I have a molecule set up in the pre prod environment. There are couple of applications deployed for performance test.
We have a shared EFS where the atom installation is done in AWS. What I am observing after running the schedules for let say 3 ~4 days the atom get crash . Even we are not able to login to the EFS .
We took the issue to AWS and the response we got is as below.
The atom process is causing multiple CLOSE_WAIT in OS which is causing the EFS to unmount. On killing the atom process the issue is getting resolved.
We took the issue to Dell Support but getting no clue if any deployed process is causing this issue ,or there is any configuration at Dell end or OS level setting need to be done.
Normally the CLOSE_WAIT situation comes where there is too many open connections. We are using web services,SAP connector,DB connector and SF connector . I have confirmed with Dell support if we have to explicitly close the connection but the answer was no.
Again a strange behaviour I have observed MULTI_THREAD has better performance than MULTI_PROCESS in our environment .Any suggestion towards it will be of great help.