cancel
Showing results for 
Search instead for 
Did you mean: 

Problem With Failed Backups - Possibly due to network issues

Mark_Williams_4
Level 2

Hello,

We've been using BE2012 for a couple of months now and we've recently been experiencing a problem with our Exchange 2007 backups (they basically fail about 3/4 of the way through with either an E000FF12 or E00084F8 error) which seem to indicate an issue with the network during backups.  These Exchange backups occur nightly and run with a number of other concurrent backups happening at the same time.  

To try and even the network load amongst jobs, I've got four NIC's in my server (2 x NC7782 and 2 x NC1020) and the server is a DL385 G1 running Windows 2008 64-bit, 16GB RAM, 2 x Dual Core processors.

Each NIC has a seperate address assigned to it which is also on the same subnet as the servers we want to backup.  So for instance NIC1 has the address 10.2.0.2, NIC2 10.2.0.3, NIC3 10.2.0.4, and NIC4 10.2.0.5.  I've tried to configure my individual jobs so that they use a specific NIC from the list to even the job-to-NIC numbers.

When the jobs fail and I check the statistics for the individual NIC's, they do show a large number of Receive, Overrun, and No Buffer errors.  So I'm assuming that our NIC's aren't up to the job of handling the amount of backup data which is being sent to it by the various remote agents.

Has anyone got any advice (apart from replacing the BE2012 server with a more up to date model) ?

Thanks

Mark

 

 

6 REPLIES 6

CraigV
Moderator
Moderator
Partner    VIP    Accredited

Hi,

 

Have you made sure that your NICs and switch ports are all hard-coded to the fastest speed of the NIC (ie. 1GB FULL)?

Have you tried to stagger your job start times a bit more?

Thanks!

Mark_Williams_4
Level 2

Craig,

Thanks for replying.  We currently have the NIC's on the server set to auto negotiate the speed and duplex and the same on the individual ports on the switch.  Checking the switch, it seems to negotiate okay with 1GB/Full and I don't see any dropped packets.

Would you still advise hard setting 1GB\Full both on the NIC's and the switch ports ?

Will look into staggering the jobs.  The only downside with BE2012 is that jobs are now a job per server rather than a job containing many servers.  So whereas before it would have worked through a list of servers in a job sequentially and couldn't start the next server until the previous one had completed, we now have more servers be backed up concurrently which is obviously meaning more network traffic for the NIC's to deal with at the same time.

 

Ken_Putnam
Level 6

For some reason, with auto on both ends, most traffic will flow OK, but the streamed data from BackupExec will have problems

 

If you site/company standards say the switches MUST be set to Auto, try hard setting the Server NICs and see what happens  

CraigV
Moderator
Moderator
Partner    VIP    Accredited

On our side, we've always set the NICs on the servers & the switch ports to a hard-coded setting. Seems to work well and takes any additional strain away from the server trying to adjust the NIC speed.

Mark_Williams_4
Level 2

Thanks to you both for your replies.  I will hardcode both the speed and duplex on the NIC's and the switch ports as suggested.  We aren't currently seeing any statistics on the Cisco switches these are plugged into to suggest dropped packets but it won't harm to hardcode at both ends. 

Ken - when you say that it's known for Backup Exec streaming data to be affected if we use auto negotiate, if this a documented issue anywhere ?

We seem to be noticing the "network" related failures more for Exchange jobs so I assume that the pattern of Exchange backup data may be different than normal file level backups and that any network delay can knock out the Exchange backup.  Does this seem feasible ?

Thanks

Mark

 

JJMcD
Not applicable

Hi,

Have you checked the Exchange database size? Has it exceeded the default set size of 50gb?

I had a situation where the database was being dismounted everynight and the backups did not complete.

I have Exchange 2003 standard and am limited to 75gb. Archiving and setting a policy in active x to delete all " deleted" items regularly, sorted the issue.

You, however have Ex 2007 - you can adjust the max size to suit your operation.

 

Regards.

john