we are experiencing odd errors with backup sessions expiring mid backup while the client is continuously sending data. This happens exactly 2 hours after the backup started. As shown in the log below the backup still runs for a while and then stops with the nsrsqlsv.exe on the client side giving the well known RPC error: "c_tcp.c:1549 Failed to write to socket 1304: An existing connection was forcibly closed by the remote host"
Client throughput was visible on the client via perfmon, as well as on the backupser side with the amount of data saved constantly increasing.
We already:
- set TCP keepalivetime to 5 minutes
- set NSR_KEEPALIVE_WAIT to 10
daemon.log:70896 05/24/13 07:57:47 AM nsrd clientname:MSSQL:databasename saving to pool 'DBNormal' (151804)
daemon.log:70896 05/24/13 08:45:44 AM nsrd clientname:MSSQL:databasename saving to pool 'DBNormal' (151806) 25 GB
daemon.log:39076 05/24/13 09:57:45 AM nsrexecd GSS warning: Session information for 'sapadm/clientname@clientname' (session number ba9e1:39e2) was requested by 'backupservername' but the session has expired and is no longer valid
daemon.log:39076 05/24/13 09:57:47 AM nsrexecd GSS warning: Session information for 'sapadm/clientname@clientname' (session number ba9ec:39e9) was requested by 'daemon:390103:2/backupservername@unknown' but the session has expired and is no longer valid
daemon.log:71659 05/24/13 11:53:40 AM nsrd clientname:MSSQL:databasename done saving to pool 'DBNormal' (151806) 254 GB
daemon.log:70896 05/24/13 12:32:23 PM nsrd clientname:MSSQL:databasename saving to pool 'DBNormal' (151806)
daemon.log:70896 05/24/13 12:35:08 PM nsrd clientname:MSSQL:databasename saving to pool 'DBNormal' (151835) 2849 MB
Is that a known issue and is it possible to increase the session timeout or monitor why the session expires?
NSR_KEEPALIVE_WAIT and the TCP Keepalive settings are already set. Currently we are investigating the firewall since we see drops that are related to the connection. Yet the firewall doesn't always drop packets when the backup aborts.
the problem has been resolved by fixing a port flapping issue on the network. Thanks for your suggestions. In this case the networker was just the victim not the culprit ☺
Von: Umraan
Gesendet: Donnerstag, 30. Mai 2013 09:23
An: Haase, Stefan
Betreff: Re: - Backup session expiring mid backup
ECN
Backup session expiring mid backup
reply from Umraan in NetWorker Support Forum - View the full discussion
Brian91_4544b4
71 Posts
1301
0
Posted May 24th, 2013 04:00
Hi Stefan
We've had issues in the past like this for some systems. As well as setting the NSR_KEEPALIVE_WAIT times, you could try the following.
How to set TCP/IP keep alive in Windows
https://solutions.emc.com/emcsolutionview.asp?id=esg60759
How to disable IPV6, TCP Chimney and Receive Side Scaling (RSS) on Windows 2008
https://solutions.emc.com/emcsolutionview.asp?id=esg118437
Hope this helps
Brian