[StarCluster] invalid instance error

[Igor Tatarinov]

I've seen a few of these intermittent errors right at start up:

PID: 29470 cli.py:158 - ERROR - InvalidInstanceID.NotFound: The instance ID
'i-2eaea843' does not exist

I think EC2 needs a few seconds after starting a new instance before it can
be queried about the new instance. Adding an initial delay seems to fix this
problem:

@@ -1168,6 +1168,7 @@
         s = Spinner()
         log.log(INFO_NO_NEWLINE, "Waiting for cluster to come up...")
         s.start()
+        time.sleep(30)
         while not self.is_cluster_up(enforce_size):
             time.sleep(30)
         s.stop()

30 seconds is probably overkill but I doubt a new instance can be spun up in
less than that.

igor
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/pipermail/starcluster/attachments/20101219/59f6a470/attachment.htm