Merge lp:~nuclearbob/utah/minus-one-retry into lp:utah

[Max Brustkern]

This change retries is_utah_done when ssh returns a -1 exit status. In my testing, this solved issues where utah would stop running after attempting to contact a machine that would not accept connections because it was in the process of rebooting. If this change is acceptable, I think it may make working with rebooting machines over ssh more reliable, which will be helpful for bootspeed.

[Max Brustkern (nuclearbob)]

Also, the current default client_install_timeout of 60 seconds seems a little tight. I can just adjust that in the job if needed, but I think a higher default could be helpful to other jobs, so I can add this or make a new merge for it if that's desired.

[Andy Doan (doanac)]

review: Approve

=== modified file 'utah/run.py'
--- utah/run.py	2013-07-11 00:24:25 +0000
+++ utah/run.py	2014-01-08 20:25:34 +0000
@@ -496,6 +496,9 @@
         if exitstatus == ClientReturnCodes.UNKNOWN:
             logging.info('UTAH client is *not* finished yet')
             raise UTAHException('UTAH is not finished', retry=True)
+        elif exitstatus == -1:
+            logging.info('SSH connection failed; retrying')
+            raise UTAHException('SSH connection failed', retry=True)
         else:
             logging.info('UTAH client is finished')
             return exitstatus