08/19 17:28:25| Ready to serve requests.
> > > 2005/08/19 17:28:26| Store rebuilding is 3.5% complete
> > > 2005/08/19 17:28:29| Starting Squid Cache version 2.5.STABLE3 for
> > > i386-redhat-linux-gnu...
> > > 2005/08/19 17:28:29| Process ID 5530
> > > 2005/08/19 17:28:29| With 1024 file descriptors available
> > > 2005/08/19 17:28:29| DNS Socket created at 0.0.0.0, port 32785, FD 4
> > > 2005/08/19 17:28:29| Adding nameserver 10.1.1.101 from
> /etc/resolv.conf
> > > 2005/08/19 17:28:29| helperOpenServers: Starting 10 'ifp_redirector'
> > > processes
> > > 2005/08/19 17:28:29| User-Agent logging is disabled.
> > > 2005/08/19 17:28:29| Referer logging is disabled.
> > > 2005/08/19 17:28:29| Unlinkd pipe opened on FD 19
> > > 2005/08/19 17:28:29| Swap maxSize 2048000 KB, estimated 157538
objects
> > > 2005/08/19 17:28:29| Target number of buckets: 7876
> > > 2005/08/19 17:28:29| Using 8192 Store buckets
> > > 2005/08/19 17:28:29| Max Mem size: 8192 KB
> > > 2005/08/19 17:28:29| Max Swap size: 2048000 KB
> > > 2005/08/19 17:28:29| Rebuilding storage in /usr/squid (DIRTY)
> > > 2005/08/19 17:28:29| Using Least Load store dir selection
> > > 2005/08/19 17:28:29| chdir: /var/spool/squid: (2) No such file or
> > > directory
> > > 2005/08/19 17:28:29| Current Directory is /root
> > > 2005/08/19 17:28:29| Loaded Icons.
> > > 2005/08/19 17:28:30| Accepting HTTP connections at 0.0.0.0, port
7034,
> > > FD 21.
> > > 2005/08/19 17:28:30| Accepting ICP messages at 0.0.0.0, port 3130,
FD
> 22.
> > > 2005/08/19 17:28:30| WCCP Disabled.
> > > 2005/08/19 17:28:30| Ready to serve requests.
> > > 2005/08/19 17:28:30| Store rebuilding is 3.5% complete

> > Squid died sometime after this entry.

> > System Reboot here. Squid active again.

> > > 2005/08/22 09:11:23| Starting Squid Cache version 2.5.STABLE3 for
> > > i386-redhat-linux-gnu...
> > > 2005/08/22 09:11:23| Process ID 2742
> > > 2005/08/22 09:11:23| With 1024 file descriptors available
> > > 2005/08/22 09:11:23| DNS Socket created at 0.0.0.0, port 32770, FD 4
> > > 2005/08/22 09:11:23| Adding nameserver 10.1.1.101 from
> /etc/resolv.conf
> > > 2005/08/22 09:11:23| helperOpenServers: Starting 10 'ifp_redirector'
> > > processes
> > > 2005/08/22 09:11:24| User-Agent logging is disabled.
> > > 2005/08/22 09:11:24| Referer logging is disabled.
> > > 2005/08/22 09:11:24| Unlinkd pipe opened on FD 19
> > > 2005/08/22 09:11:24| Swap maxSize 2048000 KB, estimated 157538
objects
> > > 2005/08/22 09:11:24| Target number of buckets: 7876
> > > 2005/08/22 09:11:24| Using 8192 Store buckets
> > > 2005/08/22 09:11:24| Max Mem size: 8192 KB
> > > 2005/08/22 09:11:24| Max Swap size: 2048000 KB
> > > 2005/08/22 09:11:24| Rebuilding storage in /usr/squid (DIRTY)
> > > 2005/08/22 09:11:24| Using Least Load store dir selection
> > > 2005/08/22 09:11:24| Set Current Directory to /usr/squid
> > > 2005/08/22 09:11:24| Loaded Icons.
> > > 2005/08/22 09:11:25| Accepting HTTP connections at 0.0.0.0, port
7034,
> > > FD 21.
> > > 2005/08/22 09:11:25| Accepting ICP messages at 0.0.0.0, port 3130,
FD
> 22.
> > > 2005/08/22 09:11:25| WCCP Disabled.
> > > 2005/08/22 09:11:25| Ready to serve requests.
> > > 2005/08/22 09:11:25| Store rebuilding is 3.5% complete
> > > 2005/08/22 09:11:27| Done reading /usr/squid swaplog (116757
entries)
> > > 2005/08/22 09:11:27| Finished rebuilding storage from disk.
> > > 2005/08/22 09:11:27| 116108 Entries scanned
> > > 2005/08/22 09:11:27| 0 Invalid entries.
> > > 2005/08/22 09:11:27| 0 With invalid flags.
> > > 2005/08/22 09:11:27| 116108 Objects loaded.
> > > 2005/08/22 09:11:27| 0 Objects expired.
> > > 2005/08/22 09:11:27| 0 Objects cancelled.
> > > 2005/08/22 09:11:27| 649 Duplicate URLs purged.
> > > 2005/08/22 09:11:27| 0 Swapfile clashes avoided.
> > > 2005/08/22 09:11:27| Took 3.0 seconds (39024.9 objects/sec).
> > > 2005/08/22 09:11:27| Beginning Validation Procedure
> > > 2005/08/22 09:11:27| Completed Validation Procedure
> > > 2005/08/22 09:11:27| Validated 115459 Entries
> > > 2005/08/22 09:11:27| store_swap_size = 1548892k
> > > 2005/08/22 09:11:27| storeLateRelease: released 0 objects
> > > 2005/08/22 11:31:10| urlParse: Illegal character in hostname
> > > 'c12_zimmerman'
> > > 2005/08/22 11:35:00| urlParse: Illegal character in hostname
> > > 'c12_zimmerman'

