[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-ha
Subject:    Re: [Linux-HA] Re: Re: Re: Re: Re: Heartbeat and RA monitor functions
From:       "Andrew Beekhof" <beekhof () gmail ! com>
Date:       2008-05-31 6:22:45
Message-ID: 26ef5e70805302322h5baa1c03qe1265bc129bbb075 () mail ! gmail ! com
[Download RAW message or body]

On Sat, May 31, 2008 at 1:46 AM, Joe Bill <pica1dilly@yahoo.com> wrote:
>> we run "probes" on new cluster nodes...
>> to make sure no service can ever be running twice
>
>>> In other words, before starting a resource on a
>>> node, for the first time after it's reboot, can
>>> we assume HB always calls a "MONITOR" or a
>>> "STOP" to certify no sub-resources used by that
>>> resource are already allocated ?
>
>> sub-resources?
>
> Other system resources (not declared as CRM resources)

By definition we can't check them because we know nothing about them.

If your RA (which does know about them) does the checks while checking
the "main" resource - then thats a different story but completely
opaque to us.

> on which a CRM resource depends upon to start, and
> that are verified at execution of the RA's MONITOR for
> a "running healthy" status to be returned.
>
>>> BTW is it a "MONITOR" or a "STOP" that is
>>> first called by HB for a resource ?
>
> So if I provide a MONITOR function in the RA, I assume
> it will be a MONITOR that will be performed as
> "probe".

yes.  in fact, you _must_ support the "monitor" operation.

>
> What if I provide several MONITOR functions, of
> different check levels, which check level will be
> performed as "probe" ?

the one actually called "monitor"
i think we also set a default value for the "check level", but its up
to you to decide if you want to make use of it
_______________________________________________
Linux-HA mailing list
Linux-HA@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
[prev in list] [next in list] [prev in thread] [next in thread]