Description of problem: ----------------------- While moving the Virt cum Gluster ( hyperconverged ) node to maintenance, by selecting an option to stop gluster process, all the gluster processes are killed abruptly by invoking the script 'stop-all-gluster-process.sh'. With RHGS 3.2.0, 'stop-all-gluster-process.sh' provides '-g' switch to gracefully kill all gluster process, as abruptly killing process may lead to inconsistent scenarios. So, while moving the gluster node to maintenance, RHV should make use of this script to stop all gluster processes Version-Release number of selected component (if applicable): ------------------------------------------------------------- NA How reproducible: ----------------- NA Steps to Reproduce: ------------------- NA Actual results: --------------- Moving the node to maintenance stops all gluster processes abruptly Expected results: ----------------- Exploit the script 'stop-all-gluster-process.sh' and make use of '-g' option to gracefully kill all gluster processes Additional info: ---------------- http://review.gluster.org/#/c/15188/
Calling the script with -g flag means that the script will keep trying till the processes are killed and does not return immediately. We need to keep polling for status of process before host maintenance can be completed - this complicates the flow, deferring for now
This bug has not been marked as blocker for oVirt 4.3.0. Since we are releasing it tomorrow, January 29th, this bug has been re-targeted to 4.3.1.
Right now there is no framework for handling a back event from the vdsm to the engine side in order to ensure that gluster processes have indeed stopped.