Issue #4268 has been updated by Markus Roberts.

File watch_type_collection.patch added
Priority changed from High to Normal
Target version set to 2.6.1

After several rounds of diagnostics with the Peter and Brice we've managed to 
get a little more clarity on the problem.  A fair fraction of the performance 
problems were algorithmic (we were recomputing certain properties for each 
defined resource multiple times, roughly proportionate to the number of defined 
resources).  These have been fixed in master for 2.6.0 in 
commit:dd03ac9fa29fce36eb64a5f831be8757f2f96f5c

The remainder of the problem appears to be over-cautious thread isolation; 
specifically, the first and second run both load / parse the entire manifest 
because they are using two separate known_type TypeCollections in two separate 
production Environments (presumed, but not established) to be in two separate 
threads.

This can be seen (but so far only at Peter's site) with the attached patch.


----------------------------------------
Bug #4268: puppet master 2.6 always imports manifests newly
http://projects.puppetlabs.com/issues/4268

Author: Peter Meier
Status: Needs more information
Priority: Normal
Assigned to: Peter Meier
Category: 
Target version: 2.6.1
Affected version: 2.6.0rc3
Keywords: 
Branch: 


Disclaimer: I'm not sure if that this is the intended behavior, but I think 
this behavior slows things very much down.

I'm currently testing 2.6 and I'm running a webrick instance by:
<pre>
/usr/bin/ruby /usr/bin/puppet master --config /srv/puppet/etc/puppet.conf 
--no-daemonize --masterport 8141 --verbose --trace
</pre>

If I run a client twice in a row I get for both runs all the autoimport 
messages, however nothing has changed on disk. So it looks to me that puppet 
master is always importing things and not only once and then reuse things it 
already loaded.

Furthermore this seems to slow down things enormously: While on 0.25.5 in a 
mongrel setup the compilation took about 10s it now takes 110s on the very same 
box which has even less load as only one master process is running and there is 
only one client connecting.

I wasn't yet able to test it with our mongrel setup, but I'm somehow a bit 
feared by this slowdown. But maybe this slowdown is also caused by something 
else of the parameters I have given to puppetmaster, but I didn't yet see which 
ones.


-- 
You have received this notification because you have either subscribed to it, 
or are involved in it.
To change your notification preferences, please click here: 
http://projects.puppetlabs.com/my/account

-- 
You received this message because you are subscribed to the Google Groups 
"Puppet Bugs" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/puppet-bugs?hl=en.

Reply via email to