Problem
Before MDM 2.2 upgrade, when one use to update or add new admin data for example tariff details, manufacturer details, etc. we had to perform “flushAll” on each managed server URL including load balancing URL.
Post MDM 2.2 upgrade, the facility to perform flushAll on each managed server has been revoked and an introduction of Global flushAll has been introduced. There is an intermittent issue where post Global flushAll some managed servers are unable to read the new or updated admin data.
The similar situation is faced during bundle deployment wherein the updated or new code is not reflected on each of the managed servers
Solution
The solution is yet to be identified but the workaround for the above problem is below:
- Rolling Restart
- Perform a rolling restart of Thread Pool Worker (TPW). Post restart check all your batch schedule are running fine if any error performs rerun action accordingly
- Perform a rolling restart of all the MDM managed servers
- Complete Restart
- Hold all the batch schedule, perform a full restart of Thread Pool Worker (TPW) and release the batch schedule
- Perform a restart of all the MDM managed servers at once
Food For Thought: Option B will cause downtime to your live system, hence the recommended option is Option A but during planned outages Option B is quicker. Decide wisely!
SOA Solutions 