Well, your stats don't look too bad. Is this representative of what you normally see? It looks like your misses are taking about as long as DNS requests (and misses are only taking a fifth of a second). On the accesses that are timing out, have you verified that the sites are indeed accessible? Spyware often tries to access pages that don't exist any more.
Another useful cachemrg page is "Cache Utilization" (about 2/3 of the way down the page). It shows requests per second and bandwidth usage. Chris -----Original Message----- From: Shawn Wright [mailto:[EMAIL PROTECTED] Sent: Tuesday, November 23, 2004 11:08 AM To: [EMAIL PROTECTED] Subject: [squid-users] Extreme delays on cache misses This may not be a squid issue, but I'm hoping to find some assistance in deciphering cachemgr output to determine if squid is part of the problem. Here is the situation: Approx 500 client PCs 10Mb full duplex internet feed, through a Linux firewall, with peak traffic of 5Mb/s, avg of . Temporary proxy was a squid box running win2k on a PII/266. I realize this is pretty slim hardware for the above load, but it was performing surprisingly well until a few weeks ago. We were running low on RAM in our firewall, so it was taken down and upgraded. We are fairly certain it is no longer a bottleneck. We have replaced the squid box with a 2.4G box, and saw little performance improvement. At times, responses were good, at others, so slow that we would see timeouts. Access log showed many cache misses with elapsed times over 20 seconds. In recent weeks, we've had some client PCs infected with spyware hammering our proxy with up to 80 hits per second (each client), all being denied, and generating hundreds of Mb of extra logs each day. (we require all access to be authenticated). At present, all of these machines have been located and shutdown. Below is the cachemgr out for general info - I'd appreciate any tips on what to focus on. THanks. Start Time: Tue, 23 Nov 2004 17:21:00 GMT Current Time: Tue, 23 Nov 2004 20:01:05 GMT Connection information for squid: Number of clients accessing cache: 244 Number of HTTP requests received: 150248 Number of ICP messages received: 0 Number of ICP messages sent: 0 Number of queued ICP replies: 0 Request failure ratio: 0.00 Average HTTP requests per minute since start: 938.6 Average ICP messages per minute since start: 0.0 Select loop called: 4937900 times, 1.945 ms avg Cache information for squid: Request Hit Ratios: 5min: 26.5%, 60min: 28.9% Byte Hit Ratios: 5min: 15.1%, 60min: 12.1% Request Memory Hit Ratios: 5min: 11.5%, 60min: 8.4% Request Disk Hit Ratios: 5min: 47.4%, 60min: 44.0% Storage Swap size: 1843193 KB Storage Mem size: 68536 KB Mean Object Size: 15.19 KB Requests given to unlinkd: 13646 Median Service Times (seconds) 5 min 60 min: HTTP Requests (All): 0.08729 0.10281 Cache Misses: 0.15048 0.19742 Cache Hits: 0.01469 0.00000 Near Hits: 0.12106 0.15888 Not-Modified Replies: 0.00000 0.00000 DNS Lookups: 0.16304 0.09117 ICP Queries: 0.00000 0.00000 Resource usage for squid: UP Time: 9604.937 seconds CPU Time: 2047.047 seconds CPU Usage: 21.31% CPU Usage, 5 minute avg: 2.34% CPU Usage, 60 minute avg: 27.24% Maximum Resident Size: 130388 KB Page faults with physical i/o: 33227 Memory accounted for: Total accounted: 92599 KB memPoolAlloc calls: 17735541 memPoolFree calls: 17014828 File descriptor usage for squid: Maximum number of file descriptors: 2048 Largest file desc currently in use: 325 Number of file desc currently in use: 187 Files queued for open: 0 Available number of file descriptors: 1861 Reserved number of file descriptors: 100 Store Disk files open: 4 Internal Data Structures: 121688 StoreEntries 15119 StoreEntries with MemObjects 15115 Hot Object Cache Items 121341 on-disk objects Generated Tue, 23 Nov 2004 20:01:05 GMT, by cachemgr.cgi/[EMAIL PROTECTED] -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Shawn Wright, I.T. Manager Shawnigan Lake School http://www.sls.bc.ca [EMAIL PROTECTED]
