Scale down is not being triggered even if the extractors are idle for more than one hour and the list of extractor instance scale-down candidates is empty.
I observed that this may be due to some unacked msg in the queue from the file upload from Humanpref extractor. I deleted the HumanPreference Index queue and now the scale down procedure is working and able to suspend the VMs. However, I see lots of Error in the log file about ssh error to the vlad-mgmt machine.
Example :2015-06-22 14:25:04,126 INFO : about to run command: "['ssh', 'vlad-mgmt.ncsa.illinois.edu', 'ssh', '-l', 'ubuntu', u'141.142.204.251', '-O', 'exit']"
No ControlPath specified for "-O" command
2015-06-22 14:25:04,263 ERROR : Running command: "['ssh', 'vlad-mgmt.ncsa.illinois.edu', 'ssh', '-l', 'ubuntu', u'141.142.204.251', '-O', 'exit']" returned error. Error: , error code: 255