My customer has multiple sites with redundant scale out architecture
(Frontend server pair, Backend server pair, all tags live on Backend, with frontend remote providers).
We just upgraded 2 sites from 8.3.7 to 8.3.10.
Here is the upgrade procedure:
- Frontend force failover
- Upgrade Frontend Primary (which takes back control)
- Upgrade Frontend Secondary
- Backend force failover
- Upgrade Backend Primary (which takes back control)
- Upgrade Backend Secondary
It seems like some configuration tags and UDTs reverted to their default values.
- config tags (Memory) that were recently FALSE were now set to default-TRUE
- UDT strings (Memory) with parameter info were set to default-BLANK
On our next upgrade, we will try to see if the incident occurs specifically on a forced failover or the actual version upgrade.
Has anyone else seen this?
Any ideas on how to prevent this for future upgrades?
Thanks