I've just try your sugestion by dividing cache into just one partition
for each disk. But it machine hangs when got a huge load
Matus UHLAR - fantomas wrote:
On 06.05.07 14:35, Yusuf Nurrachman wrote:
I have problem with my Squid Server. When cache_dir got fullfiled, squid
dies and restarted every 30 minutes. For your information my machine
config is Core2Duo Proccessor (6600) with 4GB of RAM, 1 IDE disk for
system, 2 x 250 GB SATA disk for cache. I'm using FreeBSD 6.2.
Here's my cache_dir setting :
cache_dir diskd /c1 9000 16 64 Q1=72 Q2=64
cache_dir diskd /c2 9000 16 64 Q1=72 Q2=64
cache_dir diskd /c3 9000 16 64 Q1=72 Q2=64
cache_dir diskd /c4 9000 16 64 Q1=72 Q2=64
cache_dir diskd /c5 9000 16 64 Q1=72 Q2=64
cache_dir diskd /c6 9000 16 64 Q1=72 Q2=64
cache_dir diskd /c7 9000 16 64 Q1=72 Q2=64
cache_dir diskd /c8 9000 16 64 Q1=72 Q2=64
Each of cache divided into separate partition sized 11GB.
why do you have so many partitions on each disk? That's big useless
overhead. One partition per disk should be enough
(unless you use different filesystems for different object sizes)
Turning off memory_pools not helping too. I have recompile kernel with
options "MAXDSIZ=(512*1024*1024)"
options "DFLDSIZ=(128*1024*1024)"
not enough I'd say. for 4GB machine, you can increase these parameters.
but machine hangs after some time.