Claude.Lecommandeur@Epfl.Ch writes:
>
> Hi,
>
>
> I'm using squid-1.0.5 and am having a problem for the second time.
>The problem is that suddenly the cache spool becomes almost full.
>
> The parameters in squid.conf are :
>
>cache_swap 1000
>cache_swap_low 75
>cache_swap_high 80
>
> cacgemgr.cgi indicates that 796 MB are used :
>
>Cache information for squid:
> Storage Swap size: 796 MB
> Storage Mem size: 816 KB
>
> But there is much more disk usage :
>
>Filesystem kbytes used avail %used Mounted on
>/dev/vg00/lvol4 1156885 1062139 -20943 102% /harvest
>
>
> The cache is running for several month and everything was OK
>until this morning.
>
> Is there any know reason for this ? I suspect that squid has lost
>track of files in its spool. Is there a script to reset the spool to
>what squid thinks it is ? or should I discard all the spool ? (it was
>what I did the first time it happened)
This is a problem with early versions, expired objects are orphaned
during the reload process (i.e. its too slow to unlink() them all).
More recent versions (1.0.8+) have a function to periodically remove
orphaned swap files. You might also use the scripts/check_cache.pl
script from the source distribution.
Duane W.
Received on Wed Oct 09 1996 - 10:31:12 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:33:16 MST