Opened 10 years ago

Closed 10 years ago

#39 closed (fixed)

every sliver creation at utah.geniracks.net using a pgeni.gpolab.bbn.com slice generates over 200 e-mails to gpolab testbed-ops

Reported by: chaos@bbn.com Owned by: somebody
Priority: major Milestone:
Component: Experiment Version: SPIRAL4
Keywords: Cc:
Dependencies:

Description

I hope this is an easy fix, because it is making our mailserver slightly unhappy...

After yesterday's upgrade, Luisa created 10 slivers at utah.geniracks.net. As a result, we received over 1600 e-mails which looked like this:

From: Unprivileged user <nobody@pgeni.gpolab.bbn.com>
Date: Wed, 30 May 2012 16:25:29 -0400 (EDT)
To: testbed-ops@ops.pgeni.gpolab.bbn.com
Cc: protogeni-errors@flux.utah.edu
Reply-To: testbed-ops@ops.pgeni.gpolab.bbn.com
Subject: [gpo-ops] PGENI.GPOLAB.BBN.COM: protogeni-wrapper.pl

protogeni-wrapper.pl
Invoked by nobody (Unprivileged user)
URN:     urn:publicid:IDN+utah.geniracks.net+authority+cm
Target:  urn:publicid:IDN+pgeni.gpolab.bbn.com+slice+10vmslice1                 
Module:  sa
Method:  RegisterSliver
Start:   16:25:29:579861
End:     16:25:29:662445
Elapsed: 0.082565
Error:   2
Output:  Could not create credential object
Result:
$VAR1 = 0;

Over 200 of these e-mails contain the name of each sliver, and they appear to be sent approximately once per sliver per minute for about 3 hours.

I replicated this just now with my own ecgtest slice, and we have started receiving these e-mails for my slice too.

Change History (6)

comment:1 Changed 10 years ago by chaos@bbn.com

Just to get my wild guess on the record: is it possible that some code was added to the ProtoGENI AM to support registering slivers with a slice authority, e.g. for the benefit of the clearinghouse, and that pgeni.gpolab.bbn.com, which is running stable-20120409, does not yet support the server side of this transaction?

If so, i'm going to go ahead and recommend that the AM side give up after less than 3 hours of failure. :>)

comment:2 Changed 10 years ago by chaos@bbn.com

Also, i deleted my sliver at 09:12 EDT, and that appears to have stopped the e-mails, so that's something.

comment:3 Changed 10 years ago by lnevers@bbn.com

Just a note, I was only able to set up 8 slivers successfully, not the 10 I had intended to run.

comment:4 Changed 10 years ago by chaos@bbn.com

Ah, yeah, you're right: i didn't check my own math. We received no e-mails for 10vmslice9 or 10vmslice10, which i assume are the ones that failed.

comment:5 Changed 10 years ago by lnevers@bbn.com

Yup! The 9th and 10th slice failed. The slices are named (and created) in numerical order, so that I can keep them straight.

comment:6 Changed 10 years ago by chaos@bbn.com

Resolution: fixed
Status: newclosed

I got my wild guess on the record so it could be immediately disproven by something totally obvious: boss.utah.geniracks.net had a clock which was several hours off, so various SSL discussions were failing.

This is now fixed. Thanks!

Note: See TracTickets for help on using tickets.