Re: Does squid eat much of CPU?

From: Payungsak Wanthaneeworakul <payungsw@dont-contact.us>
Date: Wed, 21 Apr 1999 17:46:14 +0700

> > Now I use squid version 2.1 PATCH2 on Linux 2.0.36. Since this morning,
> > I have suffered from the overloaded CPU. Its utilization has been over
 
> Squid does have quite a high standing CPU usage.

I have used squid in production stage since April 3, 1999. The CPU
utilization was in low percentage. It was approximately 10 %. Since
yesterday morning, I have confronted with the overloaded CPU.

> > 90 %. This makes the internet surfing through squid very slow in
> > comparison with direct connection.
>
> This does not necessarily follow. Your problem might be that
> requests are being serialised down a persistent connection, so that
> you get one image at a time, rather than a bit of each all at once.
>

Is there the limitation of the squid process about maximum concurrent
usage?

> > How can I solve this problem?
>
> First establish what is the real problem (at least tell us the cache
> size, memory size and processor type and speed, so we can tell
> whether you are overloading the system - check the page fault rate as
> well).
>
> If the standing CPU usage is the problem, I think you can reduce the
> rate at which squid does its housekeeping operations.

Cache configuration:
cache_dir /export/cache 1500 16 256

Memory of my machine is 128 MByte.

My CPU speed is Pentium II 350 MHz.

When I run command vmstat, I get the messages below.
$ vmstat 1 10
procs memory swap io system cpu
r b w swpd free buff cache si so bi bo in cs us sy id
1 0 0 0 2180 27904 18428 0 0 4 11 182 49 15 28 56
1 0 0 0 2228 27876 18412 0 0 20 0 249 46 35 62 4
3 0 0 0 2252 27872 18408 0 0 9 0 267 34 27 70 3
1 0 0 0 2244 27872 18416 0 0 10 79 274 39 29 69 2
1 0 0 0 2248 27868 18408 0 0 1 0 274 35 38 61 1
1 0 0 0 2248 27868 18404 0 0 4 0 269 67 38 62 1
1 0 0 0 2252 27868 18408 0 0 2 0 253 32 40 60 0
1 0 0 0 2248 27868 18412 0 0 1 87 319 42 41 57 2
1 0 0 0 2252 27868 18412 0 0 0 0 232 28 35 65 0
1 0 0 0 2256 27868 18412 0 0 1 0 226 18 38 61 2

In normal case, I get the messages below.
$ vmstat 1 10
procs memory swap io system cpu
r b w swpd free buff cache si so bi bo in cs us sy id
1 0 0 0 2176 28156 18272 0 0 4 11 182 48 16 28 56
1 0 0 0 2240 28116 18252 0 0 5 0 266 190 1 5 94
1 0 0 0 2240 28116 18252 0 0 0 0 246 181 1 4 95
0 0 0 0 2240 28116 18252 0 0 0 0 218 158 1 5 94
0 0 0 0 2244 28104 18264 0 0 14 56 269 175 2 7 91
0 0 0 0 2248 28104 18264 0 0 2 0 253 190 3 3 94
0 0 0 0 2256 28104 18264 0 0 0 0 208 138 0 4 96
0 0 0 0 2264 28104 18264 0 0 0 0 241 182 7 3 90
0 0 0 0 2248 28132 18260 0 0 0 0 231 166 1 4 95
0 0 0 0 2248 28132 18260 0 0 0 79 255 151 1 5 94

Since yesterday, I have stuck with slow squid in 90 % of internet
access. The 90 % of vmstat message presents the overloading CPU.

Thanks for your kindness
Payungsak Wanthaneeworakul
Received on Wed Apr 21 1999 - 05:04:57 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:45:54 MST