-
Bug
-
Resolution: Fixed
-
High
-
5.12.2, 5.12.4
-
None
-
7
-
Severity 3 - Minor
-
55
-
Issue Summary
When a node executing a delete batch operation or any other operation goes OFFLINE, entries within the AO_8542F1_IFJ_PRG_IN_PRG table are left.
This is reproducible on Data Center: (yes)
Steps to Reproduce
- Set up a Jira with 3 nodes
- Create an Assets Schema with a large number of objects
- From the Assets UI, initiate a batch deletion of all objects.
- In Jira database, check the table AO_8542F1_IFJ_PRG_IN_PRG
- There should be en entry similar to the following one tracking the status:
EXECUTED_AS_USER_KEY ID NODE_ID PROGRESS_ID STARTED STEPS_STATUS admin 10 mycluster2 deleteobjects||delete_objects_11ab891b2c-345b-4611-b449-6857c8e5c3fc 2024-02-21 11:44:25.991 {"progressId":{"resourceId":"delete_objects_11ab891be" - While the deletion is in progress, shut down the node mycluster2 as displayed on the above query.
- There should be en entry similar to the following one tracking the status:
Expected Results
If operation is canceled, entries from AO_8542F1_IFJ_PRG_IN_PRG related to the node should be removed as Assets will not attempt to perform the same task.
Actual Results
Operation is canceled and entries from AO_8542F1_IFJ_PRG_IN_PRG are left.
In some cases, a symptom could be observed where a deleteobjects process is stuck in Jira > Manage Apps on one node, while it is not visible on the other nodes.
Note
As a side effect of this bug, in a Jira DC cluster, there might be stale indexing/import tasks on the "Manage apps" page that belong to nodes not present in the current cluster setup (for example, previously removed nodes):
Assets batch process(es) are running. Be aware that upgrading, disabling or uninstalling Assets may cause data inconsistency please wait until the process(es) are completed before performing any of these changes to Assets.
Workaround
Follow the steps from Assets - ongoing indexing / import processes on "Manage apps" page belonging to old cluster nodes article.