the ganeti init script tries to start masterd and rapi on any node, not just on the master node. Reproducible: Always Steps to Reproduce: 1. /etc/init.d/ganeti start 2. /etc/init.d/ganeti stop Actual Results: node2 ~ # /etc/init.d/ganeti start * Starting ganeti-noded ... [ ok ] * Starting ganeti-masterd ... [ ok ] * Starting ganeti-confd ... [ ok ] * Starting ganeti-rapi ... [ ok ] node2 ~ # /etc/init.d/ganeti stop * Stopping ganeti-rapi ... * exit code 1 [ !! ] * Stopping ganeti-confd ... [ ok ] * Stopping ganeti-masterd ... * exit code 1 [ !! ] * Stopping ganeti-noded ... [ ok ] node2 ~ # gnt-cluster getmaster node1.domain.intern Expected Results: node2 ~ # /etc/init.d/ganeti start * Starting ganeti-noded ... [ ok ] * Starting ganeti-confd ... [ ok ] node2 ~ # /etc/init.d/ganeti stop * Stopping ganeti-noded ... [ ok ] * Stopping ganeti-confd ... [ ok ] node2 ~ # gnt-cluster getmaster node1.domain.intern attached is a patch that should fix the problem.
Created attachment 278693 [details] patch to fix service startup on non-master nodes check node status via gnt-cluster getmaster and only start masterd and rapi on if executed on master node.
(In reply to comment #1) > Created attachment 278693 [details] > patch to fix service startup on non-master nodes > > check node status via gnt-cluster getmaster and only start masterd and rapi > on if executed on master node. Thanks for the patch! I took a slightly different approach and fixed it. It's push to portage and sorry it took me so damn long to fix this.
