deltacloud-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Dies Koper (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (DTACLOUD-251) CIMI client slow to list machines
Date Mon, 22 Oct 2012 23:50:12 GMT

    [ https://issues.apache.org/jira/browse/DTACLOUD-251?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13481943#comment-13481943
] 

Dies Koper commented on DTACLOUD-251:
-------------------------------------

The problem seems to be in machines.rb and machine.rb:
For every machine operation (show, start, stop, restart), for every machine, the hardware_profiles
are loaded.
Certainly, for start, stop and restart there is no need for them.
For show, there is no need to load them again and again for each machine, rather it could
be loaded only once.

This area needs a re-design. I can take a stab at it, but not being familiar with your coding
style I doubt I can come up with a patch you'd easily accept :)

                
> CIMI client slow to list machines
> ---------------------------------
>
>                 Key: DTACLOUD-251
>                 URL: https://issues.apache.org/jira/browse/DTACLOUD-251
>             Project: DeltaCloud
>          Issue Type: Bug
>          Components: Server
>         Environment: Windows 7, FGCP driver, CIMI client
>            Reporter: Dies Koper
>            Assignee: Michal Fojtik
>
> When I list "instances" in the DC GUI with the FGCP driver and look at the requests made
by the driver client to the FGCP API server, I see that the driver's instances method was
called once. With a user account on a system with many (>50) instances spread over 30 virtual
systems, it takes 15 seconds.
> When I list cimi/machines on the cimi client with the FGCP driver and look at the requests
made by the driver client, it first invokes hardware_profiles three times, then invokes the
driver's instances method once, then calls the hardware_profiles operation 410 times. This
takes 160 seconds so the browser shows a timeout page (RestClient::RequestTimeout at /cimi/machines).
> Looking at only the latter case's results, the hardware_profiles could have been called
directly from the cimi client or internally (from images) but as it doesn't happen with the
normal DC GUI, I'm reporting this against the CIMI client first.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message