Hey all,
I recently built a PoC environment with Horizon 7.2 used for only RDS hosted apps. Pool is set to use only BLAST protocol and not allow users to choose others. It worked fine for a few days after implementation, and then the customer reported that their RDS hosts were going into agent unreachable status. Issue was reproducible by connecting to an application, closing the horizon client, and attempting to reconnect from another PC. The user would get a message saying the remote host is not ready to accept connections, and then a few minutes later the RDS host would go into agent unreachable status in View administrator. I checked the View Agent service status and it was still running on the RDS host, and there were no references to it shutting down in the log or debug log.
Today, I found some odd messages in the blast-worker-sessionid2.log file at the time of the issue. This repeated every 10 seconds until the time of me capturing the log files:
2017-10-25 15:34:47.625-0400 [WARN ] 0x1178 bora::Warning: socket 113 state: CLOSE_WAIT local: 0.0.0.0:49152 remote: 10.1.10.56:61769
2017-10-25 15:34:57.323-0400 [WARN ] 0x1178 bora::Warning: socket 113 state: CLOSE_WAIT local: 0.0.0.0:49152 remote: 10.1.10.56:61769
2017-10-25 15:35:07.021-0400 [WARN ] 0x1178 bora::Warning: socket 113 state: CLOSE_WAIT local: 0.0.0.0:49152 remote: 10.1.10.56:61769
I then changed the pool to use only PCoIP and so far, have not been able to reproduce the issue. Has anyone seen anything like this with BLAST?
Thanks!