Bug 232300 - rgmanager should discover migration if someone migrates a vm out from under it
rgmanager should discover migration if someone migrates a vm out from under it
Product: Red Hat Enterprise Linux 5
Classification: Red Hat
Component: rgmanager (Show other bugs)
All Linux
medium Severity medium
: ---
: ---
Assigned To: Lon Hohberger
Cluster QE
Depends On:
  Show dependency treegraph
Reported: 2007-03-14 15:30 EDT by Lon Hohberger
Modified: 2009-04-16 18:37 EDT (History)
1 user (show)

See Also:
Fixed In Version: RHBA-2007-0580
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
Last Closed: 2007-11-07 11:45:50 EST
Type: ---
Regression: ---
Mount Type: ---
Documentation: ---
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---

Attachments (Terms of Use)

  None (edit)
Description Lon Hohberger 2007-03-14 15:30:57 EDT
Description of problem:

xm migrate ... 

...will cause rgmanager to think the VM was stopped on the node it was running
on and restart it.  If migration succeeded and the VM is restarted, this ends up
with two copies of the VM running on the cluster.

The thought was "don't do that", but we can protect against this in the
following manner:

For a service which is "migratory" - i.e. a VM with no children, if status check
fails, we can then:
* ask each node to check for the "migratory" service.
* if it turns up, we can simply update rgmanager's state to reflect the new owner.
* if it does not turn up, take recovery as normal
Comment 1 Kiersten (Kerri) Anderson 2007-04-23 13:23:29 EDT
Fixing Product Name.  Cluster Suite was integrated into the Enterprise Linux for
version 5.0.
Comment 3 Lon Hohberger 2007-06-29 13:52:38 EDT
Warning: testing of this is dangerous.

On 5.0, if you migrate a virtual machine off of a host node while rgmanager is
watching the VM, rgmanager will see this as a failure and restart the VM -
potentially ending up with the VM running in two places on the cluster.

On 5.1, if you migrate a VM off of a host node, rgmanager will first query other
nodes in the cluster - and then restart it.

Additionally, if you create a VM on a cluster node running rgmanager - and
rgmanager has an associated service, a subsequent 'enable' request will not
start the VM a second time - it will simply flag the VM as 'started' on the node
which is already running it.
Comment 5 errata-xmlrpc 2007-11-07 11:45:50 EST
An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on the solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.


Note You need to log in before you can comment on or make changes to this bug.