On 18/06/12 18:22, Jo Rhett wrote:
> On Jun 17, 2012, at 4:22 AM, Dominic Cleal wrote:
>> Yes, prior to 3.0 it is a lot slower. The reason's simple, it loads and
>> parses every file unless you explicitly tell it (via lens/incl params)
>> which file you intend to edit.
>>
>> 3.0 has an optimisatio
On Jun 17, 2012, at 4:22 AM, Dominic Cleal wrote:
> Yes, prior to 3.0 it is a lot slower. The reason's simple, it loads and
> parses every file unless you explicitly tell it (via lens/incl params)
> which file you intend to edit.
>
> 3.0 has an optimisation (#7285) so if you use the context param
On 16/06/12 20:42, Patrick Carlisle wrote:
>
>
> On Sat, Jun 16, 2012 at 9:11 AM, Brice Figureau
> mailto:brice-pup...@daysofwonder.com>>
> wrote:
>
> On 13/06/12 23:46, Daniel Pittman wrote:
> > On Wed, Jun 13, 2012 at 1:32 PM, Peter Meier
> mailto:peter.me...@immerda.ch>> wrote:
>
On 12/06/12 11:47, Trevor Vaughan wrote:
> 3) Augeas is slower that the native concat type that we posted to
> Github (though that needs some polish)
Yes, prior to 3.0 it is a lot slower. The reason's simple, it loads and
parses every file unless you explicitly tell it (via lens/incl params)
whic
On Sat, Jun 16, 2012 at 9:11 AM, Brice Figureau <
brice-pup...@daysofwonder.com> wrote:
> On 13/06/12 23:46, Daniel Pittman wrote:
> > On Wed, Jun 13, 2012 at 1:32 PM, Peter Meier
> wrote:
> >>
> >>> Actually, this is a group where y'all probably *can* deliver on
> >>> this request:
> >>>
> >>> I
On 13/06/12 23:46, Daniel Pittman wrote:
> On Wed, Jun 13, 2012 at 1:32 PM, Peter Meier wrote:
>>
>>> Actually, this is a group where y'all probably *can* deliver on
>>> this request:
>>>
>>> If you are experiencing poor performance, it would be awesome if
>>> you could generate some profiling dat
On Wed, 2012-06-13 at 13:52 -0700, Luke Kanies wrote:
> On Jun 13, 2012, at 1:22 PM, Peter Meier wrote:
>
> > -BEGIN PGP SIGNED MESSAGE-
> > Hash: SHA1
> >
> >> I would really appreciate that. I recently moved a larger
> >> installation to 2.7 (latest) and I'm probably going back, as the
This would be a great wiki page ;-)
On Jun 13, 2012, at 2:46 PM, Daniel Pittman wrote:
> On Wed, Jun 13, 2012 at 1:32 PM, Peter Meier wrote:
>>
>>> Actually, this is a group where y'all probably *can* deliver on
>>> this request:
>>>
>>> If you are experiencing poor performance, it would be awe
On Jun 13, 2012, at 8:52 AM, Luke Kanies wrote:
> Would you be willing to share your configs with us so we can do some
> performance analysis on them? Starting with known-slow configs would
> make a big difference.
I am no longer an employee at the shop, but a long time ago I sent our entire
ca
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
>> Although, I do a lot of hiera lookups now, but also switched
>> from full storeconfigs to thin storeconfigs, I think the
>> difference is huge and I'm wondering what might cause it.
>
> fwiw hiera isn't caching - there was a caching yaml backend
>
- Original Message -
> From: "Peter Meier"
> To: puppet-dev@googlegroups.com
> Sent: Wednesday, June 13, 2012 11:21:47 PM
> Subject: Re: [Puppet-dev] why I don't feel that ENCs are very functional
> right now (performance)
>
> -BEGIN PGP SIGNE
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
> I committed the caching YAML code to master, and it should be
> merged. Just boarding a plane, but can link when I land :)
I thought that have already be included in one of the 1.0.0 hiera
rc-releases? Or are you talking about a different cache?
I'
On Wed, Jun 13, 2012 at 1:32 PM, Peter Meier wrote:
>
>> Actually, this is a group where y'all probably *can* deliver on
>> this request:
>>
>> If you are experiencing poor performance, it would be awesome if
>> you could generate some profiling data for us.
