Fail evacuate flow with deleted VM
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
masakari |
New
|
Undecided
|
Unassigned |
Bug Description
Today my lab environment got exception like this: "Resource could not be found". My case is very rare where the compute server unfortunately rebooted as soon as it received the request to delete the VM but could not do it, specifically look up the following behavior, note that it has concurrency:
1) Server 'controller' received request to delete VM (normal or amphora).
2) The 'compute' server is off, the above request is still unprocessed just stop at the waiting queue. The 'Masakari Engine' on the 'controller' server has listed the VMs located on the server that just crashed.
3) The 'compute' server is back up and running the request to delete the VM.
4) The 'Maskari Engine' on the 'controller' server continues execution to the step in the source code called 'Task Evacuate'. Right at the line of code that gets the VM information through the Nova SDK, there is an error right before the 'spawning evacuate' line for that VM, of course the error will be described that the resource cannot be found.
Note: I'm based on Yoga branch.
https:/
-------
for instance_id in instance_list:
msg = "Evacuation of instance started: '%s'" % instance_id
-> instance = self.novaclient
-------
...
2023-06-11 05:02:04.326 7 ERROR masakari.
2023-06-11 05:02:04.326 7 ERROR masakari.
2023-06-11 05:02:04.326 7 ERROR masakari.
2023-06-11 05:02:04.326 7 ERROR masakari.
2023-06-11 05:02:04.326 7 ERROR masakari.
2023-06-11 05:02:04.326 7 ERROR masakari.
2023-06-11 05:02:04.326 7 ERROR masakari.
2023-06-11 05:02:04.326 7 ERROR masakari.
2023-06-11 05:02:04.326 7 ERROR masakari.
2023-06-11 05:02:04.326 7 ERROR masakari.
2023-06-11 05:02:04.326 7 ERROR masakari.
...
description: | updated |
description: | updated |
description: | updated |
description: | updated |