Thanks, here's the additional info. Could there be a bug?
I've changed the high/low level, but it seems that only some cache_dir lowers their storage. Those who hit 99-100% stills keeps hovering between that level until the warning comes (lowered automatically by squid to prevent crashing).
Filesystem 1k-blocks Used Available Use% Mounted on
/dev/sda1 505605 106853 372648 23% /
none 1293776 0 1293776 0% /dev/shm
/dev/sda6 505605 8239 471262 2% /tmp
/dev/sda5 4127076 788332 3129100 21% /usr
/dev/sda8 23640176 1279368 21159952 6% /usr/local
/dev/sda3 5162828 472364 4428204 10% /usr/local/squid/logs
/dev/sda7 505605 67777 411724 15% /var
/dev/sdb1 17211212 12727952 3594868 78% /usr/local/squid/cache2
/dev/sdb2 17218732 16049828 280108 99% /usr/local/squid/cache3
/dev/sdc1 17359192 12754736 3708436 78% /usr/local/squid/cache4
/dev/sdc2 17359220 12701628 3761568 78% /usr/local/squid/cache5
/dev/sdc3 17359220 15735748 727448 96% /usr/local/squid/cache6
/dev/sdc4 17359220 15883264 579932 97% /usr/local/squid/cache7
/dev/sdd1 17359192 15863172 600000 97% /usr/local/squid/cache8
/dev/sdd2 17359220 15476828 986368 95% /usr/local/squid/cache9
/dev/sdd3 17359220 15574196 889000 95% /usr/local/squid/cache10
/dev/sdd4 17359220 15551112 912084 95% /usr/local/squid/cache11
/dev/sde1 17359192 16106176 356996 98% /usr/local/squid/cache12
/dev/sde2 17359220 15987532 475664 98% /usr/local/squid/cache13
/dev/sde3 17359220 15422344 1040852 94% /usr/local/squid/cache14
/dev/sde4 17359220 16254176 209020 99% /usr/local/squid/cache15
/dev/sdf1 17359192 15659120 804052 96% /usr/local/squid/cache16
/dev/sdf2 17359220 15555732 907464 95% /usr/local/squid/cache17
/dev/sdf3 17359220 14615468 1847728 89% /usr/local/squid/cache18
/dev/sdf4 17359220 16296124 167072 99% /usr/local/squid/cache19
====================
cache_dir ufs /usr/local/squid/cache2 15360 32 256 cache_dir ufs /usr/local/squid/cache3 15360 32 256 cache_dir ufs /usr/local/squid/cache4 15360 32 256 cache_dir ufs /usr/local/squid/cache5 15360 32 256 cache_dir ufs /usr/local/squid/cache6 15360 32 256 cache_dir ufs /usr/local/squid/cache7 15360 32 256 cache_dir ufs /usr/local/squid/cache8 15360 32 256 cache_dir ufs /usr/local/squid/cache9 15360 32 256 cache_dir ufs /usr/local/squid/cache10 15360 32 256 cache_dir ufs /usr/local/squid/cache11 15360 32 256 cache_dir ufs /usr/local/squid/cache12 15360 32 256 cache_dir ufs /usr/local/squid/cache13 15360 32 256 cache_dir ufs /usr/local/squid/cache14 15360 32 256 cache_dir ufs /usr/local/squid/cache15 15360 32 256 cache_dir ufs /usr/local/squid/cache16 15360 32 256 cache_dir ufs /usr/local/squid/cache17 15360 32 256 cache_dir ufs /usr/local/squid/cache18 15360 32 256 cache_dir ufs /usr/local/squid/cache19 15360 32 256
=================================
cache_swap_low 80 cache_swap_high 85
==================================
From: Hendrik Voigtl�nder <[EMAIL PROTECTED]> To: Lizzy Dizzy <[EMAIL PROTECTED]> CC: [EMAIL PROTECTED] Subject: Re: [squid-users] ./squid -z questions Date: Mon, 28 Jun 2004 08:21:32 +0200
Lizzy Dizzy wrote:Hi, I have a caching server with 6 SCSI disks. The first disk contains the squid application and kernel while the other 5 are just cache storage.I don't think that you need to recreate the cache dirs.
Some times back, the first disk crashed and I recreated the OS and squid application on another similar machine and put the disk back. I ensure that partitioning in the cache_dir (inside squid.conf) is the same as what is partitioned inside the 5 cache disk.
Lately I notice the following:
2004/06/28 10:05:40| diskHandleWrite: FD 1028: disk write error: (28) No space left on device
2004/06/28 10:05:40| storeUfsWriteDone: got failure (-6)
2004/06/28 10:05:40| storeSwapOutFileClosed: dirno 12, swapfile 0004689A, errflag=-6
(28) No space left on device
/dev/sde3 17359220 15373680 1089516 94% /usr/local/squid/cache14
/dev/sde4 17359220 16283368 179828 100% /usr/local/squid/cache15
/dev/sdf1 17359192 15388208 1074964 94% /usr/local/squid/cache16
I've set the low and high mark from 90 and 93 to
cache_swap_low 80 cache_swap_high 85
but they do'nt seems to help. I am wondering if I need to force squid recreate the cache directories (by using squid -z). But would I loose the cache data? Do I need to remove the existing directories first? Can I use squid -z without creating directories in the harddisks frst (i.e. let squid create them based on the cache_dir lists)
IMHO you would loose the data as you need to remove at least the dir you want to recreate. You can recreate a single cache_dir, I use a modified squid.conf (with only one cache_dir) for that.
Erasing a cache_dir take ages. If you need to keep the squid up, deactivate the cache_dir in squid.conf. Creating a new filesystem is definitely faster.
There might be another problem. Please post your cache_dir settings.
Regards, Hendrik Voigtl�nder
_________________________________________________________________
Keep track of Singapore & Malaysia stock prices. http://www.msn.com.sg/money/
