Showing results for 
Search instead for 
Did you mean: 

MS_Exchange_Server backup policy failed on DAG mail servers with error 13 file read failed

Level 2

Hi everyone.

I have a big problem to backup 21 exchange databases GRT enable, at the begining of policy started get error az below

      "socket read failed, an existing connection forcibly close by remote host . ( 10054 )"

after that

       "file read failed  13" 

screenshots are attached.

I did all microsoft solution for event 10054 about keepalivetime and connection time out but not wokring.

I tried to split 21 databases to different policy, some policy worked but almost failed.

Exchange ver: 2016

Exchange sevres os: WIndows DataCenter 2016

Anybody can help us ?!


Partner    VIP    Accredited Certified


was it ever working, or it is a new installation? If it was working earlier, what was changed?

You mention GRT - it means that without GRT it is running fine? (I guess not, because it is failing in Snapshot phase..).

Review VSS state on DAG nodes with "vssadmin list writers". If issues found , google MS/Veritas recommendation for resolving.

You are in DAG, so try also another copy in Policy Attributes (Passive instead Active etc.)



Hi Michal

Thank you so much for reply.

It is new installation.

Yes without GRT work fine.

I have checked VSS state on DAG nodes no error exist as below attached.

I have selected DAG virtual name and Netbackup selects passive node to backup automatically.


What is the NetBackup version on the Exchange servers?

If you look in the Event log, do you find a bpfis exception?

What errors do you find in the bpfis log? If bpfis crashes, what are the last few lines in the log?

My guess, pending your answers, is that bpfis on the client can't query the NetBackup Framework Discovery Service on the master server. If I down the service, I can produce a bpfis crash on the second database it tries to process for GRT.

In this case, you would find "ExchangeResearcher::queryDiscService(): Unable to get iterator for researching" in the bpfis log, and the bpfis log would end with a CORBA exception. I think it applies only to version 8.2. The solution would be to make sure the discovery service is running on both client and master server, and look for reasons the query from client to server would be blocked.