We upgraded UIM v8.5.1 to UIM v9.0.2 on 26th November and from that point forward, SLA data is not coming in through one of the HUBs. After 26th November 5:50 AM MT we are not seeing any data.
Dashboards became red where this data will be seen by the higher management. Every month end we will pull the data and provide this to the higher management.
- SLO data mismatch
Release : 9.0.2
Component : UNIFIED INFRASTRUCTURE MGMT
If data has changed, you may have to merge the old QOS data with where it is being saved now. Using the SLM portlet ->Database Status via the drop down and check specifically under the QOS_NET_CONNECT qos to see the latest sampledata timestamp ('Last sample' column) - where its being collected and is current. The source/target may have been changed, e.g., uppercase to lowercase for the source, different target, etc. In that case you can merge the data for 'like' QOS in the proper direction, e.g., from the previous source/target TO the current source/target (where the data is currently being collected), otherwise for now you may just choose to edit the QOS objects in the SLOs for the SLA if you're certain how to correct it. Via the MERGE process, when the old data is merged with the new (current data based on date/timestamp), e.g., old unused source/target to the newly changed source/target names, and mapped to the current new table then the SLAs were reporting as expected again.
Notes on merging QOS:
- You can only merge two 'LIKE' QOS objects.
- Normally the blue 'direction' arrow should point to the Last sample - where the data is currently being saved for the correct QoS target.
- You can click the box to choose to Delete the 'source' QoS after the merge operation.
- Effects on PRD/SLA will show the data/historic data being filled in where it was missing or there were previously gaps.
Please make sure the underlying SLOs are pointing to the correct qos, source and target, and then afterwards, rt-click on the SLA and recalculate it for that time period. Then check in about 30 minutes.
Also, make sure there are no unexpected 'Excluded Periods'
When recalculating the SLA, first confirm that the underlying SLOs of the SLA are pointing to the correct QOS, source machine (robot), and target, - you should see current values, and then if so, afterwards, rt-click on the SLA and recalculate it for that time period and note that Create historic data should be selected.
In the sla_engine set the loglevel to 5 and logsize to 200000, then increase the java heap memory to the sla_engine, if you have a number of SLAs/SLOs, it should be at 2048/4096 respectively (as long as you have the physical memory available on the Primary), Then cold start the sla_engine. If you still have a problem open a support case and attach the log after a recalculate showing any/all errors/exceptions/failures.