>
______________________________________________________________________________

> Press return for more...

> EMAIL>

> > Yes:
> > -F Don't serve any requests until store is rebuilt.

> > My guess is that squid is trying to rebuild the store, but is too busy
> > servicing requests.
> > If that doesn't work, you can manually clear the cache, the store file
> and
> > have squid rebuild the cache hierarchy (squid -z).

> > Tim Rainier
> > Information Services, Kalsec, INC

> > "John R. Van Lanen, Network Operations - TCCSA" <VANLANEN@tccsa.net>
> > 08/29/2005 12:12 PM

> > To
> > squid-users@squid-cache.org
> > cc
> > VANLANEN@tccsa.net
> > Subject
> > [squid-users] dead squid, Still looking,

> > Me again, sorry,

> > my squid dies, when it does the cache.log most always shows the same
> last
> > entry
> > of

> > Store rebuilding is 4.4% (or some other number) complete

> > Then nothing, The Squid status shows

> > squid dead but pid file exists,

> > Does anyone know in ES 3 if there is a problem with processes just
> > disappearing?

> > I've Tryed increasing Cache_mem from 8 to 768
> > disabled store.log
> > Increase ifp_redirector processes from 10 to 30.

> > Any other settings anyone could think of?

> > Orginally posted;

> > I'm running Redhat ES 3.0,

> > Running Squid Stable 2.5 ver 3 that comes with the Redhat ES 3.0

> > The issue I have is that squid will stop dead,

> > Do a squid status and it shows

> > squid dead but pid file exists
> > squid: ERROR Could not send signal 0 to process 5530: (3) No such
> process,

> > Have seen this before, restart squid or rebooting solves the issue for
> > awhile,
> > the 5530 number changes.

> > Cache.log shows nothing but normal stuff.

> > Searching the net shows others with the same issues but can't find a
> > solution.

> > Let me know if you can help. Thanks.
> >
>
______________________________________________________________________________
> > | John R. Van Lanen, Manager of Network Operations Voice: (330)
> 264-6047
> > |
> > | Tri-County Computer Services Association (TCCSA) Fax: (330)
> 264-5703
> > |
> > | "Do not meddle in the affairs of dragons, E-mail:vanlanen@TCCSA.NET|
> > | because you are crunchy and taste good with ketchup" |
> >
>
------------------------------------------------------------------------------

>
______________________________________________________________________________
> | John R. Van Lanen, Manager of Network Operations Voice: (330)
264-6047
> |
> | Tri-County Computer Services Association (TCCSA) Fax: (330)
264-5703
> |
> | "Do not meddle in the affairs of dragons, E-mail:vanlanen@TCCSA.NET|
> | because you are crunchy and taste good with ketchup" |
>
------------------------------------------------------------------------------

______________________________________________________________________________
| John R. Van Lanen, Manager of Network Operations Voice: (330) 264-6047
 |
| Tri-County Computer Services Association (TCCSA) Fax: (330) 264-5703
 |
| "Do not meddle in the affairs of dragons, E-mail:vanlanen@TCCSA.NET|
| because you are crunchy and taste good with ketchup" |
------------------------------------------------------------------------------
Received on Mon Aug 29 2005 - 15:18:10 MDT