Re: All entries in store.log are RELEASE'd immediatley

Greg Cunningham writes:
> > -----Original Message-----
> > From: Reuben Farrelly [SMTP:reuben-squid@mira.net]
> > Sent: Sunday, June 27, 1999 7:01 PM
> > To: Greg Cunningham; 'squid-users@ircache.net'
> > Subject: RE: All entries in store.log are RELEASE'd immediatley
...
> > What does the tag for cache_dir read like? [Presumably you do
> > actually
> > have some free space on the drive]
> [Greg Cunningham] cache is on 3 x 9GB drives viz:
> cache_dir /usr/local/squid/cache/sdb1 8096 64 256
> cache_dir /usr/local/squid/cache/sdc1 7796 64 256
> cache_dir /usr/local/squid/cache/sdd1 8096 64 256

These are probably excessively obvious questions, but did you make sure
that the drives are actually mounted at those points, are mounted
writeable, and that squid has user or group write permissions from the
mount point? Have you fscked the disks and made sure none of them has
a hardware problem? Does running squid -z to rebuild the cache
directories say anything unusual?

> here is a down -> up sequence
> 1999/06/26 21:50:19| Preparing for shutdown after 0 requests
> 1999/06/26 21:50:19| Waiting 30 seconds for active connections
> to finish
> 1999/06/26 21:50:19| FD 17 Closing HTTP connection
> 1999/06/26 21:50:19| Shutting down...
> 1999/06/26 21:50:19| FD 18 Closing ICP connection
> 1999/06/26 21:50:19| Closing unlinkd pipe on FD 14
> 1999/06/26 21:50:19| storeDirWriteCleanLogs: Starting...
> 1999/06/26 21:50:19| Finished. Wrote 0 entries.
> 1999/06/26 21:50:19| Took 0 seconds ( 0.0 entries/sec).
> CPU Usage: 5679.330 seconds
> Maximum Resident Size: 0 KB
> Page faults with physical i/o: 229

I think page faults should be 0. (Unless I greatly misunderstand
Linux.) The CPU usage also seems bizarrely high to me.

...
> 1999/06/26 21:54:23| Starting Squid Cache version 2.2.STABLE2
> for i586-pc-linux-
> gnu...
> 1999/06/26 21:54:23| Process ID 5130
> 1999/06/26 21:54:23| With 256 file descriptors available
> 1999/06/26 21:54:23| Performing DNS Tests...
> 1999/06/26 21:54:23| Successful DNS name lookup tests...
> 1999/06/26 21:54:23| helperOpenServers: Starting 5 'dnsserver'
> processes
> 1999/06/26 21:54:23| Unlinkd pipe opened on FD 14
> 1999/06/26 21:54:23| Swap maxSize 24563712 KB, estimated 1889516
> objects
> 1999/06/26 21:54:23| Target number of buckets: 37790
> 1999/06/26 21:54:23| Using 65536 Store buckets, replacement runs
> every 1 second
> 1999/06/26 21:54:23| Max Mem size: 65536 KB
> 1999/06/26 21:54:23| Max Swap size: 24563712 KB
> 1999/06/26 21:54:23| Rebuilding storage in Cache Dir #0 (CLEAN)
> 1999/06/26 21:54:23| Rebuilding storage in Cache Dir #1 (CLEAN)
> 1999/06/26 21:54:23| Rebuilding storage in Cache Dir #2 (CLEAN)
> 1999/06/26 21:54:23| Set Current Directory to
> /usr/local/squid/cache/sdb1
> 1999/06/26 21:54:23| Loaded Icons.
> 1999/06/26 21:54:23| Accepting HTTP connections on port 8080, FD
> 17.
> 1999/06/26 21:54:23| Accepting ICP messages on port 3130, FD 18.
> 1999/06/26 21:54:23| Ready to serve requests.
> 1999/06/26 21:54:23| storeRebuildFromDirectory: DIR #2 done!
> 1999/06/26 21:56:34| storeRebuildFromDirectory: DIR #1 done!
> 1999/06/26 21:58:49| Configuring Parent
> cache.vision.net.au/8080/3130
> 1999/06/26 21:58:49| Configuring Parent
> cache2.vision.net.au/8080/3130
> 1999/06/26 21:58:49| Configuring Sibling beowulf/8080/3130
> 1999/06/26 21:58:49| storeRebuildFromDirectory: DIR #0 done!
> 1999/06/26 22:01:07| Finished rebuilding storage disk.
> 1999/06/26 22:01:07| 0 Entries read from previous
> logfile.

Here's where it really ought to be showing something. My cache.log
shows the number of entries read from previous logfile as increasing
steadily from the first boot of the server. If yours are showing 0,
that has got to be related.

...
> 1999/06/26 22:01:07| 0 Swapfile clashes avoided.
> 1999/06/26 22:01:07| Took 404 seconds ( 0.0 objects/sec).

This also seems strange. I've got 1.6GB allocated to my test cache,
with 29K objects in them currently; it takes 3 seconds to rebuild.
Yours is roughly 15 times the size, but is taking roughly 130 times as
long, in real-time, to do nothing. Unfortunately, I can't say anything
more constructive than "something is seriously wrong here." Maybe there
is some lower level problem?

  -- Clifton

-- 
 Clifton Royston  --  LavaNet Systems Architect --  cliftonr@lava.net
        "An absolute monarch would be absolutely wise and good.  
           But no man is strong enough to have no interest.  
             Therefore the best king would be Pure Chance.  
              It is Pure Chance that rules the Universe; 
          therefore, and only therefore, life is good." - AC