Recently we have encountered a phenomenon wehere values where reset while upgrading different integration policies from one version to another. 
I'm not sure if this is intended but it is also not consistent when leaving the upgrading menu and entering again. Sometimes, the existing values exist and sometimes they are reset to default values (localhost, etc.). Another problem I've encountered that might or might not be related to this is that sometimes the Filters of Dashboards vanish. Restarting the Browser and clearing cache helps. I'm interested if someone had encountered the same issues as described.
From my point of view it is problematic because:
-
Inconsistency makes it hard to trust the integration while upgrading them. Currently, I always have to compare the values from the existing integration with the values of the integration in the upgrade menu to see if things like the listen address, listen port, dataset name, ingest pipeline, etc. are still correct.
-
Many integrations lead to an excessive amount of time required to upgrade all the integrations and double-check if the values are still correct.
Integrations where I had the issue:
- Custom TCP Logs
- Custom UDP Logs
- Windows
- Custom Windows Event Logs
- Palo Alto Networks Logs
OS: macOS 12.5
Browser: Edge 103.0.1264.62
Kibana: 8.3.2
Elasticsearch: 8.3.2
Fleet-Agents: 8.3.2
OS Elastic Server: Debian 11.3
Happy to provide you further information if needed.