Re: [K12OSN] squid box can't handle load
Eric Harrison (eharrison@mail.mesd.k12.or.us)
Mon, 3 Feb 2003 08:08:30 -0800 (PST)
On Mon, 3 Feb 2003, Mike Rambo wrote:
>Heath Henderson wrote:
>>
>> Squid should be able to handle that kind of load easily though.
>>
>> How many computers are making connections to this machine?
>>
>
>We have 43 buildings - possibly a potential for 4000-6000 users at any
>given time (as a rough guess).
>
>I ran a top after I first posted and found...
>
>>
>> 9:09am up 2 days, 9:55, 3 users, load average: 2.44, 2.36,
>> 2.29 57 processes: 54
>> sleeping, 3 running, 0 zombie, 0 stopped
>> CPU states: 6.9% user, 93.0% system, 0.0% nice, 0.0% idle
>> Mem: 255908K av, 252448K used, 3460K free, 0K shrd, 30588K
>> buff
>> Swap: 526296K av, 0K used, 526296K free 153784K
>> cached
>>
>> PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
>> 8146 squid 25 0 12668 12M 1252 R 71.4 4.9 3:50 squid
>> 8152 squid 15 0 2112 2112 688 S 2.7 0.8 0:06 squidGuard
>> 8154 squid 15 0 2076 2076 688 S 1.7 0.8 0:06 squidGuard
>> 8165 squid 16 0 2064 2064 688 S 1.7 0.8 0:04 squidGuard
>> 8160 squid 15 0 2088 2088 688 S 1.5 0.8 0:05 squidGuard
>> 8183 root 16 0 1044 1044 848 R 1.5 0.4 0:00 top
>> 8149 squid 15 0 2104 2104 688 S 1.3 0.8 0:06 squidGuard
>> 8150 squid 15 0 2092 2092 688 S 1.3 0.8 0:05 squidGuard
>> 8151 squid 15 0 2104 2104 688 S 1.3 0.8 0:07 squidGuard
>> 8153 squid 15 0 2100 2100 688 S 1.3 0.8 0:05 squidGuard
>> 8159 squid 16 0 2104 2104 688 S 1.3 0.8 0:05 squidGuard
>> 8161 squid 15 0 2104 2104 688 S 1.3 0.8 0:05 squidGuard
>> 8148 squid 15 0 2092 2092 688 S 1.1 0.8 0:07 squidGuard
>> 8158 squid 15 0 2084 2084 688 S 1.1 0.8 0:05 squidGuard
>> 8166 squid 15 0 2080 2080 688 S 1.1 0.8 0:03 squidGuard
>> 3 root 34 19 0 0 0 RWN 0.9 0.0 2:17
>> ksoftirqd_CPU0
>> 8147 squid 15 0 2100 2100 688 S 0.9 0.8 0:09 squidGuard
>> 8156 squid 15 0 2088 2088 688 S 0.9 0.8 0:05 squidGuard
>> 8157 squid 15 0 2104 2104 688 S 0.9 0.8 0:05 squidGuard
>> 8155 squid 15 0 2080 2080 688 S 0.7 0.8 0:05 squidGuard
>> 8163 squid 15 0 2096 2096 688 S 0.7 0.8 0:04 squidGuard
>> 8164 squid 15 0 2092 2092 688 S 0.7 0.8 0:04 squidGuard
>> 144 root 15 0 0 0 0 SW 0.5 0.0 0:05 kjournald
>> 8162 squid 15 0 2076 2076 688 S 0.5 0.8 0:05 squidGuard
>> 1 root 15 0 480 480 420 S 0.0 0.1 0:08 init
>> 2 root 15 0 0 0 0 SW 0.0 0.0 0:00 keventd
>> 4 root 15 0 0 0 0 SW 0.0 0.0 0:09 kswapd
>>
>
>
>It would appear the box isn't fast enough for the load. I did a service
>ipchains stop to remove the port 80 redirection to squid and internet
>access has returned to normal. Of course filtering is off line now too.
>
>
>Mike
We're running ~100 sites on five boxes, each with a PIII 550, 512M ram,
20 Gig IDE drives. We have run fine on as few as three of these servers,
we'd probably be ok with two.
Caches are memory sensitive, you will want more than 256M.
For squid configuration, here's my entries that are relevent to performance:
cache_mem 300 MB
maximum_object_size 16384 KB
cache_replacement_policy heap LFUDA
memory_replacement_policy heap LFUDA
cache_dir diskd /var/spool/squid 15000 30 256
cache_store_log none
half_closed_clients on
-Eric
_______________________________________________
K12OSN mailing list
K12OSN@redhat.com
https://listman.redhat.com/mailman/listinfo/k12osn
For more info see