Is there a supported way to attach a Tape Drive to OpenVMS 9.2-2

[pocketprobe]

Are there presently any supported ways to attach a(n LTO) tape drive to a system running 9.2-2?
Ideally via SAS/SCSI

[pocketprobe]

If I understand your problem correctly, you are seeing crashes while trying to get SYSMAN to configure and set up the driver to the actual tape hardware.

That is correct. It either crashes immediately after asking sysman to configure it, or right after "%STDRV-I-STARTUP, OpenVMS startup begun at" when the system is rebooted. Thankfully the drive is external and can be simply not present to allow it to boot.

The patch notes did mention fixes in IO_ROUTINES which was where my crash was, regarding data buffer locations; And fixes in the SCSI driver for one of the controllers I'm using. The large file copy bug also affected the bare metal machine, and mentioned memory boundaries so it was worth a retest overall.

I'm not sure why I focused on the magtape fix when there were a few that may affect this.

[pocketprobe]

This now works in V9.2-3!

Device now is identified correctly. This is actually an HP LTO-5 drive strangely.

Code: Select all

$ show dev mga0 /full

Magtape $2$MGA0: (HELDEN), device type HP Ultrium 5-SCSI, is online, file-
    oriented device, available to cluster, error logging is enabled, device
    supports fastskip (per_io).

    Error count                    0    Operations completed                  0
    Owner process                 ""    Owner UIC                      [SYSTEM]
    Owner process ID        00000000    Dev Prot            S:RWPL,O:RWPL,G:R,W
    Reference count                0    Default buffer size                2048
    WWID   02000008:5001-4380-00AA-5984
    Density                  default    Format                        Normal-11
    Allocation class               2

  Volume status:  no-unload on dismount, position lost, odd parity.

$ init/media=comp mga0: test

Code: Select all

$ bac/med=comp/ignore=label/block=65532/ver/record disk$users:[000000...]*.*;* $2$mga0:test.bck/save
%MOUNT-I-MOUNTED, TEST mounted on _$2$MGA0: (HELDEN)

Backup field works correctly too.

Code: Select all

$ dir disk$users:[users]dana.dir /full

Directory DISK$USERS:[USERS]

DANA.DIR;1                    File ID:  (82495,2,0)
Size:           14/16         Owner:    [DANA]
Created:    20-MAR-2024 16:12:13.99
Modified:   20-NOV-2024 12:16:41.90 (50)
Expires:    <None specified>
Backup:     20-NOV-2024 17:40:22.26
Effective:  <None specified>
Recording:  <None specified>
Accessed:   20-NOV-2024 21:51:16.82
Attr Mod:   10-NOV-2024 19:55:52.49
Data Mod:   20-NOV-2024 12:16:41.90
Linkcount:  1
File organization:  Sequential
Shelved state:      Online
Caching attribute:  Writethrough
File attributes:    Allocation: 16, Extend: 0, Global buffer count: 0
                    No default version limit, Contiguous, Directory file
Record format:      Variable length, maximum 512 bytes, longest 512 bytes
Record attributes:  No carriage control, Non-spanned
RMS attributes:     None
Journaling enabled: None
File protection:    System:RWE, Owner:RWE, Group:RE, World:
Access Cntrl List:  None
Client attributes:  None

Total of 1 file, 14/16 blocks.
$

Code: Select all

$ create/dir disk$junkdrawer:[restoretest]
$ backup $2$mga0:test.bck/rewind disk$junkdrawer:[restoretest...]/by_owner=orig
%MOUNT-I-MOUNTED, TEST mounted on _$2$MGA0: (HELDEN)

Code: Select all

$ dir disk$users:[000000]

Directory DISK$USERS:[000000]

000000.DIR;1        BACKUP.SYS;1        BADBLK.SYS;1        BADLOG.SYS;1
BITMAP.SYS;1        CONTIN.SYS;1        CORIMG.SYS;1        DEVELOPMENT.DIR;1
GAMES.DIR;1         GPT.SYS;1           INDEXF.SYS;1        PUBLIC.DIR;1
QUOTA.SYS;1         SCRATCH.DIR;1       SECURITY.SYS;1      SENDFILES.DIR;1
SYSLOST.DIR;1       T4$DATA.DIR;1       TEMP.DIR;1          USERS.DIR;1
VOLSET.SYS;1

Total of 21 files.
$

Scratch and t4$data weren't created, but they were empty. I'll call it user error based upon my arguments.

