Question:
Why does it occure that when a SpectroServer has a very high volume of Alarms, aka Alarm Storm, over a short period of time that it can happen that Alarm Clear events are not received in SOI ?
Alarms that are cleared in Spectrum already are still in the SOI Console.
Answer:
Looking into the SpectrumIM Connector logfile for the Model_Handle fo the Spectrum Model that had the Alarm, one can see the “Alarm { NEW” message but no “Alarm { Delete ” message.
This situation can happen when the SpectroServer runs into a problem to update its Corba Clients.
Following configuration for the Spectrum Corba Client Notifications Cache can be applied. The values of the parameters can be adjusted as needed.
In $SPECROOT/SS/.vnmrc are the following default settings for the CORBA-layer (orb_/request timeout und queue settings:
orb_timeout=60
itc_request_timeout=60000
max_pending_corba_client_notification=10000
In case of a massive Alarm Workload this settings can be increase as the following suggestion:
orb_timeout=90
itc_request_timeout=90000,
max_pending_corba_client_notification=50000
the "max_pending_corba_client_notification" queue would be 5 time bigger and the timeout increased to 90 seconds and the Alarm new, update, delete will be aged (removed from the corba client cache) after 90000ms/90sec.