On Tuesday, November 4, 2014 9:50:15 AM UTC-5, jcbollinger wrote:
>
> By far the biggest difference is is File retrieval time.  This will be for 
> File resources where you specify content via a 'source' parameter rather 
> than via a 'content' property.  The agent must make a separate request to 
> the master for each such file, and those are collectively taking a long 
> time.  Most likely they are backing up behind a bottleneck, so that much of 
> the time consumed per node is actually spent waiting for service.
>

I agree, there is something odd about File retrieval. We are creating a 
total of 10 autofs maps, however when I run trace, it looks like every 
separate entry creates a concat::fragment (puppetlabs-concat) which is 
later combined into a single file. I'm not too familliar with what 
concat::fragment does, but a quick scan of the code shows a lot of "source" 
references. We might have to just go to template files, instead of using 
the autofs module. That might take care of all our issues.

 

> If the CPU is not overloaded and you have free physical RAM then It seems 
> to me that the system service (i.e. httpd) and the I/O subsystem are your 
> remaining candidates for the locus of the issue.  As you attempt to 
> identify the bottleneck, do not ignore the total number of transactions 
> serviced.  That could be as important as -- or even more important than -- 
> the volume of data exchanged.
>
I've started graphing iostat and after a day of data will switch back to 30 
minute interval to see what is hit the hardest. 
 
Henrik, I might have to go to 1.9.3 regardless, because foreman-rake does 
not work with the new rake in 2.1. Something to do with routes... I'm not a 
ruby/rake person so can't fix it myself. 

-- 
You received this message because you are subscribed to the Google Groups 
"Puppet Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/puppet-users/1ed2e033-5261-49db-83ca-e21315742acf%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to