While running a deployment they are getting an error during the artifact distribution phase. The error: Failed to download artifact [ Artifact [ <name of artifact>]] to retrieval agent [<retrieval agent hostname>]. Local agent error is [Artifact download task was cancelled, agent [<retrieval agent hostname>] does not respond].

CA Release Automation 6.4.0.10072

By running "netstat -aonp | grep 6600" on the execution server we confirmed there were zero connections which is a big red flag. At a minimum you should see the CA Release Automation process binding to the network adapter on port 6600 since that is the port that the NIMI module uses and binds to. That is the local connection that should be shown (always) when the execution server is running. Then you should see remote connections being made (by all of the agents that report to this execution server) to the 6600 port on the execution server. 

The strange thing is that ASAP -> Agent Management doesn't show this execution server as critical (or its agents). This was somewhat explained by running "netstat -aonp | grep 8080" and "netstat -aonp | grep 61616" on the execution server which was showing the local bind to the network adapter that the execution server makes as well as the remote connections by the management server. If port 8080 and 61616 are available then that explains why the management server "thinks" everything is okay even though there is a big problem with NIMI. In this case, which is extremely rare, NIMI appears to not be running which means that the execution server is not going to reflect any kind of agent update status. 

1. We stopped the CA Release Automation services on the execution server.
2. Then we cleaned up files/folders outlined in this KB article on the execution server: Safely deleting files from CA Release Automation Agent install folder
3. Then we deleted the folder: <ExecutionServerInstallDir>/activemq-data/nes/LevelDB
4. Then we started CA Release Automation services on the execution server, waited a moment and confirmed that connections on 6600 were being made, confirmed artifact distributions worked and that there were no more unexpected problems with NIMI.