Re: lib/41019: pthreads: high load when using varnish (locking/unlocking causes 98% WCPU)

[Andrew Doran <ad%netbsd.org@localhost>]

The following reply was made to PR lib/41019; it has been noted by GNATS.

From: Andrew Doran <ad%netbsd.org@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: lib/41019: pthreads: high load when using varnish 
(locking/unlocking causes 98% WCPU)
Date: Mon, 16 Mar 2009 00:06:14 +0000

 On Sun, Mar 15, 2009 at 09:35:02PM +0000, Michael Stapelberg wrote:
 
 >  kdump consists of *LOTS* of calls like this:
 >   21193      2 varnishd RET   clock_gettime 0
 >   21193      2 varnishd CALL  clock_gettime(0,0x7f7ff7bffab0)
 
 Presumably that's from the debug output you have enabled.
 
 >   21193      2 varnishd RET   clock_gettime 0
 >   21193      2 varnishd CALL  clock_gettime(0,0x7f7ff7bffab0)
 >   21193      2 varnishd RET   clock_gettime 0
 >   21193      2 varnishd CALL  clock_gettime(0,0x7f7ff7bffab0)
 >   21193      2 varnishd RET   clock_gettime 0
 >  
 >  >  This sleeps for wthread_purge_delay after garbage collecting a thread.
 >  >  It also sleeps for the same between scans.
 >  >  What do you have it set to?
 >  param.show thread_pool_purge_delay
 >  200 828     
 
 Not sure how to interpret that.
 
 Is it making it to any of the usleep() calls in the code?
 I'm not seeing a threading bug (yet).