Thanks for you suggestion. Your suggestion hit the point: It's the logging
problem.
I cut it down by adjusting the logging level of one process. I will try it
on other processes.

The new top info:
  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND


 2068 admin     15   0  157m  11m 1588 S 49.5  0.2 391:49.01 supervisord

python version: Python 2.7.2

supervisord version: 3.0a12

uname -a : Linux service1 2.6.18-274.el5xen #1 SMP Fri Jul 8 17:45:44 EDT
2011 x86_64 x86_64 x86_64 GNU/Linux

ps flax:
1   500  2068     1  16   0 161020 11664 -      Ss   ?        392:02
/home/admin/depLib/python/bin/python
/home/admin/depLib/python/bin/supervisord -c /home/
0   500  2070  2068  18   0 1269428 5424 futex_ Sl   ?          2:24  \_
tradenotify.bin
0   500  2071  2068  15   0 476364 19692 923320 Sl   ?          1:22  \_
python user_consumer.py 1
0   500  2072  2068  15   0 476360 19688 923320 Sl   ?          1:22  \_
python user_consumer.py 0
0   500  2073  2068  15   0 476364 19692 923320 Sl   ?          1:21  \_
python user_consumer.py 3
0   500  2074  2068  15   0 541888 19692 923320 Sl   ?          1:21  \_
python user_consumer.py 2
0   500  2076  2068  15   0 21517156 64676 351886 Sl ?         59:18  \_
python trade_process_realtime.py 4
0   500  2077  2068  15   0 25943696 72644 351886 Sl ?         62:24  \_
python trade_process_realtime.py 0
0   500  2078  2068  15   0 4177976 33480 351886 Sl  ?         58:45  \_
python trade_process_realtime.py 1
0   500  2079  2068  15   0 9711332 42824 147035 Sl  ?         58:59  \_
python trade_process_realtime.py 2
0   500  2080  2068  15   0 3993472 33216 351886 Sl  ?         58:51  \_
python trade_process_realtime.py 3
0   500  2081  2068  15   0 6569172 32068 429496 Sl  ?          4:21  \_
python trade_consumer.py 1
0   500  2082  2068  15   0 6692240 32176 351886 Sl  ?          4:20  \_
python trade_consumer.py 0
0   500  2083  2068  15   0 6876608 32788 351886 Sl  ?          4:23  \_
python trade_consumer.py 3
0   500  2084  2068  15   0 6999684 32836 351886 Sl  ?          4:20  \_
python trade_consumer.py 2
0   500  2085  2068  18   0 1519028 57196 futex_ Sl  ?         10:33  \_
service_publisher.bin
0   500 13470  2068  25   0 755100 30404 184466 Sl   ?          3:35  \_
transmit_service.bin
0   500 14807  2068  15   0 456696 20500 351886 Sl   ?          0:26  \_
python trade_processor.py -n 5 -f
0   500 14808  2068  15   0 456704 20500 560123 Sl   ?          0:26  \_
python trade_processor.py -n 4 -f
0   500 14825  2068  15   0 456696 20504 351886 Sl   ?          0:27  \_
python trade_processor.py -n 7 -f
0   500 14835  2068  15   0 456700 20488 351886 Sl   ?          0:27  \_
python trade_processor.py -n 6 -f
0   500 14844  2068  15   0 456700 20512 351886 Sl   ?          0:27  \_
python trade_processor.py -n 1 -f
0   500 14853  2068  15   0 456700 20504 351886 Sl   ?          0:26  \_
python trade_processor.py -n 0 -f
0   500 14862  2068  15   0 456700 20504 505984 Sl   ?          0:27  \_
python trade_processor.py -n 3 -f
0   500 14871  2068  15   0 456700 20508 351886 Sl   ?          0:26  \_
python trade_processor.py -n 2 -f
0   500 14886  2068  15   0 418796 19428 184466 Sl   ?          1:12  \_
python logon_process.py


Thank you very much.





2012/11/5 Timothy Jones <[email protected]>

>  My first suggestion would be to check for excessive output from the
> child processes.  Sometimes someone will turn on debug logging in the child
> processes, and supervisor read its all, and echos it to more logfiles.****
>
> ** **
>
> When you reply, please include the output of ‘ps flax’, starting with the
> supervisor process and including all of its child processes.****
>
> Also, include platform, contents of ‘uname –a’ and /etc/issue, python
> version, supervisor version, etc.****
>
> ** **
>
> tlj****
>
> *From:* [email protected] [mailto:
> [email protected]] *On Behalf Of *??
> *Sent:* Monday, November 05, 2012 12:40 AM
> *To:* [email protected]
> *Subject:* [BULK] [Supervisor-users] Help: The Supervisord cpu load is
> too high. How to cut it down?****
>
> ** **
>
> The supervisord occupied 96.3% CPU. The machine has 4 cores. ****
>
> How to cut it down, Is there any method?****
>
> Thanks.****
>
> ** **
>
> The top info:****
>
>  ****
>
>   PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
>
>       ****
>
>  2068 admin     18   0  157m  11m 1588 S 96.3  0.2 353:22.63 supervisord *
> ***
>
_______________________________________________
Supervisor-users mailing list
[email protected]
http://lists.supervisord.org/mailman/listinfo/supervisor-users

Reply via email to