can't connect on socket

[oracle999]

Backup started Sat Dec 27 12:49:44 2014

12:50:29 Initiating backup
12:55:34 INF - Server status = 25
12:57:22 INF - Backup by oracle on client bumisty3 using policy bucyclone_ORACLE
_sdblive, sched Weekly_Full_Offline:cannot connect on socket
misty3 #

any ideas appreciated

Paul

[inn_kam]

add host file name 

check firewall

[oracle999]

firewall is open

host entries are correct

rgds

Paul

[inn_kam]

"Server Status:  cannot connect on socket "

Comms with database backups work different in comparison with OS backup.

When rman starts, a backup request is sent from the client directly to bprd on the master server (via vnetd).
The master then needs to resolve the client's IP address to a hostname that appears in Client list of policy.

Please ensure that you have bprd log folder on the master. If it does not exist, create the folder and restart NBU. After next failure, check bprd log on master for incoming request from client's IP. If master is unable to resolve client's IP, add to hosts file. Ensure that master can resolve client's IP address to the Client name contained in the policy.

Please also create the following log folders on the client: bpcd dbclient.
Remember to chmod 777 dbclient.

https://www-secure.symantec.com/connect/forums/oracle-rman-backup-issue#comment-6036911

[inn_kam]

Oracle RMAN backups failing with error Server Status: Cannot connect on Socket

Article:TECH147617

|

Created: 2011-01-06

|

Updated: 2013-10-26

|

Article URL http://www.symantec.com/docs/TECH147617

[Marianne]

Wow! Thanks for providing the link to my post that you have copied 'as is'!

[Marianne]

More troubleshooting steps:

Test comms between master and client:

1. Hostname lookup:
Use 'bpclntcmd -hn <name>' and 'bpclntcmd ip <IP' in both directions to verify forward and reverse lookup.
See:

How to verify name resolution for NetBackup (tm) systems, using the "bpclntcmd" command 
http://www.symantec.com/docs/TECH27430 

2. Port connectivity (port 1556 is used in NBU 7.6)
Ensure bprd log folder exists on the master and bpcd log folder on the client. (NBU on master must be restarted to enable bprd log.)
On master:
bptestbpcd -client <client-name> -verbose -debug 
(Check output of command and client's bpcd log. If nothing is logged in bpcd folder on client, there is for sure a firewall problem - check for OS firewall on client as well.)
On client :
bpclntcmd -pn
If nothing is returned, there is either port connectivity problem with master or else the master cannot find a policy for the client name that is resoved from the client's IP address.
Check bprd log on the master for request from client's IP address.

[oracle999]

this is cuurenlty being investigated by oracle networking thanks

[Michael_G_Ander]

Could be a timeout issue, as there over 5 minutes between the Initiating message and the first INF message

For database backups you often have to increase the CLIENT_READ_TIMEOUT and the CLIENT_CONNECT_TIMEOUT

[Marianne]

Troubleshooting steps that I have posted on 6 Jan should not take longer than 5 minutes... maybe 10...

[jim_dalton]

Simplify the problem: can you do  a flat file backup of data on same host? In fact this could be flat file, we cant tell, despite it saying oracle regularly, so might be of benefit to tell us the policy type, and if Oracle have you checked oracle specific instructions followed?

Any history to this? Used to work , now doesnt, perhaps. Tell us about the infrastructure...

Jim 

[oracle999]

this issue was reolved by a restart of the solaris zone