x
login Signup

How can I restart one agent of many on a host?

We have an EA agent that has stopped, is now in the Error state and reports that it last sent an A2E_GET_DIRECTORY_ENTRIES call some time back. The original build job was terminated, but this remains. How do we restart this rogue agent without running /etc/init.d/ecagent restart?

avatar image By kev holmes 668 asked Apr 24, 2012 at 11:02 AM
more ▼
(comments are locked)
avatar image eric melski ♦♦ Apr 24, 2012 at 04:30 PM

Can you show us the output of cmtool runAgentCmd 'session state' for this agent?

avatar image kev holmes Apr 24, 2012 at 05:44 PM

Not onsite anymore, will capture and send the next time we encounter it.

10|750 characters needed characters left

1 answer: sort voted first

Generally speaking I would say that using "/etc/init.d/ecagent restart" is the best way to do that. There is a whole bunch of cleanup and setup, including diskcache, mount points etc.. I assume you don't want to abort potentially running agents? My recommendation is to use the standard way, and let emake recover from the restarted agents.

avatar image By sven ♦♦ 3.3k answered Apr 24, 2012 at 04:27 PM
more ▼
(comments are locked)
avatar image eric melski ♦♦ Apr 24, 2012 at 04:30 PM

To be completely safe, you should first disable the affected agents in the Cluster Manager UI, then restart the agents, and finally re-enable the agents in the Cluster Manager UI.

10|750 characters needed characters left
Your answer
toggle preview:

Up to 8 attachments (including images) can be used with a maximum of 524.3 kB each and 1.0 MB total.