[Asterisk-Users] Top and asterisk performance
Julian Lyndon-Smith
asterisk at dotr.com
Fri Oct 28 16:33:21 MST 2005
Oh man, stupid of me - I forgot to include that info at the start.
However, there are a couple of things that come out of the analysis:
1) I've restarted the * server. I now have
top - 00:29:08 up 13 days, 10:50, 1 user, load average: 0.00, 0.01, 1.04
Tasks: 83 total, 1 running, 82 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.3% us, 0.0% sy, 0.0% ni, 99.7% id, 0.0% wa, 0.0% hi, 0.0% si
Mem: 1034640k total, 270408k used, 764232k free, 42524k buffers
Swap: 2031608k total, 0k used, 2031608k free, 153256k cached
*much* better :)
2) There were about 10 channels that were constantly trying to call: It
turns out that the agentcallbacklogin has had a change of parameters in
the past couple of days, one that basically meant that the original
dialplan caused the system to start eating cpu cycles. So my old dial
plan seemed to cause the issues. I did realise that there was a problem
early on, in that agents couldn't login. After a (quick!) bit of
research, I discovered that the dialplan needed changing, changed it,
did a "extensions reload" and voila! Agents could log in. I forgot about
the 10 or so "crashed" (see below) channels.
foxtrot*CLI> show channels
Channel Location State Application(Data)
Local/6041 at AgentQ-a2 6041 at AgentQ:3 Busy Busy()
Local/6041 at AgentQ-a2 s at AgentQ:1 Down (None)
Local/6036 at AgentQ-33 6036 at AgentQ:3 Busy Busy()
Local/6036 at AgentQ-33 s at AgentQ:1 Down (None)
SIP/711-6416 8888 at from-sip:1 Ring
AgentCallbackLogin(711|@AgentQ
SIP/711-8145 *8888 at from-sip:1 Ring
AgentCallbackLogin(8888|711 at Ag
SIP/711-5ff9 8888 at from-sip:1 Ring
AgentCallbackLogin(711|@AgentQ
SIP/6035-3c31 *6035 at from-sip:1 Up
AgentCallbackLogin(6035|6035 at A
SIP/6033-a08c *6033 at from-sip:1 Up
AgentCallbackLogin(6033|6033 at A
SIP/6025-c55c *6025 at from-sip:1 Up
AgentCallbackLogin(6025|6025 at A
SIP/6035-4ef0 *6035 at from-sip:1 Up
AgentCallbackLogin(6035|6035 at A
At midnight, * automatically restarts, and now top is showing the
results above. I can only assume that it was the result of those channels.
[root at foxtrot html]# vmstat 5 5
procs -----------memory---------- ---swap-- -----io---- --system--
----cpu----
r b swpd free buff cache si so bi bo in cs us
sy id wa
0 0 0 764280 42584 153196 0 0 1 17 29 14 5
4 91 0
0 0 0 764280 42584 153196 0 0 0 9 2051 143 0
0 99 1
0 0 0 764280 42584 153196 0 0 0 30 2026 124 0
0 100 0
0 0 0 764280 42584 153196 0 0 0 0 2022 122 0
0 100 0
0 0 0 764280 42588 153192 0 0 0 18 2023 134 0
0 99 0
I also assume that the following is normal (something about threads ...)
[root at foxtrot html]# ps -ef|fgrep asterisk
root 31462 1 0 00:00 ? 00:00:00 /bin/sh
/usr/sbin/safe_asterisk
root 31473 31462 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31514 31473 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31516 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31517 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31518 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31519 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31520 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31521 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31522 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31523 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31524 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31525 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31526 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31527 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31528 31514 0 00:00 ? 00:00:01 /usr/sbin/asterisk -vvvg -c
root 31529 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31530 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31531 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31532 31514 0 00:00 ? 00:00:00 /usr/sbin/asterisk -vvvg -c
root 31646 30821 0 00:34 pts/0 00:00:00 fgrep asterisk
Many thanks for all the help.
Julian.
Steve Kann wrote:
> Julian Lyndon-Smith wrote:
>
>> We had to move from a old * server to a new one in a hurry (hardware
>> failure). The old server was a dual pentium 700 with 512MB ram running
>> fedora core 2, the new one is a single 3GHz Pentium with 1gb ram.
>>
>> The same number of people are connected to the new server as the old,
>> the same number of inbound calls to the isdn30 etc (on average 20
>> calls active at any time (SIP and ZAP)). Basically, just a server
>> swapout.
>>
>> I must be reading top wrong, because the old server had a idle of
>> approx 30%, whereas the new server is
>>
>> top - 13:35:21 up 12 days, 23:57, 1 user, load average: 7.11, 7.20,
>> 7.21
>> Tasks: 98 total, 9 running, 89 sleeping, 0 stopped, 0 zombie
>> Cpu(s): 99.0% us, 1.0% sy, 0.0% ni, 0.0% id, 0.0% wa, 0.0% hi,
>> 0.0% si
>> Mem: 1034640k total, 144792k used, 889848k free, 21952k buffers
>> Swap: 2031608k total, 0k used, 2031608k free, 61248k cached
>>
>> Notice the 99.0% us. This fluctuates between 80 and 99%.
>>
>> The other difference is that the new server is on cvs-head as of today
>> - I did say that it was an emergency :) whereas the old server was
>> cvs-head from june sometime.
>>
>> Is it just me, or is there a problem ?
>
>
>
> Well, I suppose it depends on what's using all that CPU -- you leave
> that part out.
>
> Assuming it's asterisk threads causing the 100% cpu usage, and the load
> average of 7, then, yes, that's a lot of CPU.
>
> But, you could have some other program/processes doing that, and if
> they're batch processes (they're clearly not niced), they may end up
> with a lower dynamic priority and not affect asterisk too much.
>
> -SteveK
>
>>
>> Julian.
>> _______________________________________________
>> --Bandwidth and Colocation sponsored by Easynews.com --
>>
>> Asterisk-Users mailing list
>> Asterisk-Users at lists.digium.com
>> http://lists.digium.com/mailman/listinfo/asterisk-users
>> To UNSUBSCRIBE or update options visit:
>> http://lists.digium.com/mailman/listinfo/asterisk-users
>>
>
> _______________________________________________
> --Bandwidth and Colocation sponsored by Easynews.com --
>
> Asterisk-Users mailing list
> Asterisk-Users at lists.digium.com
> http://lists.digium.com/mailman/listinfo/asterisk-users
> To UNSUBSCRIBE or update options visit:
> http://lists.digium.com/mailman/listinfo/asterisk-users
>
>
More information about the asterisk-users
mailing list