SSM restart by SD

Use case: we should ask SD try to restart SSM should it detect non-responsive SSM after a certain time of period (or after tried a number of times). Not only after detecting SSM's PID is gone or the underlying network connection is broken.

The length of the waiting time or the number of times it should try (and the waiting time between the tries) should be tunables which user can customize according to own situation.

Also after the waiting time expires or the number of re-tries exhausted, SD should log a final error and ask for Admin intervention. Business justification: I have made some further research and discussed with the team. It has been found out it is only the following 2 conditions will trigger SD to failover (restart) SSM. I have also tested the conditions in a test cluster in which both situations SSM was restarted.

Previous SSM pid disappeared Physical network connection to SSM from SD broke at TCP connection level In our case here, for the following error message we see: [...] It was the API connection from SD to SSM querying information failed. However, the TCP connection is still alive. This is very much likely to be the case why the error message reported by SSM was not restarted by SD.

it should be the SD's responsibility to detect whether to restart SSM and under what circumstances. So far we've found SD will only do this for broken connection or missing SSM PID. It must have been designed this way for a reason in the first place. We might need your help to log an RFE if you think there are some situation you would like SD to have SSM restarted.