In 3.1 there is no way the administrator can cancel task. The idea is to have a right click (or button) on the tasks tab of "cancel task" Open questions: - should it apply only to VDSM tasks ? - should we enable cancellation of task group ? - are there any tasks that we specifically don't want to enable to cancel?
(In reply to comment #0) > In 3.1 there is no way the administrator can cancel task. > The idea is to have a right click (or button) on the tasks tab of "cancel > task" > > Open questions: > - should it apply only to VDSM tasks ? VDSM is doing the heavy lifting anyway, are there any non-vdsm tasks that can be relevant here? With external tasks coming in, we may need to see how we incorporate those into this picture though > - should we enable cancellation of task group ? afaik we have a job that consists of a set of tasks, we should be able to cancel/rollback an entire job > - are there any tasks that we specifically don't want to enable to cancel? I'd like to see an overall list of tasks, before we start choosing how we treat each of them.
Hi guys, This feature seems to have slept for quite some significant amount of time and to be honest makes even more sense with the improvement of product in Storage area. Support wise we face quite often situations where rollback/foward commit dies and the task is still running in hypervisor (reported as unfinished in manager as well). Wouldn't it make sense to actually implement manual triggering of canceling task with unsuccessful reason 'manual interrupt'. Example - LSM is interrupted by reboot of SPM hypervisor and manager at the same time - none of the two can report finished/unfinished task, task hangs (very rare scenario I know). We have to go, delete task from task repository, restart vdsm and possibly even manually edit DB and unlock locked resources. Also for long-lasting storage tasks it could be beneficial.
*** Bug 748408 has been marked as a duplicate of this bug. ***
I just experienced this limitation in a pretty ugly way. Being a RHV noob, my expectation when I created a VM from a template was either that it would default to a thin disk copy or would make the selection obvious if not mandatory. Instead, the default is clone, and the option to change it is buried in the advanced options dialogs. The result was inadvertently creating a VM that would need to copy 50+ GB of data on the Gluster backend. With no apparent ability to cancel this process, I'm just stuck working on my thumb-twiddling skills.
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.
4.3.1 has been released, please re-target this bug as soon as possible.
There is no infrastructure which would link commands/jobs/tasks running on RHV Manager with tasks running in VDSM, meaning that each RHV Manager command can run multiple tasks, but RHV Manager is able only to monitor them and detect they finished with success or failure, there is no common infrastructure to cancel existing task. The main reason that VDSM tasks are usually running low level operations which in many cases cannot even be cancelled, so possibility to cancel task needs to be implemented for each such task. Based on above it doesn't make sense to implement common infrastructure to cancel running tasks, because ability to cancel the task needs to be provided by each concrete task and there might be different ways how to cancel such task. So please file specific RFE to add ability to cancel specific task. For example we have already implemented specialized cancel option for live migration flow. So suggesting closing this RFE and WONTFIX. Moran?
Peter, Should we close this RFE?
(In reply to Martin Perina from comment #24) > There is no infrastructure which would link commands/jobs/tasks running on > RHV Manager with tasks running in VDSM, meaning that each RHV Manager > command can run multiple tasks, but RHV Manager is able only to monitor them > and detect they finished with success or failure, there is no common > infrastructure to cancel existing task. The main reason that VDSM tasks are > usually running low level operations which in many cases cannot even be > cancelled, so possibility to cancel task needs to be implemented for each > such task. > > Based on above it doesn't make sense to implement common infrastructure to > cancel running tasks, because ability to cancel the task needs to be > provided by each concrete task and there might be different ways how to > cancel such task. So please file specific RFE to add ability to cancel > specific task. For example we have already implemented specialized cancel > option for live migration flow. > > So suggesting closing this RFE and WONTFIX. Moran? Peter/Martin?
As mentioned above it's not possible to provide a common infrastructure to cancel a task, but each flow, which should support cancellation, need to implement specific cancel action. As mentioned above Live Storage Migration has been identified as a flow where cancellation is needed, so moving this command bug into a specific flow bug, so moving to storage team for further evaluation. Please create specific bugs for each flow where cancellation is needed.
This request is not currently committed to 4.4.z, moving it to 4.5