[Pacemaker] Notification when a node is down

[Steven Hale]

On 15 September 2014 16:49, David Vossel <dvossel at redhat.com> wrote:
>
> This might be a useful reference.
>
> http://clusterlabs.org/doc/en-US/Pacemaker/1.1-pcs/html-single/Pacemaker_Explained/index.html#idm207039249856

I've been having trouble with this too, and I spent ages on the above
link trying to make it work.

I've got the MailTo resource setup as a clone on all nodes.  I get
emails telling me when a resource moves to or away from a node.  But I
get nothing at all if a resource has failed for some reason and just
ends up in the "stopped" state.

I can't get ClusterMon to work with email notifications.  Looking
through the source I can't see how it can ever work.

Looking at "/usr/lib/ocf/resource.d/pacemaker/ClusterMon" the "start"
method runs a command like this:

${HA_SBIN_DIR}/crm_mon -p $OCF_RESKEY_pidfile -d -i $OCF_RESKEY_update
$OCF_RESKEY_extra_options -h $OCF_RESKEY_htmlfile"

The example in the documentation uses the "extra_options" parameter to
pass through the email settings.

<nvpair id="ClusterMon-instance_attributes-extra_options"
name="extra_options" value="-T pacemaker at example.com -F
pacemaker at node2.example.com -P PACEMAKER -H mail.example.com"/>

But the man page for "crm_mon" doesn't say anything about -T, -F, or
-P options!  crm_mon has no ability at all to send email.  I
eventually gave up and assumed that the documentation for this feature
was either out of date or just plain wrong.

This is with Pacemaker 1.1.10 from CentOS7 default repository.

pacemaker-cli-1.1.10-32.el7_0.x86_64

If the documentation for email notifications (7.2) *is* correct, then
could someone give me a hint as to what I'm doing wrong?

Thanks!