cancel
Showing results for 
Search instead for 
Did you mean: 

Mode sense taking a long time > 40 seconds in Robtest

ReverbD
Level 2
Hi all,

need some idea's on how to track down the cause of my Netbackup issues. I am running Solaris 10 X86 and NB 6.5.3, Spectralogic library. Currently only have the master seeing 2 drives, and robotic control on drive 1 lun 1. I've been trying to track down some slowness and scsi timeouts. I find that when I run robtest, mode sense takes greater than 40 seconds to come back and allow me to see drives/slots etc. I am thinking that the issue is with either the HBA used to talk to the drive or somewhere along the path. Any help would be appreciated.

thanks,

D
3 REPLIES 3

Anonymous
Not applicable
Things to tick/check:

1) Your hardware is supported in the NetBackup latest HCL.
2) In addition firmware levels/revisions checked. Release notes may reveal known issues you experiencing.
Just like this technote reveals:
http://seer.entsupport.symantec.com/docs/270136.htm
3) OS/Hardware drivers up to date
4) If a SAN, verify your zoning.

Does the netbackup 'sgscan' command work OK at reporting device discovered.
http://seer.entsupport.symantec.com/docs/291443.htm

Omar_Villa
Level 6
Employee
Check performance is something complicated, but a solaris haves a bunch of nice tools who can help you to troubleshoot this, just like iostat -exnDl st1, where st1 is the tape drive address, with this you can see at OS level what the speed of the drive is between OS and drive, if robtest takes too long to respond check the library firmware version, probably the issue is with the library and not with NBU or Solaris.

Nicolai
Moderator
Moderator
Partner    VIP   

If you use SCSI HBA make sure you:
  • Aren't using SCSI cables that are too long. Intern cabling count too  (Allowed cabel lengts at  http://www.ramelectronics.net/scsi_connectors.ep )
  • You're SCSI bus terminators are good.
  • Don't have duplicate SCSI id's.
  • Try re-seating SCSI cables and check no connector leg are bend.
You can also swap around cables/terminators to see if the problem follow a pattern.