[openstack-dev] [nova] gauging feelings on a new config option for powervm_lpar_operation_timeout

Matt Riedemann mriedem at us.ibm.com
Mon Aug 12 15:23:48 UTC 2013

Russell, with powervm the hypervisor (IVM) is running on a detached system 
and depending on how many compute nodes are going through the same 
hypervisor I would think the load would vary as it processes multiple 
requests.  In my runs with Tempest I haven't seen any timeouts in the 
power_off operation, but I have seen other problems which are making me 
conscious of timeout issues with the powervm driver, mainly around taking 
a snapshot of a running instance (but that's a different issue related 
more to disk performance on the IVM and network topology - still 

Honestly, I just don't like hard-coded timeouts which can't be configured 
if the need arises.  I don't know why there is a timeout argument in the 
code that defaults to 30 seconds if it can't be overridden (or is 
overridden in the code).  We could do like the libvirt driver and just not 
have a timeout on stop/start, but that scares me for some reason.


Advisory Software Engineer
Cloud Solutions and OpenStack Development

Phone: 1-507-253-7622 | Mobile: 1-507-990-1889
E-mail: mriedem at us.ibm.com

3605 Hwy 52 N
Rochester, MN 55901-1407
United States

From:   Russell Bryant <rbryant at redhat.com>
To:     openstack-dev at lists.openstack.org, 
Date:   08/12/2013 09:49 AM
Subject:        Re: [openstack-dev] [nova] gauging feelings on a new 
config option for powervm_lpar_operation_timeout

On 08/11/2013 04:04 PM, Matt Riedemann wrote:
> While working on a patch to implement hard reboot for the powervm driver
> [1], I noticed that the stop_lpar (power_off) method has a timeout
> argument with a default value of 30 seconds but it's not overridden
> anywhere in the code and it's not configurable.  The start_lpar
> (power_on) method doesn't have a timeout at all.  I was thinking about
> creating a patch to (1) make start_lpar poll until the instance is
> running and (2) making the stop/start timeouts configurable with a new
> config option, something like powervm_lpar_operation_timeout that
> defaults to 60 seconds.

Why would someone change this?  What makes one person's environment need
a different timeout than another?  If those questions have good answers,
it's probably fine IMO.

Russell Bryant

OpenStack-dev mailing list
OpenStack-dev at lists.openstack.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20130812/1d29fea2/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/gif
Size: 1851 bytes
Desc: not available
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20130812/1d29fea2/attachment.gif>

More information about the OpenStack-dev mailing list