3719 Hardware Errors on New Tapes

[halomerchant]

We have relatively new tapes in our library and randomly get the dreaded 3719 unable to write to media errors almost nightly on 2 or 3 out of 140 backups. I don't get it, sometimes the backups never error and then bam we'll get a bunch of these. All relatively new StorageTek LTO1 drives in an L700 and all diagnostics check out alright. CA keeps saying its a timeout error on the drive or some type of overheating is occurring. StorageTek says its the communication within Arcserv and the tape library. I can't win this battle with them. Why do these guys partner if they can't fix problems for big companies? We're probably dropping Arcserv for TSM because of this issue, just not reliable.

 

[vschumpy]

You can start here if you haven't already:

http://supportconnect.ca.com/sc/redir.jsp?reqPage=search&searchID=TEC342813

CA should be able to log a call with STK or vice-versa via TSANET so that they can work together on this if indeed they are not already.

You can also use STK's utilities to see if any hard or soft errors are logged against any of the drive units.

 

[davidmichel]

If the Tape Engine is set to Detail or Summary mode those errors will be in the log. Copy some of them here and lets take a look.

For the next failure get the time stamp it happened from the activity log and then cross reference that with the tape log. The tape log is criptic but errors are usually fairely easy to spot. Once found copy the section with the error here.

 

[davidmichel]

oh and by the way this type of thing comes up often, software support sees Hardware Error and says its the hardware, hardware support runs diagnostics with no problem and says its the software. Sure you can use another program and chances are the there will be no more errors. Of course it does not necessarily mean the errors are not happening, they could still be happening but just not reported. Any program that uses the OS tape device driver can not capture the level of errors that ARCserve does because ARCserve uses SCSI commands direct to the device.