UCMDB 10.22 CUP3.334 CP18 HA - Results not sending to writer after seven days
I have been looking into a persistent issue with an instance of UMCDB where the discovery and integration works okay for several days, before the unsent results queue on the discovery probes. Performing a failover to the other server node (either through the JMX or by stopping services on the writer), resolves the immediate issue, but it will reoccur in about a week.
Here is a high-level description of our configuration
UCMDB Server Version: 10.22 CUP3.334
Content Pack: 18.00.74
UCMDB server cluster: 2 (one writer/one reader)
No. of UCMDB Probes: 21 (All are version 10.22 CUP3.334)
Performing integration from CSV of CIs (via discovery job)
Population and Push Integration: Asset Manager Population job, Data Push jobs for Service Manager
Discovery of server hardware through Powershell
Discovery of Desktops and Laptops via UD Agent.
Happy to provide further information. I am currently trying to documentation/release notes of which release (Version, CUP or CP) resolves this issue if it was a known one
You have to see if the issue is based on performance and slow processing of the results, or maybe it is caused by some stuck job, which doesn't let the next results to be sent.
See the DFP Gateway log on the probe with the queue to see how ofter it sent results to the UCMDB server. also check the UCMDB cmdb.reconciliation.audit log to see how fast it processes the results.
I would recommend checking a setup of Load Balancer, there is a chance that switch of a direction for a writer not performed well. Check what URL is used to determine which server is a writer. Could be a key...
-Dmitry Gomel, PMP
Click the Like button at the bottom to say 'Thanks'.