incubator-deltacloud-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mark McLoughlin <mar...@redhat.com>
Subject Re: [jira] [Commented] (DTACLOUD-58) api/instance query performance with 8 vms in rhevm
Date Thu, 14 Jul 2011 09:17:03 GMT
On Thu, 2011-07-14 at 09:53 +0200, Michal Fojtik wrote:
> On Jul 14, 2011, at 12:16 AM, David Lutterkort (JIRA) wrote:
> 
> > 
> >    [ https://issues.apache.org/jira/browse/DTACLOUD-58?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13064913#comment-13064913
] 
> > 
> > David Lutterkort commented on DTACLOUD-58:
> > ------------------------------------------
> > 
> > The 'heartbeat' isn't coming from deltacloud, but from aeolus; 
> > reloading the list of all VM's every 30s is way too often.

This isn't what I expected actually

In aeolus, condor is what's monitoring running instances and I thought
it did that be GETing each instance individually every 90 seconds

> > At the same time, I don't understand why it takes RHEV-M this long to list 8 VM's
> > 
> > I am afraid deltacloud is just the messenger in this issue
> 
> Right, we're just forwarding requests to RHEV-M Rest API written on top of PowerShell
interface
> to RHEV-M.
> You can see how much it takes for backend to process your request by looking at response
headers.
> Search for 'X-Backend-Runtime:' header.

There's a known performance regression with the powershell
implementation of the API. Apparently this branch:

  https://github.com/markmc/rhevm-api/tree/0.9-milestone9.1-pmg

helps a lot, but we haven't been able to diagnose it further.

Cheers,
Mark.


Mime
View raw message