Re: [openstack-dev] [TripleO][Heat] using convergence_engine to deploy overcloud stack

[Ben Nemec]

On 08/09/2017 05:49 AM, Rabi Mishra wrote:
On Wed, Aug 9, 2017 at 1:41 PM, Smigielski, Radoslaw (Nokia - IE) 
mailto:radoslaw.smigiel...@nokia.com>> 
wrote:


Hi there!

I have a question about heat "convergence_engine" option, it's
present in heat config since quite a long time but still not enabled.

Well, convergence is enabled by default in heat since newton. However, 
Tripleo does not use it yet, as convergence engine memory usage is 
higher than that of legacy engine.


There has been number of optimizations in the last two cycles to improve 
that situation. However, AFAIK, when using a single node undercloud, 
memory usage would always be more with convergence.


I think there are plans for Tripleo to move to convergence in Queens as 
discussed in this ML thread.


http://lists.openstack.org/pipermail/openstack-dev/2017-June/118237.html


Yeah, my understanding is that we'll look at turning on convergence for 
the undercloud by default in Queens.  It was just getting too late in a 
cycle of full of huge architectural changes to flip the switch in Pike.


If there's no indication of trouble from the experimental job we should 
definitely turn it on ASAP once Queens opens though.




And I am wondering if anyone has tried enabled it and deploy
overcloud? I did it myself and it seems to be working.

The main reason why I am looking at this options are problems with
scaling out, adding computes, replacing failed controllers on setups
with 50+ computes and number of nested stack 1000+.


You've to probably scale out undercloud heat to get most of the benefits 
of convergence that comes with the distributed architecture.


Is there any reason what holds Heat switch to convergence architecture?


_
*Radosław Śmigielski*
Nokia CBIS R&D

__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe:
openstack-dev-requ...@lists.openstack.org?subject:unsubscribe

http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev





--
Regards,
Rabi Misra



__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev



__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Re: [openstack-dev] [TripleO][Heat] using convergence_engine to deploy overcloud stack

[Sagi Shnaidman]

Hi there

On Wed, Aug 9, 2017 at 1:49 PM, Rabi Mishra  wrote:

> On Wed, Aug 9, 2017 at 1:41 PM, Smigielski, Radoslaw (Nokia - IE) <
> radoslaw.smigiel...@nokia.com> wrote:
>
>> Hi there!
>>
>>I have a question about heat "convergence_engine" option, it's
>> present in heat config since quite a long time but still not enabled.
>>
> Well, convergence is enabled by default in heat since newton. However,
> Tripleo does not use it yet, as convergence engine memory usage is higher
> than that of legacy engine.
>
>
TripleO CI has heat-convergence job running on heat patches in experimental
pipeline [1] It runs there during last year at least.
No any high memory usage was detected in last few months when I watched it.

[1]
https://github.com/openstack-infra/project-config/blob/master/zuul/layout.yaml#L10407

Thanks

-- 
Best regards
Sagi Shnaidman
__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Re: [openstack-dev] [TripleO][Heat] using convergence_engine to deploy overcloud stack

[Rabi Mishra]

On Wed, Aug 9, 2017 at 1:41 PM, Smigielski, Radoslaw (Nokia - IE) <
radoslaw.smigiel...@nokia.com> wrote:

> Hi there!
>
>I have a question about heat "convergence_engine" option, it's present
> in heat config since quite a long time but still not enabled.
>
Well, convergence is enabled by default in heat since newton. However,
Tripleo does not use it yet, as convergence engine memory usage is higher
than that of legacy engine.

There has been number of optimizations in the last two cycles to improve
that situation. However, AFAIK, when using a single node undercloud, memory
usage would always be more with convergence.

I think there are plans for Tripleo to move to convergence in Queens as
discussed in this ML thread.

http://lists.openstack.org/pipermail/openstack-dev/2017-June/118237.html

And I am wondering if anyone has tried enabled it and deploy overcloud? I
> did it myself and it seems to be working.
>
> The main reason why I am looking at this options are problems with scaling
> out, adding computes, replacing failed controllers on setups with 50+
> computes and number of nested stack 1000+.
>

You've to probably scale out undercloud heat to get most of the benefits of
convergence that comes with the distributed architecture.

> Is there any reason what holds Heat switch to convergence architecture?
>
>
> _
> *Radosław Śmigielski*
> Nokia CBIS R&D
>
> __
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
>


-- 
Regards,
Rabi Misra
__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev