Details
-
Bug Report
-
Resolution: Done
-
L3 - Default
-
None
-
None
-
M
Description
Brief summary of the bug. What is it ? Where is it ?
In cases where a cloud user is migrating from a Cluster without Optimize to a cluster with Optimize (e.g. from 1.1.x to 1.2.x) old definition records may actually not exist anymore so Optimize can't import these definitions.
There might still be process instance records though referring to these definitions and the import would get stuck on them indefinitely retrying to fetch the definition.
Steps to reproduce:
- Start a Zeebe Environment
- Deploy a definition and start and instance
- delete the exportet definition entry from the zeeber exporter index
- Start Optimize
Actual result :
Observe the error logs:
org.camunda.optimize.service.exceptions.OptimizeRuntimeException: The process definition with id2251799813685353 has not yet been imported to Optimize
Thrown repeatedly as the definition will never be present in Optimize
Expected result:
These situations can always occur if Optimize hits an incomplete export history (may it be due to the fact that is was not present in that environment before or because the import was blocked due to other issues for some time)
I see two options here:
1. skip those entries after the failure reoccurred a configurable amount of times
2. still import the entities (but we can't as the key/id that is getting resolved is part of the target index name, this would require removing the unavailable key from the instance index name)
3. make the zeebe exporter add the process definition key (id in their terms)
mgm-controller-panel
This is the controller panel for Smart Panels app
Attachments
1.
|
Remove dependency on imported definitions for incident records |
|
Done | Unassigned |
2.
|
Remove dependency on imported definitions for variable records |
|
Done | Unassigned |