629694 - modify runslave.py to call back to nagios

Status: RESOLVED FIXED

(Release Engineering :: General, defect, P4)

Description

[Dustin J. Mitchell [:dustin] (he/him)]

Per bug 627126, runslave.py is the single common point for all (desktop) slaves, and we want it to report its status to nagios, as a passive check.

This should go into runslave.py *before* the rollout to the various win* systems, since that's a manual process.

Comment 1

[Dustin J. Mitchell [:dustin] (he/him)]

Attachment: m629694-puppet-manifests-r1.patch

I'm putting this in staging now.  Let's see if it hangs any of the slaves' startup - it shouldn't!

I considered complicating this so that it times out the connect, read, and write operations -- but that's hard to do portably on Windows.  If nagios is *down*, this will fail silently and move on.  If nagios is hanging, this will hang, too - but then we've got bigger problems.

Comment 2

[Dustin J. Mitchell [:dustin] (he/him)]

linux-ix-slave05 has rebooted a bunch of times with this patch applied, with no ill effects.  I'll leave it in staging over the weekend, but this looks promising, so plz 2 review.

Comment 3

[Mike Taylor [:bear]]

Comment on attachment 529218 [details] [diff] [review]
m629694-puppet-manifests-r1.patch

One thing that I am wondering is the buf+= sk.recv() loop in do_send_notice() routine is how you would prevent an endless loop if the remote end of the socket is closed but your local socket remains open.  You would receive 0 bytes until something else happened.

Other than that (and given how much of a low-level single-purpose tool) r+

Comment 4

[Dustin J. Mitchell [:dustin] (he/him)]

Attachment: m629694-puppet-manifests-r2.patch

updated version on bear's advice.  Checked in and will be deployed shortly along with the patch for bug 648665.

Comment 5

[Dustin J. Mitchell [:dustin] (he/him)]

deployed.