>>
>> perftools.rb is the nicest profil
On Wednesday, June 13, 2012, Peter Meier wrote:
> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA1
>
> > Actually, this is a group where y'all probably *can* deliver on
> > this request:
> >
> > If you are experiencing poor performance, it would be awesome if
> > you could generate some profiling d
On Jun 13, 2012, at 1:22 PM, Peter Meier wrote:
> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA1
>
>> I would really appreciate that. I recently moved a larger
>> installation to 2.7 (latest) and I'm probably going back, as the
>> slow down + extended memory usage is just not worth it.
>>
>> Co
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
> Actually, this is a group where y'all probably *can* deliver on
> this request:
>
> If you are experiencing poor performance, it would be awesome if
> you could generate some profiling data for us.
>
> perftools.rb is the nicest profiler for Ruby I
- Original Message -
> From: "Peter Meier"
> To: puppet-dev@googlegroups.com
> Sent: Wednesday, June 13, 2012 9:22:30 PM
> Subject: Re: [Puppet-dev] why I don't feel that ENCs are very functional
> right now (performance)
>
> -BEGIN PGP SIGNE
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
> I would really appreciate that. I recently moved a larger
> installation to 2.7 (latest) and I'm probably going back, as the
> slow down + extended memory usage is just not worth it.
>
> Compile time doubled in my case (~60s compared to 27s). We
> m
On Wed, Jun 13, 2012 at 8:52 AM, Luke Kanies wrote:
> On Jun 12, 2012, at 8:55 PM, Peter Meier wrote:
>>
I'm hoping that 3.0 brings some significant performance
improvements.
>>>
>>> We discovered some functionality regressions that we are working
>>> on right now and hope to have out o
On Jun 12, 2012, at 4:55 PM, Trevor Vaughan wrote:
> The ENC interface appears to be slow in general.
>
> I can't say much about Dashboard and Foreman given their use of
> storedconfigs, so I'm not sure where the bottlenecks are there
> exactly.
>
> It *seems* to be that the base ENC gets read
On Jun 12, 2012, at 8:55 PM, Peter Meier wrote:
> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA1
>
>>> I'm hoping that 3.0 brings some significant performance
>>> improvements.
>>>
>>
>> We discovered some functionality regressions that we are working
>> on right now and hope to have out of the
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
>> I'm hoping that 3.0 brings some significant performance
>> improvements.
>>
>
> We discovered some functionality regressions that we are working
> on right now and hope to have out of the way soon. We've also made
> the decision to hold off on re
The ENC interface appears to be slow in general.
I can't say much about Dashboard and Foreman given their use of
storedconfigs, so I'm not sure where the bottlenecks are there
exactly.
It *seems* to be that the base ENC gets read multiple times instead of
cached the first time, but that might jus
On Jun 12, 2012, at 3:47 AM, Trevor Vaughan wrote:
> Hi Jo,
>
> I've experienced some of the same issues that you're highlighting
> while using Puppet internally at my company.
>
> My experience has shown:
>
> 1) 2.7 does slow things down significantly
Unfortunate, but true. If I understand co
Oh, I meant writing your own native types to replace the execs.
Look at the puppet-stdlib 'file_line' type for an easy to follow example.
If you can do this, you'll probably cut your run time dramatically.
Trevor
On Tue, Jun 12, 2012 at 3:48 PM, Jo Rhett wrote:
> On Jun 12, 2012, at 3:47 AM, T
On Jun 12, 2012, at 3:47 AM, Trevor Vaughan wrote:
> 4) Calling a large number of Exec's and Service's will cause some sort
> of undetermined queueing backup on your clients
Yeah, that was our diagnosis of the problem. At the former site, it was
necessary to use Execs to test package update issu
Trevor, Jo, all,
Thanks for your thoughts on the state of ENCs. We agree there are
numerous problems and are in fact in the process of building a new ENC
we hope will solve many if not most of these problems. Certainly, it's
still in the prototyping phase and is not nearly ready for production
use
Hi Jo,
I've experienced some of the same issues that you're highlighting
while using Puppet internally at my company.
My experience has shown:
1) 2.7 does slow things down significantly
2) ENC's are too slow to scale except in the case of LDAP where the
performance seems relatively reasonable as
I'm sure this thread is going to take some heat. Obviously what I post here is
very YMMV, and I guess that's my point -- MMDV and it just doesn't seem equal
just yet. That said, I'd be very happy to be enlightened about things I may
have overlooked. So please reply to this thread with a chance t
29 matches
Mail list logo
