Someone enabled 2 probes that were connected to websphere_mq and this added 1.3 million CI's.
This load is preventing the SOI UIM connector from starting.
This added approx. 1.3 million ci's _ci_name) from the websphere_mq application.
The (Systems Operations Insight) SOI connector could not handle that load, and cpu was very high.
The customer disabled the probes and they need to know how to get the ci's out of the uim db so they can get the ci count back down to a reasonable amount, e.g., 500, other than the Mercator1 ci_name-related entires there were 60000 ci_names.
SELECT COUNT(*) FROM CM_CONFIGURATION_ITEM
SELECT COUNT FROM CM_CONFIGURATION_ITEM where ci_name LIKE '%MERCATOR1%'
DELETE TOP (1000) FROM CM_CONFIGURATION_ITEM where ci_name LIKE '%MERCATOR1%'
We progressively increased the query's TOP <value> after trying a smaller value to see how the DB responded and it went well, (fast) so we ended up deleting all of the unwanted data pretty quickly.
As a result, CPU usage was no longer a problem and averaged around 2% even though SOI was busy again but in an expected way - since it was deleting all of the obsolete entries that no longer have a ci_name associated with them and this is expected.
It will probably take 1-2 hours to normalize by deleting the obsolete entries.