couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Cliffano Subagio <cliff...@gmail.com>
Subject high futex on strace
Date Thu, 03 Nov 2011 06:01:19 GMT
Hi,

My environment has 2 couchdb servers replicating to each other. Each
couchdb instance has 2 databases, 1 database has no view, the other has 9
views which constantly get updated due to heavy document creates & updates.
Each server has 2 cpu cores, and I beam.smp process is using 140% cpu, with
2 couchjs processes using ~10% each.

Is the high cpu usage by beam.smp normal?
Is there any config tweaking that can be done to reduce beam.smp cpu usage?


>From strace, I found out that there's a high futex (almost 70%) which
indicates some sort of locking. ps: my system-fu is low - please CMIIW.
% time     seconds  usecs/call     calls    errors syscall
------ ----------- ----------- --------- --------- ----------------
 67.78  132.723912          42   3128236    339042 futex
 25.45   49.834421      155733       320         1 select
  2.48    4.849559        6154       788           fsync
  2.22    4.352599           5    944252           epoll_wait
  0.58    1.143640           1    877069           sched_yield

Is this high futex normal?



Cheers,
Cliff.

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message