You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@cloudstack.apache.org by "Wido den Hollander (JIRA)" <ji...@apache.org> on 2014/09/29 00:42:34 UTC

[jira] [Closed] (CLOUDSTACK-7641) Do not always ask libvirt to refresh storage pools

     [ https://issues.apache.org/jira/browse/CLOUDSTACK-7641?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Wido den Hollander closed CLOUDSTACK-7641.
------------------------------------------
    Resolution: Fixed

Merged into master

> Do not always ask libvirt to refresh storage pools
> --------------------------------------------------
>
>                 Key: CLOUDSTACK-7641
>                 URL: https://issues.apache.org/jira/browse/CLOUDSTACK-7641
>             Project: CloudStack
>          Issue Type: Improvement
>      Security Level: Public(Anyone can view this level - this is the default.) 
>          Components: KVM
>    Affects Versions: Future
>            Reporter: Wido den Hollander
>            Assignee: Wido den Hollander
>            Priority: Minor
>              Labels: ceph, libvirt, rbd
>             Fix For: 4.5.0
>
>
> When getStoragePool() in the LibvirtStorageAdapter is called it asks libvirt for information regarding that storage pool.
> In the process the StorageAdapter asks libvirt to 'refresh' the storage pool.
> Refreshing a storage pool means that libvirt will scan the whole pool and get information about all the volumes which it then caches.
> getStoragePool() is called for almost every storage operation in KVM and that causes delays.
> On big (RBD) storage pools such a refresh can take up to 2 or 3 minutes.
> The StatsCollector in the Management Server asks all the hosts about the storage pools as well, so we should only refresh the pool when the StatsCollector wants storage statistics.
> This way we keep the information in libvirt up to date, but we also speed up storage tasks on KVM.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)