If you are on a Solaris2.6 box this is a known disk fragmentation problem in Sol2.6
during heavy file creation/deletion as with Squid.
The problem has been discussed before on this list. I don't think there was a solution
except lowering arg 2 of cache_dir to say 85% of actual disk capacity..
P�l
----------------------------------------------------------------
Addr: P�l Baltzersen, ElTele �st AS, Fredrik Selmers vei 2, Postboks 6299 Etterstad,
N-0603 Oslo.
Phone: +47 23 18 10 00 Direct: +47 23 18 11 74
fax: +47 23 18 10 01 Mobile: +47 93 08 11 74
mail: [EMAIL PROTECTED] [EMAIL PROTECTED]
>>> List Server Account <[EMAIL PROTECTED]> 12:06:10 15.03.99 >>>
I am running squid 2.1 patch 2 + mem leak patch. (Although the same
happens with 2.2 pre 3). I have looked at the FAQ, and can't find any
information about this.
I also adjusted "store_avg_object_size from 13 KB down to 11 KB, as per
an article I read (but did not understand), but it seems to have made no
difference. I have 4 cache partitions, all 2050 Megs. Once one shrinks,
the
others follow. If I restart the cache, the caches will start off back at
a max of 2050 megs, and decrease again.
My Squid cache automatically shrinks. Initially 1 directory will
shrink, then all of them seem to follow. I recently turned debugging on
and it tells me :
At 21:28:03
diskHandleWrite: FD56: disk write error: (28) No space left on device
storeSwapOutHandle: SwapOut failure (err code = -6).
WARNING: Shrinking cache_dir #2 to 1490620 KB
I just did discovered the problem, and did a df (at 23:08:07)
/dev/sda2 2136576 1578058 558518 /cache2 (which is cache dir #2)
I did reformat the partition (and there is only the squid cache on the
partition), to allow 0% for root only space.
My Config Includes :
cache_swap_low 97
cache_swap_high 98
maximum_object_size 12000 KB
cache_dir /cache2 2050 16 256
Any ideas, this is driving me nuts !!!
Thanks,
David
