Description of problem: force-stop tried from user and and under broker using rhc-admin-ctrl-app, and app failed to stop. mcollective.log Output: I, [2012-07-31T10:47:09.972609 #30896] INFO -- : stickshift.rb:271:in `cartridge_do_action' cartridge_do_action call / request = #<MCollective::RPC::Request:0x7f37f82a14f8 @action="cartridge_do", @agent="stickshift", @caller="uid=0", @data= {:cartridge=>"jbossas-7", :process_results=>true, :action=>"force-stop", :args=>"'showcase' 'richfaces' '9ef9c83478954bed9add90b160c4b5a2'"}, @sender="mcollect.cloud.redhat.com", @time=1343746029, @uniqid="b45dd9dd094ad2cb60a746cb9f2b12a4"> I, [2012-07-31T10:47:09.973140 #30896] INFO -- : stickshift.rb:272:in `cartridge_do_action' cartridge_do_action validation = jbossas-7 force-stop 'showcase' 'richfaces' '9ef9c83478954bed9add90b160c4b5a2' I, [2012-07-31T10:47:25.220335 #30896] INFO -- : stickshift.rb:329:in `cartridge_do_action' cartridge_do_action ERROR (254) ------ 9ef9c83478954bed9add90b160c4b5a2 9ef9c83478954bed9add90b160c4b5a2 sh: fork: retry: Resource temporarily unavailable sh: fork: retry: Resource temporarily unavailable sh: fork: retry: Resource temporarily unavailable sh: fork: retry: Resource temporarily unavailable sh: fork: Resource temporarily unavailable
trying to stop on the app node using: service libra stopuser UUID, also failed with: Stopping 9ef9c83478954bed9add90b160c4b5a2... showcase_ctl.sh bash: fork: retry: Resource temporarily unavailable bash: fork: retry: Resource temporarily unavailable bash: fork: retry: Resource temporarily unavailable bash: fork: retry: Resource temporarily unavailable bash: fork: Resource temporarily unavailable
Might be worth double checking here to see if the force-stop is running as root or not. This is one of the very few hooks that _should_ run the kill as root (so make sure it's not getting any flags passed to it)
Rob expects to look at today
Bug 838365 addressed this issue in development and the fix should be in production shortly. The force-stop function had a component that ran as the gear UID. This would fail if the gear had maxed out its memory or allowed number of processes. The component was re-written to run entirely as root. *** This bug has been marked as a duplicate of bug 838365 ***