Cannot connnect on socket fixed by making bpcd and bprd log directories?

[Wile_E__Coyote]

7.1.0.4

Working on a DR exercise this morning, and we had a client that would not connect to the Netbackup servers.

Getting a "cannot connect on socket" error.

Able to telnet to ports 13724 and 13782 from client to server, and from server to client.

Still getting "cannot connect on socket" error.

So, I created the bpcd and bprd log directories to troubleshoot, and all of a sudden, it was working!

It's repeatable. Same problem happened on another client. I was asked "What did you do to make it work?" so we went through the same process as above, checking for connectivity after every step.

Creating the bpcd and bprd log directories definitely fixed the problem. The question is, why?

[Nicolai]

Netbackup 7.x  uses port 1556 for communication and 13724 for backward comparability.

I have not heard of such a fix before.

[Marianne]

NBU refreshes host cache once every hour. 

Changes/updates to hosts files and DNS are not read by master and/or media servers right away.
This can be fixed by running 'bpclntcmd -clear_host_cache' on the servers.

Chances are that host cache automatically refreshed in the time it took you to create log folders.

See http://www.symantec.com/docs/TECH136792 

When and how to clear the NetBackup host cache

[mph999]

I agree with Marianne - something else must be happening. I can't think of any possible link between 'Can't connect on socket' and log files. If this really is caused by the log files, it would be quite a 'deep routed' issue, by which I mean I would expect it to be something related to the code - and as we know, it works for everyone else, including your production servers. I guess it is easy t prove. Create your log files Clear host cache Run job and show it works Remove bprd log, restart bprd and rerun job - still working ? Put bprd log back, restart bprd - test again ? Remove bpcd log - test again ? M