[
https://issues.apache.org/jira/browse/CLOUDSTACK-8575?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Wilder Rodrigues resolved CLOUDSTACK-8575.
------------------------------------------
Resolution: Won't Fix
We are facing problem with the current behaviour of CloudStack.
When we kill the database, the management server doesn't die. However, when
database is back, not all the management server tasks are running.
For example:
1. Assuming the database is empty
a. start management server
b kill database
c. wait for 1 minute and restart the management server
d. deploy a datacenter
Remarks: zone, POD, cluster, host, primary/secondary storage. However, the SSVM
and CPVM are never created.
2. Assuming a datacenter exists
a. start management server
b kill database
c. wait for 1 minute and restart the management server
d. try to create a virtual machine
Remarks: router VM is never created; virtual machine creation fails
I will give it until today's (22nd June) to identify which task dies and if
that would be simple to fix. Otherwise, the task will be closed as a "Won't
fix" and a new approach will be suggested.
Cheers,
Wilder
> Manager Server dies if connection with database is lost
> -------------------------------------------------------
>
> Key: CLOUDSTACK-8575
> URL: https://issues.apache.org/jira/browse/CLOUDSTACK-8575
> Project: CloudStack
> Issue Type: Bug
> Security Level: Public(Anyone can view this level - this is the
> default.)
> Components: Management Server
> Reporter: Wilder Rodrigues
> Assignee: Wilder Rodrigues
> Labels: MySQL
>
> As an admin user I would like to be able to interrupt the connection between
> the management server and the DBMS and have it attempt to automatically
> recover for a configurable period of time
> =====================
> If for some reason (e.g.DBMS is down, or network connectivity is poor) the
> management server cannot reach the DBMS, then it errors out and stops.
> From an operational stand point, it is sometimes necessary to interrupt the
> connection between the management server and the DBMS. As long as the
> interruption is temporary, the management server should be able to retry and
> re-establish the connection within a given time interval. During this period
> all DB calls will error, but the management system should continue to run.
> We plan to provide a highly available DBMS to cloudstack with a 'floating' ip
> - however host failures or maintenance will still interrupt database
> connections. Ideally cloudstack would reconnect and continue (with the
> floating ip active on another database node).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
