Page fault too high.

From: Benarson Behajaina <[email protected]>
Date: Tue, 6 Jul 1999 12:41:08 +0200 (MET DST)

        Hello squid users,

I'm running Squid 2.2s3,
Linux (SuSE), kernel 2.2.0, RAM 64 MB, 3x2 GB
When I start my Squid, it works fine, but after 3 hours
the Page faults with physical i/o is growing
to 10 times greater than the Number of HTTP requests.

From my cache manager info:
- Number of HTTP requests received: 4127
- Page faults with physical i/o : 44056

I read the FAQ, and tried:
- to reduce the cache_mem parameter (from 21 MB to 16 MB)
- to reduce cache_swap (low 80, high 85)
- to set memory_pools off
- to reduce the maximum_object_size (1024 KB)
- my squid was compiled with --enable-dlmalloc

but squid is still running too slow and with high page fault.

from my squid.conf:
cache_dir /cache1 1900 16 256
cache_dir /cache2 1900 16 256
cache_dir /cache3 1900 16 256

The cache manager show the following lines after 3 days:
- Number of HTTP requests received: 440127
- Page faults with physical i/o : 2030896

Should I upgrade my RAM to higher than 64 MB , or should
I reduce the disk amount size in cache_dir ?

Any help would be appreciated.

---
Benarson Behajaina.
Received on Tue Jul 06 1999 - 04:48:13 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:47:18 MST