MACALLSTR@vax1.physics.oxford.ac.UK (12/17/87)
This is an update to the TA78/79 controller drop-out problems we are experiencing. The following information has been obtained only after several days of monitoring and testing the tape drives. The good news is that, with careful operational procedures, you may avoid the problem most of the time. The bad news is that this drop-out is a feature ( Can you believe it?! ) of the TA78/79/HSC setup. To remind you of our straight-forward setup : we have a VAX8700 + HSC50 + TA78/79 with four TU78/79 drives. What you must AVOID doing : DON'T switch the tape drive OFF-LINE/ON-LINE while it is still software mounted and active. If you do, the TA78/79 controller will DROP-OUT - GUARANTEED! You may initially think things are OK as other jobs using the other drives will continue running BUT as soon as you attempt to software remout the drive AND any other drives in use have dismounted the tapes in use at the time the TA78/79 will DISCONNECT - GONE! This means that any jobs which have dismounted their current tapes for any reason but are not yet finished using the tape drive will fail e.g. dismount/mount between continuation volumes and jobs using continuation volumes are usually the long tedious ones. If no other drives are active the TA78/79 will drop-out immediately. The TA78 can be brought back to life by either toggling the PORT button or by pressing the internal white reset button under the lower front cover of the TA78/79 cabinet. On the positive side, having had our TA78's upgraded to TA79's ('Get Well FCO') there has been a dramatic improvement in the robustness of the drives to error conditions and, in the course of testing, every imaginable error situation was contrived and the drives remained alive. We can now write tape after tape ( e.g. in full back-ups ) with, consistently, error counts of zero. Previously, on the TA78's, even the best quality tapes were recording error counts in the range 0-10 per tape. Although, good operational procedures can reduce the incidence of someone switching an active drive OFF-LINE/ON-LINE, even experienced operators can make mistakes ( especially under pressure ) and if general users have direct access to the tape drives such occurrences are guranteed from time to time. The impact of this TA78/79 feature will vary from site to site. Why an OFF-LINE/ON-LINE condition on one drive should cause the TA78/79 controller to drop-out and so disconnect ALL FOUR TA78/79 drives beats me?! I might accept temporary disconnection of the one drive on which the OFF-LINE/ON-LINE condition occurred as an OFF-LINE/ON-LINE interrupt is an abnormal operational situation but not disconnection ( even temporary ) of all four drives. I hope DEC will provide a microcode fix to handle this error situation in a more satisfactory manner. I hope this information will enable TA78/79 users to enjoy their drives. I hope, too, that, if you're troubled with this TA78/79 disconnect problem, you'll bang on DEC's door. Fixes will only be implemented if DEC consider it's generally worthwhile and something is usually commercially worthwhile only when it satisfies a significant number of customers ( or a number of significant customers! ). John