SDM is not detecting SDC process down. Spectrum console is displaying SDC as green and no alarm is triggered. Such behavior is only observed, when SDC is initiating connection. Steps to recreate fault: 1. Discover SDC host as host type via SNMP 2. Configure SDM to accept connection from SDC ( remoteaccept xxxxxx), other parameters set as default 3. Import SDM configuration 4. Configure SDC to initiate connection (connect xxxxxx), other parameters set as default 5. Restart SDC to read configuration 6. SDC is seen as green 7. Kill SDC process (the SDC host is reachable by SNMP) 8. SDM is not changing the status for SDC (devices behind the SDC goes to red (unavailable))
Expectaton is that SDC will be shown as unavailable.
The same test for SDC configured as accept connection from SDM works perfectly (SDC is marked as unavailable)
Here is the sample of working configuration (SDC down is detected and alarm raised)
1. working configuration: sdc.config: # Place config entries here, see README for options -accept IP_LS_Primary:6844 -accept IP_LS_Secondary:6844
sdm.config: -remoteconnect IP_SDC
and the not working configuration (SDC down is not detected and so no alarm)
2. not working configuration: sdc.config # Place config entries here, see README for options -connect IP_LS_Primary:6844 -connect IP_LS_Secondary:6844
sdm.config: -remoteaccept IP_SDC
Environment
Spectrum 10.3 on Linux platform
Cause
Since the SDC process (initiated the connection) itself is getting shutdown, the endpoint gets deleted . SDM will check for keepalive messages for the entries present in epMap and since the entry is not present for this endpoint, it is not able to reflect the status of SDC host.
Resolution
Fix for this issue will be available in release 10.3.2