Large oracle restores fail through windows Media Servers - 2801

[gpar]

Hello,

We are having an issue where large oracle restores fail through our windows media servers, restores through Linux Media servers are succesful.

Netbackup 7.6.1.2

Not an alternate server restore

KeepAliveTime set to 2 hours on windows media server as well as linux

From the DBclient log 

10:36:43.158 [94913] <16> readFromServer: ERR - recv() from server on socket failed: Connection reset by peer (104)
10:36:43.158 [94913] <16> dbc_get: ERR - failed reading data from server, bytes read = -1

After this point the backup fails with an error 2801.

Any suggestions?

[Tape_Archived]

This is connectivity issue for sure. 

* Ensure that the client server list contains entries for the media servers that can be used during a backup or restore.
* Check ownership and permission on the Oracle instance that you restore.
* Check the NetBackup dbclient and user_ops logs for additional failure information.

Check that the Oracle user on the client can successfully communicate with the master server by using the bpclntcmd -pn command.
* Check that the Oracle user on the client can see the backup images on the master by using the bplist command.
* On UNIX and Linux hosts, check that the oracle user has sufficient permissions to read /etc/services, /etc/nsswitch.conf, and /usr/openv/netbackup/bp.conf.

[Marianne]

2 lines from one log does not help to troubleshoot.

Entire dbclient log is needed to follow the process from client side.

bpbrm and bptm logs on the media server is needed to check for comms, timeouts and data transfer.

What is Client Read timeout on the Windows media server set to?
Database restore normally needs a large Client Connect and Client Read timeout to cater for db activities that NBU is not aware of.