607814 - Ganglia grid for Socorro

Status: RESOLVED FIXED

(mozilla.org Graveyard :: Server Operations, task)

Description

[Laura Thomson :laura]

- Set up a basic ganglia grid for Socorro
- Get gmond on all the boxes for basic monitors
- File bugs when you need help from the Socorro team to implement monitors on as many of the criteria at https://wiki.mozilla.org/Socorro:OperationalMetrics as possible.

Comment 1

[Justin Dow [:jabba]]

So it doesn't get lost, need to add this to the socorro-base class in puppet:

class yourclass {
   $ganglia_cluster_name = "Socorro"
   $ganglia_mcast_addr = "239.2.11.7X"
   include ganglia::gmond
}

Comment 2

[Justin Dow [:jabba]]

bkero: This might have to wait a few days for me to get puppet working nicely for socorro, but I'll go ahead and give you the bug, since you'll probably know how to get the cluster up in ganglia.

Comment 3

[Laura Thomson :laura]

I saw
http://im-ganglia01/ganglia/?c=mpt-collectors&m=load_one&r=hour&s=descending&hc=4&mc=2

(Where are the other two collectors?)
What's the status on getting the rest of the boxes monitored?

Comment 4

[Ben Kero [:bkero]]

Laura: I see 6 hosts in mpt-collectors, which should be the correct number.  Is there something that I'm missing about it, or did the issue get resolved?

Jabba: I don't see a socorro class for puppet.  Did you ever get around to this?

Comment 5

[Justin Dow [:jabba]]

It's its own module. puppet/trunk/modules/socorro. Aside from the 6 collectors, we want:

cm-breakpad02
cm-breakpad03
cm-breakpad04
dm-bp-mware01
tm-breakpad01-master01

and the generic cluster.

Probably also stage would be good:
dm-breakpad-stage01
pm-collector-stage1
tm-stage01-master01

Comment 6

[Justin Dow [:jabba]]

This is now done for phx and added to puppet. Once SJC gets repurposed for staging, and set up via puppet, this will also "just work".

The Ganglia Grid for Socorro is available at http://sp-admin01.phx1.mozilla.com/ganglia (available from mpt-vpn).