cloudstack-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CLOUDSTACK-9175) [VMware DRS] Adding new host to DRS cluster does not participate in load balancing
Date Mon, 13 Mar 2017 02:03:04 GMT

    [ https://issues.apache.org/jira/browse/CLOUDSTACK-9175?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15906775#comment-15906775
] 

ASF GitHub Bot commented on CLOUDSTACK-9175:
--------------------------------------------

Github user serg38 commented on the issue:

    https://github.com/apache/cloudstack/pull/1257
  
    @sureshanaparti I've been thinking if there is better way to re-sync all portgroups and
I couldn't come up with easy alternative. With that said if we are copying portgroups for
newly added host I propose to do the same for host returning from Maintenance (CancelMaintenance).
It is very possible that host failed or was rebuild so it might not have all original portgroups
and DRS will be failing for it as well.



> [VMware DRS] Adding new host to DRS cluster does not participate in load balancing
> ----------------------------------------------------------------------------------
>
>                 Key: CLOUDSTACK-9175
>                 URL: https://issues.apache.org/jira/browse/CLOUDSTACK-9175
>             Project: CloudStack
>          Issue Type: Bug
>      Security Level: Public(Anyone can view this level - this is the default.) 
>          Components: Management Server, VMware
>    Affects Versions: 4.5.2
>            Reporter: Suresh Kumar Anaparti
>            Assignee: Suresh Kumar Anaparti
>
> When a new VMware host is added into a cluster, Cloudstack, by default, doesn't create
all the port groups present in the cluster. And since it doesn't have all the necessary networking
port groups (existing VM's port groups) it is not eligible to participate in DRS load balancing
or HA.
> Steps:
> 1. Have a DRS and HA cluster in fully automated mode, with two hosts H1 and H2 created
in the vCenter.
> 2. Configure this cluster in Cloudstack and create couple of VMs.
> 3. Start stressing the host by running some cpu hogging scripts in each of the VM.
> 4. Enable maintenance mode on one of the host - say H1 from Cloudstack.
> 5. Also, quickly enable maintenance mode on host H1 from vCenter.
> (This should migrate all the VMs to host H2) Make sure none of the VMs are present on
host H1.
> 6. Add host H3 into DRS cluster from vCenter and from Cloudstack as well.
> 7. At this point, the load is definitely imbalanced. This can be verified from vCenter
( Click on cluster -> Go to Summary tab -> under vSphere DRS section, it should show
'Load imbalanced'
> Now, as per DRS rules, the load should be balanced across all the available hosts.
> In this case, even after adding new host, the load is imbalanced. 
> The reason for the load imbalance is VMs (created from Cloudstack) are not eligible to
migrate to new host because networks or the cloud portgroups are not available on the new
host H3 (except for private).



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Mime
View raw message