Backup Exec 12.5 with HP MSL2024 920 SAS can't run 2 jobs at once.

[BCIKevin]

Forgive me for the length of this, but I wanted to supply all the needed information.

We’re running Backup Exec 12.5 Rev 2213with all the appropriate agents.
This is on a Windows 2003 sp2 system (32 bit)
The server is an HP ProLiant DL 360G6
The library is a HP MSL2024 SAS 920 library with 2 drives and 30 tape capacity, using LTO3 tapes.
Driver for the library is 1.8.0.0 using the backup exec library driver tool.
Driver for the HBA is 1.28.2.1

When we first got the library back in Sept, we were running on an older DL360G4, on which we had Backup Exec 12.0. The HBA was a HP SC44Ge. After installation, we were fine unless we tried to run two backup jobs at the same time. After the second job would start, load the tape, and begin writing data the first job would immediately stop writing data to the tape, and eventually time out with an unkown failure. Event logs would report events 11 and 127 LSI_SAS saying “The driver detected a controller error on \Device\RaidPort1”

We did a lot of troubleshooting, including updating firmware and drivers, disabling the HP Insight tools, the registry fix for the Storport driver, to name a few. Eventually the HP tech we were working with told us that the DL360 wasn’t in the compatibility matrix with the library we got. This was obviously annoying, but since our old library (an Overland Neo 2000 series) was still functional, I decided to get by using that. We could run two jobs at once with it with no problems, and as my budget was short for a new server we pressed on.

Eventually, we got a new server ordered that complied with the compatibility matrix (the DL360G6) and installed a fresh install of the 12.5 software. Much to my annoyance, we had the same problem. I called HP once again, and ran through all the same steps we did before, updating drivers and firmware, disabling HP Insight tools, registry fix, etc. After all that, he told me the HBA (the same model SC44Ge as previous server) was compatible with the server, but not with the library and recommended a different model, SC08Ge. This also required a different cable which I ordered as well once I saw the cable for the SC44Ge wouldn’t fit. We got that installed, device manager was happy, I ran two backup jobs and basically have the same problem. The only difference now is the first job keeps going, while the second job times out after a while with the same errors and takes the drive off-line.

Any suggestions that anyone would have would be appreciated. I’m running our backups on this server now with the new library as the Overland finally started having issues after 5 years of faithful service. As long as I time out my backup jobs so that 2 don’t run at once I’m fine, but it would be more convenient to be able to have that ability (not to mention that’s what we paid for). I know the HP techs will eventually get around to blaming Backup Exec again (they tried that with the old server too, but lost that argument when I told them the Overland worked fine). I just want to ensure that I cover everything before I start pitching a fit over their hardware again.

[Ken_Putnam]

We’re running Backup Exec 12.5 Rev 2213with all the appropriate agents.

Does that include one LEO?

Out of the box, BackupExec will support  one Library/Loader with one drive.  Adding a extra drives or a second loader (with one drive) requires one LEO each

[BCIKevin]

Yes, we have the library expansion option as well.

[Ken_Putnam]

Is the Library higher on the channel that the two drives?

if you shut down all BackupExec services, can NTBackup see and access both drives at the same time?

[BCIKevin]

Not sure exactly what you mean.  If you mean by looking on the devices tab, the library is above the two drives on that.

As for NTBackup, I've not played with that much.  I stopped the Backup Exec services but how do I see if NTBackup can access both drives at once? 

Thanks for the responses.

[Ken_Putnam]

Haven't played with SAS much, but with SCSI, you need to have the library with a lower SCSI ID than the drives (higher priority)

Once backupExec is stopped, just execute NTBAckup and see if it sees both drives.  if so mount a couple of scratch tapes and try two small test jobs at the same time

[BCIKevin]

It doesn't appear that it will.  I tried running a backup job, then kicked off another right afterwards.  The second stayed at "Mounting the media" until the first completed, then kicked off.  Also, under computer management in the removable storage area, it has the library shown as active, and one TEAC DV-28S-V device which I assume is one of the drives, but it doesn't show another one.  So while Device manager shows both drives, it appears that NTBackup is only seeing one of them.

I haven't been able to determine how to find the SCSI id yet, still researching that one.

[Ben_L_]

Can follow the directions in this technote to get the hardware configuration and post the file up so that I can review the data?

http://seer.entsupport.symantec.com/docs/322814.htm

[BCIKevin · Accepted solution]

Thanks for the suggestions gentlemen.  I was intending on proceeding with the directions Ben L. supplied, however I ended up on the phone with an HP tech and it appears we resolved the problem.

There was a more recent update for the firmware for the library and drives, using HP's LT&T software, we upgraded that.  It brought the library to ver 4.6 and the drives to C25W if I recall correctly. We also added some additional registry entries for the drivers for the HBA.  I unfortunately did not write them down as we went, as I was just going through things as directed and was doubtful that we would resolve it today.  However it involved adding an AutoRun DWord value set to 0 in HKEY Local Machine, System, Services for the drivers listed for the library and tape drives.  It also involved adding a new key string for Storport with a Dword value called BusyRetryCount with a hex value of 250.  I believe that was in HKEY Local Machine, System, Current Control Set, Enum, SCSI, Sequential&Ven_HP&Prod_Ultrium.  We did that for each string relating to the 5&2740 strings under Device Parameters.   I believe there is a technote out for this process, as we had followed it before but after the installation of the new HBA I hadn't done it again.

Regardless, after a reboot I ran two test jobs and they both ran at the same time and completed successfully.  I'll be monitoring it over the weekend to ensure there are no more issues, but it appears to be resolved.

Thanks again for the assistance.  After several months of dealing with this I honestly didn't think HP would come through, but it appears they did.