As discussed on the list we'd like to increase the swap on the boxes to 1.5G / proc and change the resolution of the notifications so we don't get too many false positives. I'm suggesting:
10 failures. checking every 5 minutes on failure. This means swap will have to be 15% or less free for at least 50 straight minutes before we get notified.
Ok, the nagios side of this is done (I think) we'll see how it goes before I call it done done.
This is all set, both Swap on the builders and nagios have been configured as described.
to comment on this ticket.