Many CallHomeEvents and bad performance of discovery

Dear community,

I was reading this discussion : 

/it_ops_mgt/ucmdb/f/sws-cms_sup/353579/ucmdb-support-tip---callhome-inventory-by-scanner-job-scheduling-slow-down-the-discovery-cycle

And i want your advices.

I'm using callhome jobs and Discovery jobs trought MZ. 

Discovery job are scheduled and cal home job are scheduled too.

I think it's the cause of the bad discovery performance.

Can you give me some tips ?

Thanks.

  • Verified Answer

    Call home processing is a very processor intensive job. There are many writes and deletes involved with this job. I would optimize your use of Call Home so that it is only used for the most dynamic hosts in your inventory collection.

    Do you have your IP Ranges configured by Datacenter and Client ranges?
    If you have Client ranges defined then, stop Datacenter (IP Lease Time = Permanent or long) Call Home:
    globalSettings.xml

    <!--Switch for if allowing data center IP to call home--> <property name="allowDataCenterCallhome">false</property>

    How frequent is your Inventory Discovery by Scanner interval?
    If your discovery is every 14 days then, set call home interval to 336 hours.
    If your discovery is every 7 days then, set call home interval to 168 hours.
    globalSettings.xml

    <!--Time interval in hours that is allowed between two call home requests from the same host --> <property name="allowCallhomeInterval">168</property>

    Do you run the IP MAC harvesting job (SNMP or ARP cache)?
    These jobs also create Call Home event so set the interval to match your Inventory Discovery schedule.
    For instance, Inventory Discovery begins on Monday and interval is every 7 days. Then schedule IP MAC for once a week on Wednesday or Thursday. The reason is that Agent Call home will be close to Monday start time and will update IP Address. The IP MAC job should then be several days later to update IP address of still parked inventory jobs before the job invocation time is reset.

    If you do not use any UD Agent based jobs then, turn off Call Home.
    UD Agent based jobs are:

    • Install UD Agent
    • Uninstall UD Agent
    • Update UD Agent
    • Inventory Discovery by Scanner

    globalSettings.xml

    <!--Switch for if allowing to call home--> <property name="allowCallhome">false</property>

    Call Home processing is only used for unparking the scheduled UD Agent jobs. If you do not use any of these jobs then you have no need to process call home requests. You can then stop all the processing overhead on the UCMDB server by turning off call home.

  • Hello Florian,

    I presume that the main scope of your inquiry is the general performance degradation induced by the heavy CallHome tasks.

    This was noticed in the past also. To add to the information provided by , some of the performance degradation happens on the History updates on the CallHome history tables.

    A good option is to disable the entire CallHome CIT History tracking so whatever updates we have on these CIs we won't have any history statements behind them so the DB and overall data-in will perform much better. This is done with a special qualifier on the CIT level.

    When the CIT_NOT_TRACKED_FOR_HISTORY qualifier is assigned to an object CIT, no CI events (add, update, delete) will be stored to history for all instances of the selected CIT as well as all incoming and outgoing links related.

    More details on the documentation page at https://docs.microfocus.com/itom/Universal_CMDB:2020.02/SAHstryDisable