any recommendations for tuning cache_mem parameters?

From: Joe Ramey <ramey@dont-contact.us>
Date: Wed, 25 Sep 1996 14:37:15 -0500

I'm running squid-1.0.16. Sometimes I get messages like this:

[24/Sep/1996:13:41:38 -0500] store.c:1979: storeGetMemSpace: Allocating beyond the high water mark with total size of 15102414
[24/Sep/1996:13:41:38 -0500] store.c:1979: storeGetMemSpace: Allocating beyond the high water mark with total size of 15106382

And occasionally like this:

[24/Sep/1996:13:44:11 -0500] store.c:1989: storeGetMemSpace: Allocating beyond the MAX Store with total size of 16777912
[24/Sep/1996:13:44:11 -0500] store.c:1991: Start Deleting Behind for every pending objects
[24/Sep/1996:13:44:11 -0500] store.c:1992: You should really adjust your cache_mem, high/low water mark,
[24/Sep/1996:13:44:11 -0500] store.c:1993: max object size to suit your need.

I recall some discussion earlier about adjusting cache_swap_low and
cache_swap_high to be close to each other. What about cache_mem_low
and cache_mem_high? Would adjusting them to be close to each other
help avoid this error message? Or, do I need to reduce them both?

I tried setting cache_mem_high to 85 and leaving cache_mem_low at 75,
and I am still getting the errors, but (it seems) not quite as often.

Any suggestions? Should I increase cache_mem from the default 16 to
32? I think I have enough memory to handle that. Would it help
prevent this problem?
Received on Wed Sep 25 1996 - 12:38:40 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:33:04 MST