[Linux-cluster] Automated tool for cluster testing?

[Andrew Beekhof]

On Thu, Nov 26, 2009 at 4:30 PM, Rafael Micó Miranda
<rmicmirregs at gmail.com> wrote:
> Hi Andrew,
>
> El jue, 26-11-2009 a las 11:35 +0100, Andrew Beekhof escribió:
>> On Wed, Nov 25, 2009 at 8:41 PM, Rafael Micó Miranda
>> <rmicmirregs at gmail.com> wrote:
>> > Hi all
>> >
>> > Is there any automated tool to test the functionality and availability
>> > of a cluster configuration? Which test would it made?
>> >
>> > I was thinking on this subjects:
>> > - Resource start, monitor and stop operations
>> > - Resource migration
>> > - Communications/network failures
>> > - Node failures
>> > - Service daemons failures
>> > - Fencing device failures
>> > - Qdisk failure
>> > - etc.
>> >
>> > If the answer is "no", which is the battery of test you would apply over
>> > your cluster before considering it "stable and ready" for production
>> > purposes?
>>
>> Pacemaker comes with one that does almost all this, you could probably
>> write a rgmanager module for it.
>>    http://hg.clusterlabs.org/pacemaker/stable-1.0/file/tip/cts
>>
>> Let me know if you'd like more info.
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>
> I used CTS once in one of my old Linux-HA clusters, but writing a
> rgmanager module sounds difficult.

Yep, non-trivial. But possible.

> In the other hand, if you can provide me a simple list of the different
> tests CTS applies it will help a lot just to check if my list of checks
> is as complete as it can.

Look for the class keyword in CTStests.py