Code: Select all

$ dir disk$junkdrawer:[restoretest]

Directory DISK$JUNKDRAWER:[RESTORETEST]

000000.DIR;1        DEVELOPMENT.DIR;1   GAMES.DIR;1         PUBLIC.DIR;1
SENDFILES.DIR;1     SYSLOST.DIR;1       TEMP.DIR;1          USERS.DIR;1

Total of 8 files.
$

Thanks for the new release, and the fix after my report.

[roberbrooks]

We are not claiming support for tapes yet in V9.2-3 because we have not tested it fully.

Fixes were made to the tape class driver SYS$MKDRIVER.EXE. We're pretty confident that the
driver is now correctly ported to X86. Prior to V9.2-3, some work had been done to get it to build
cleanly, and then we sort of forgot about it, because it was less important than other work.

However, once we got fibre channel passthrough to work correctly on ESXi, it was pretty straightforward
to figure out the remaining porting issues.

We expect that we'll announce support for tapes some time in 2025, likely in Q1 or Q2 at the latest.

[issinoho]

Sorry to x-post but OP can you verify that bare metal VMS is supported on an HP Proliant ML150 Gen 9 Xeon Server? Thanks.

[mberryman]

Sadly, I note that while directly attached tapes may work, even if not yet supported, fibre channel attached tapes still do not. Are you able to say when this is expected to work?

I do backups to tape drives on the SAN. Once that is working (and SDL is released for x86) I can probably shut down my non-x86 hosts.

[pocketprobe]

Sorry to x-post but OP can you verify that bare metal VMS is supported on an HP Proliant ML150 Gen 9 Xeon Server? Thanks.

Running x86 OpenVMS isn't supported on bare metal period as I understand it. There was early work on having it boot on DL380 Gen9 servers, it's a touch clunky in spots from not having been fully validated. I'm using that ML150 as it's a quiet tower server and very similar hardware.

I have run x86 VMS on other bare metal as well but it is not guaranteed to work correctly or at all. I simply like sharing my successes and failures with this community.

[issinoho]

I guess "supported" wasn't exactly what I was going for. "Possible" might be more accurate.
ML150s are pretty affordable units. Interesting. Thanks.

[roberbrooks]

> Sadly, I note that while directly attached tapes may work, even if not yet supported,
> fibre channel attached tapes still do not.

Hmmm.

A few notes ago, I wrote . . .
> We are not claiming support for tapes yet in V9.2-3 because we have not tested it fully.
> Fixes were made to the tape class driver SYS$MKDRIVER.EXE.
> We're pretty confident that the driver is now correctly ported to X86.

It works for us -- what operation are you trying, and what error are you getting?

When I say "works for us", our configuration is V9.2-3 using ESXi with PCI passthrough.
Everything we've tried works. As I said, we're not claiming support yet because we have not
done the battery of tests we need to feel confident that everything is OK, but so far, it looks good.

Added in 2 minutes 43 seconds:
> Running x86 OpenVMS isn't supported on bare metal period as I understand it.
> There was early work on having it boot on DL380 Gen9 servers, it's a touch clunky in
> spots from not having been fully validated.

On gen 9, you might have success getting a local disk to boot.

On a gen 10, you'll only have success using fibre channel; we don't have a driver for the adapters
that ship with gen 10.

As noted, bare metal is not supported, but we boot bare metal to do performance testing on a DL380 gen 10, using
fibre channel to boot.

[mberryman]

I am running ESXi on a DL380 G9 with a Qlogic QLE2692 HBA using PCI passthru. In this setup, disk access works great.

My tape library is an HP MSL4048 with two tape drives, an LTO4 and LTO7.

After installing V9.2-3, I attempted a SYSMAN IO FIND_WWID command with no result (and no error). I even tried copying the SYS$DEVICES.DAT file over from an Integrity system and rebooting but still no mag tape devices were found.

[roberbrooks]

> I am running ESXi on a DL380 G9 with a Qlogic QLE2692 HBA using PCI passthru. In this setup, disk access works great.

What version of ESXi; what firmware version on the fibre HBA?

I'm grasping at straws in trying to understand what isn't working for you.