Sorry for top-posting. Afflicted with Outlook.

Here is the STONITH agent. I've tested it on both old and new and it
seems fine. 

Regards
Darren

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Dejan
Muhamedagic
Sent: 08 September 2009 18:27
To: General Linux-HA mailing list
Subject: Re: [Linux-HA] ibmrsa-telnet STONITH Not Working

Hi Darren,

On Tue, Sep 08, 2009 at 04:52:31PM +0100, [email protected]
wrote:
> > 
> > Hi Darren,
> 
> Hi Dejan.
> 
> 
> > 
> > On Fri, Sep 04, 2009 at 04:48:30PM +0100,
[email protected]
> > wrote:
> > > The ibmrsa plugin requires mpcli to be installed and the ibmrsa-
> > telnet
> > > plugin doesn't work with newer RSA boards. It seems the hostlist
> > error
> > > was a dead herring leading me up the garden path.
> > >
> > > I've altered the ibmrsa-telnet plugin a little to get it working
> with
> > > the new format of the CLI (username : changed to login: > changed
to
> > > system> etc). Not sure of how to officially submit patches or
> > anything
> > > so if it's useful I've attached my changed version.
> > 
> > Darn, why did they have to change the text, as if "login" were so
> > much better than "username".
> > 
> > The only reasonable way would be to check the RSA release and
> > then to wait for the correct string. But first we have to figure
> > out when the change took place, that may be close to impossible.
> > I doubt that such information is in the release notes ;->
> > 
> > Looking at the Telnet documentation, there is another method
> > which waits for a set of regular expressions:
> > 
> >     Telnet.expect(list...)
> > 
> > Perhaps we could modify the plugin to use that instead of
> > read_until(). I don't have an RSA board available though. Could
> > you try this out?
> 
> I could certainly try this, right about now is the perfect time to do
it
> too. 

Great!

> > 
> > BTW, you also increased timeouts. Was that while you were testing
> > or is it really needed? At any rate, it's better to go with
> > longer timeouts.
> 
> The prompts take quite a long time to appear, certainly longer than
the
> original timeouts. I think 20s should be long enough to allow for the
> prompt but to time out in a reasonable time if something is wrong.
I'll
> have a look and see what I can do. Alternatively perhaps there can be
a
> ibmrsa-telnet1 and ibmrsa-telnet2 to allow for both revisions?

I'd like to avoid that if at all possible, it will be source of
endless confusion and a maintenance problem. As far as timeouts
go, it's OK if we adopt the longer ones. If a fence action
times out, that happens only once and ends in error. It may take
a bit longer for that to happen, but it should be fairly seldom
on a well administered cluster.

Cheers,

Dejan

> 
> > 
> > Cheers,
> > 
> > Dejan
> > 
> 
> Darren
> _______________________________________________
> Linux-HA mailing list
> [email protected]
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Attachment: ibmrsa-telnet
Description: ibmrsa-telnet

_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to