[comp.os.vms] disk/TK50 interaction?

ADA16%jrs@ICS.UCI.EDU (Mark) (02/11/88)

We are having problems with our TK50 and BACKUP.  We put big disk drives on
our MicroVAX II in November, and since then we have been unable to complete an
image backup.  The tape unit gets a fatal error or announces that it is
"not software enabled" before the backup completes.  We are also getting a
very high error rate, even with successful incremental backups.

Here are the details of the configuration:
MicroVAX II in small chassis (BA23-A)
12 MBytes add-on memory
DEQNA (Ethernet)                        TK50 (Allegedly latest rev levels)
>>>Systems Industries (SI) QDA4EC disk controller<<<
DZQ11 (4-line terminal controller)      RQDX3 (DEC Disk controller)

The pieces above are the controllers, in their positions on the bus.
The 70MB disk that came with the system is the system disk.  The Systems
Industries controller is controlling four MAXTOR 4380 disk drives (380MB),
which SI calls SI55's.  These are bound as a volume set (DUB0 - DUB3).

The command
$ BACKUP/image dub0: mua0:user.bak/blocksize=32768/buff=5
results in about one tape error every 12 I/Os, with an eventual crash of the
program.  The same command on the system disk gives about an error every 100
I/Os.  On our other MicroVAX, with Fujitsu Eagles, we get about an error
every 300 I/Os.  (Note that these are tape errors; we haven't had ANY errors
on the disk.)

Anything that slows down the I/O seems to lower the error rate.  Using
a blocksize of 2048, for example, gives an error only about every 250 I/O's.
Using fewer buffers, using COPY instead of BACKUP, and doing BACKUP on
selected files (/SINCE=...) allhave a lower error frequency.  (Note that the
MAXTOR drives are considerably faster than the system disk, and somewhat
faster than the Fujitsus.)

Has this happened to you?  The TK50 drive has been replaced twice
and the controller has been replaced once.  The SI controller has also been
replaced.  Any ideas?

		Mark Wadsworth