Re: slurpd high system load

[John Beamon <jbeamon@franklinamerican.com>]

That fixed it.  It just fixed it.  I changed the replogfile from 
"slurpd.replog" to "slapd.replog" in the same directory.  slapd.replog 
stays at 0 bytes, but my CPU utilization is now at about 40% and my 
system load's cut in half.

As an open inquiry, why is this not documented anywhere?  Every 
documentation I've ever read says to use

replog  /path/to/slurpd.replog

at the bottom of slapd.conf, below my replicas.  Not one document 
anywhere has ever told me to let slurpd choose that one by default and 
set another in slapd.conf.  I don't know the mechanics well enough to 
justify it, but this change just worked.

-j


Andreas wrote:
> On Thu, May 29, 2003 at 10:07:48AM -0500, John Beamon wrote:
>
> > time and never relinquishes it.  It does not grab "91%" and hold that 
> > number, but it maintains a 0.0% idle CPU.  This is a recent observation 
>
>
> I can think of two possible causes regarding configuration errors:
>
> 1) The slave is also configured to replicate to the master. This will
> get you a nice loop, and the replication log will never empty.
>
> 2) Have the master use a reserved name for the replogfile. Slurpd uses
> slurpd.replog for its private copy. If you happen to set the replog
> directive to slurpd.replog, then slurpd will loop.


--

John Beamon
Systems Administrator
Franklin American Mortgage
eml: jbeamon@franklinamerican.com
web: www.franklinamerican.com