Bug 589 - Abnormal load to one of threads
Abnormal load to one of threads
Status: ASSIGNED
Product: NSD
Classification: Unclassified
Component: NSD Code
4.0.x
Other Linux
: P5 normal
Assigned To: NSD team
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2014-06-30 09:55 CEST by Sergei Mamonov
Modified: 2014-06-30 10:41 CEST (History)
1 user (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Sergei Mamonov 2014-06-30 09:55:43 CEST
All nsd threds use ~1% cpu, one about 5% and one use 10-10+% !
nsd-control status
version: 4.0.4
verbosity: 2
ratelimit: 50

ps aux | grep "nsd -c"
bind     22350  0.3  2.0 451744 416784 ?       Ss   11:44   0:01 nsd -c /etc/nsd/nsd.conf
bind     22502  0.0  3.5 2272176 730692 ?      S    11:45   0:00 nsd -c /etc/nsd/nsd.conf
bind     22503  0.8  3.6 2272608 758932 ?      S    11:45   0:02 nsd -c /etc/nsd/nsd.conf
bind     22504  0.6  3.6 2272608 751048 ?      S    11:45   0:01 nsd -c /etc/nsd/nsd.conf
bind     22505  5.4  3.7 2273320 762352 ?      S    11:45   0:14 nsd -c /etc/nsd/nsd.conf
bind     22506 12.0  3.5 2272608 733668 ?      S    11:45   0:31 nsd -c /etc/nsd/nsd.conf

And one thread worked with more querise - 
nsd-control stats_noreset
server0.queries=35659
server1.queries=12900
server2.queries=189366
server3.queries=9835
num.queries=247760
....
nsd-control stats_noreset
server0.queries=61680
server1.queries=22108
server2.queries=325446
server3.queries=11440
num.queries=420674


We try restart it, but server2.queries have more more queries  all the same.

And it have 6 process, but in nsd.conf " server-count: 4".
Comment 1 Wouter Wijngaards 2014-06-30 10:41:34 CEST
Hi Sergei,

This is the operating system that is dividing the workload.  Because most threads are unused (largely unused), it can choose whatever it likes, and does not choose to divide the load equally.

Best regards,
   Wouter