Memory problems

Hi Squid Users,
                I have a couple of question about squids memory usage.

I am running squid 1.1beta20 on an SGI (Irix 5.3), and used gcc 2.7.2 to
compile squid. The system has 512Mb memory and I've allowed for a cache
size of 10Gb. Cache_memory is set to 32Mb.

Soon after starting squid from cold, the process occupies about 120Mb of
memory. After 24 hours, the process size has grown by 120Mb to 240Mb and
it continues to grow at the same rate until it is using most of the
memory on the system, at which point swapping and paging start with the
resultant degradation. At this point (after nearly 4 days), it is
necessary to do a shutdown/restart to free up the memory and start all
over again.

The 'Cache Information' page from cachemgr shows that it is the memory
allocated to 'Free ordinary blocks' that is growing most rapidly. As an
example, when squid is using 386Mb memory, there is 257Mb for 'Free
ordinary blocks', and the total free is 263Mb.

As of version 1.1beta20, I have turned off the 'memory_pools' option in
order to get squid to release memory that it is not using, but this has
not made any difference.

My questions are:
1) Why is squid not re-using all the free memory that it has allocated
to itself and continue to grab more? Is there a memory
'leak' in that squid is not keeping track of freed memory blocks ?
2) Why is it not releasing unneeded memory back to the OS when the
memory_pools option is set to 'off'

Thanks for any helpful suggestions received.

David

PS Sorry about the misdirect 'subscribe' email
Received on Wed Nov 27 1996 - 18:14:56 MST