You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@cloudstack.apache.org by "Rohit Yadav (JIRA)" <ji...@apache.org> on 2014/10/17 17:16:34 UTC

[jira] [Commented] (CLOUDSTACK-7365) Upgrading without proper systemvm template corrupt cloudstack management server

    [ https://issues.apache.org/jira/browse/CLOUDSTACK-7365?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14175135#comment-14175135 ] 

Rohit Yadav commented on CLOUDSTACK-7365:
-----------------------------------------

For XenServer, make sure you register/upload this template:
http://cloudstack.apt-get.eu/systemvm/4.4/systemvm64template-4.4.0-6-xen.vhd.bz2
with name = "systemvm-xenserver-4.4" (watchout for spaces since the upgrade path uses this hardcoded value in a hashmap).

And check that its checksum is 7a8a6c4f8478147f1c6aa0d20f5e62b9 (this is of the unextracted vhd.bz2 file). Note that you should upload all systemvm templates for all kinds of hypervisor clusters you have and not just for the default one (look the at the upgrade path code).

I read the upgrade path again, so the issue is this upgrade path assumes that we're upgrading from 4.4.0 (not necessarily true), so it would fail in case 4.4.0 systemvm template is not found. Why it doing this? ([~dahn] should tell us since he added the upgrade systemvm method in 00c2696e7a1d1930a4088af6fc085b523b0b3589)
>From code, looks like it is simply changing the systemvm template user to SYSTEM user, updates template IDs and updates some configuration options.

The solution to this I think should be that it searches only for 4.4.0, so if no systemvms are available we should at least update the configuration update statements, but don't throw exception since a user maybe upgrading from 4.3.1 etc. Also because if I'm upgrading to 4.4.1 or above I won't be interested in using/having 4.4.0 systemvm templates, at some point I would upgrade to those newer templates.

Comments?

> Upgrading without proper systemvm template corrupt cloudstack management server
> -------------------------------------------------------------------------------
>
>                 Key: CLOUDSTACK-7365
>                 URL: https://issues.apache.org/jira/browse/CLOUDSTACK-7365
>             Project: CloudStack
>          Issue Type: Bug
>      Security Level: Public(Anyone can view this level - this is the default.) 
>          Components: Management Server
>    Affects Versions: 4.3.0, 4.4.0, 4.4.1
>            Reporter: Pierre-Luc Dion
>              Labels: upgrade
>         Attachments: 4.4.0to4.4.1_mgtlogissue.txt
>
>
> Since 4.3.0, also affecting 4.4.0 and 4.4.1. When upgrading CloudStack management server, it is required to have systemvm template properly named prior to the upgrade. otherwise the management server will fail to restart with after upgrading database schema.
> The possible repair method is to revert packages to previously installed CloudStack and restore the database which have been upgraded.
> This is not a viable upgrade path since management servers packages could be accidentally upgraded after a  "yum upgrade" or "apt-get update".
> Upgrading CloudStack management-server without previously uploading systemvm template should not fail to start the management-server. if the systemvm template is not in place, then the management-server cannot start.



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