Situation: squid 1.1.16, 1 gig cache, 64M ram, Linux 2.0.31, stock
libc 5.3.12
Cache information:
Squid Object Cache: Version 1.1.16
Start Time: Wed, 01 Oct 1997 03:31:41 GMT
Current Time: Wed, 29 Oct 1997 01:35:54 GMT
Connection information for squid:
Number of TCP connections: 313149
Number of UDP connections: 0
Connections per hour: 467.3
Select loop called: 7681584 times, 314.031 ms avg
Cache information for squid:
Storage Swap size: 808 MB
Storage Mem size: 4906 KB
Storage LRU Expiration Age: 365.00 days
Requests given to unlinkd: 0
Unused fileno stack count: 0
Resource usage for squid:
CPU Time: 3605 seconds (1960 user 1645 sys)
CPU Usage: 0%
Maximum Resident Size: 0 KB
Page faults with physical i/o: 174809
Memory usage for squid via mallinfo():
Total space in arena: 39240 KB
Ordinary blocks: 15366 KB 16011 blks
Small blocks: 0 KB 0 blks
Holding blocks: 260 KB 1 blks
Free Small blocks: 0 KB
Free Ordinary blocks: 23873 KB
Total in use: 15626 KB 40%
Total free: 23873 KB 61%
The system is now starting to allocate swap space, performance is
nevertheless quite good, but I'm now wondering why 23M of memory are
reported free, but still locked by squid. I suspect a libc/malloc
problem, and I'd certainly love to see at least 20 of those megs
actually returned to the OS. Is there anything I can do? From the
page at http://cache.is.co.za/squid/opt/malloc.html I've found a list
of alternative malloc implementations, and I was wondering if any of
the list subscribers had any experience in this regard. As a last
note.. "Maximum Resident Size", it's reported as set to 0. What is it
exactly, and how can I change it? Doesn't seem to be "cache_mem" in
squid.conf, which is set at 8 megs and according to the notes before
it should be concerned only with in memory objects.
thanks
-- danilo
Received on Tue Oct 28 1997 - 17:51:28 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:37:21 MST