Task stalls in state RUNNING in case of service dies
Bug #1502120 reported by
Roman Dobosz
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Mistral |
Fix Released
|
Undecided
|
Nikolay Makhotkin | ||
Mitaka |
Fix Released
|
High
|
Nikolay Makhotkin |
Bug Description
I've tested Mistral (two Mistral services on two controllers) on devstack, to check what happens if executor dies during executing an action, just to check if there is an HA for tasks, and discovered that the task is stuck in RUNNING state.
Steps to reproduce:
1. Create a workflow, that will run for some time (in my case I've created custom action that run for 20 seconds)
2. Execute the workflow and check which service is executing it
3. Kill that service
4. Issue "mistral execution-list" command and observe that task is in RUNNING state. Forever.
Changed in mistral: | |
status: | New → In Progress |
To post a comment you must log in.
Roman,
Thanks for letting us know. We've been aware of this issue since the very beginning when we started using oslo.messaging. Since oslo doesn't support post-processing message acknowledgement yet we have this gap. And that's the reason why have a number of patches with "RPC" in their commit messages on review. Currently we're working with oslo team to fix this problem. Here's a WIP that addresses that problem: https:/ /review. openstack. org/#/c/ 229186/