[Swift-devel] Coaster CPU-time consumption issue

Mihael Hategan hategan at mcs.anl.gov
Mon Jul 13 14:34:04 CDT 2009


On Mon, 2009-07-13 at 14:30 -0500, Mihael Hategan wrote:
> On Mon, 2009-07-13 at 14:18 -0500, Michael Wilde wrote:
> > On 7/13/09 2:06 PM, Mihael Hategan wrote:
> > > A while ago I committed a patch to run the service process with a lower
> > > priority. Is that in use?
> > > 
> > > Also, is logging reduced or is it the default?
> > > 
> > > Is the 97% CPU usage a spike, or does it stay there on average?
> > 
> > In the test I observed Sarah running last Thu, it stayed close to 100%
> > during the whole run - many minutes, solid near-100% CPU. During that
> > time a tail of the coaster log showed a burst of a few messages every
> > few seconds - not intensive enough to explain the overhead as all due to
> > logging.
> 
> Ok, that does look like a problem. I need to see the log from that.

However, I want to stress out that it may NOT be the same problem in all
cases of high CPU usage. So reduced logging should still be used before
trying to reproduce this specific problem.

> 
> In addition, when you observe this SPECIFIC behavior (solid near 100%
> CPU, burst of a few messages every few seconds and not much else in the
> logs), please do a jstack <pid> on the process in question and send the
> output of that.
> 
> 
> _______________________________________________
> Swift-devel mailing list
> Swift-devel at ci.uchicago.edu
> http://mail.ci.uchicago.edu/mailman/listinfo/swift-devel




More information about the Swift-devel mailing list