sysman error when issuing io replace mga6 --- illegal I/O function code

[Lucie]

I thought someone might have an answer before I put a case into HP.
System is OpenVMS 8.3 1h1.
It is part of a 3 alpha node cluster

I issued an io replace mga6 and received the following error.

SYSMAN-I-NODERR, error returned from node N058V7-SYSTEM-F-ILLIOFUNC, illegal I/O function code

The sys$devices.dat was updated but when I issued an io list I received the following output.

On port _FGA0:, the following tape WWIDs are not yet configured:
Target 15, LUN 6, COMPAQ SDLT320
WWID=02000008:500E-09E0-0018-325B

On port _FGC0:, the following tape WWIDs are not yet configured:
Target 15, LUN 6, COMPAQ SDLT320
WWID=02000008:500E-09E0-0018-325B

Show device/full mga6 indicates N058V7 has access to the tape drive via MSCP Server not the fiber.

Does anyone have any suggestions as to what must be done to correct this?


Thanks!
Lucie

[Hoff]

You have an HP support contract? Um, why not use it? Your company paid money for that support contract, and paid money to get to the front of the question queues to get your answers, and specifically paid for fast (and official) answers for cases where you hit errors with OpenVMS.

If you're willing to follow what is usually a slower approach for support (ITRC), what's the physical connection out to the SDLT320 drive? This is a connection via NSR or MBR, right?

What sequence of SYSMAN commands and what REPLACE command did you use here? With the WWID or not?


And if you've not already seen this, here is the standard documentation for the sequence....

IO

REPLACE_WWID

Description

You can use the two parameters, devnam_string and wwid_string,
with the REPLACE_WWID command to replace a broken tape device
with a new device. The command automatically updates the data
structures that record the new devnam-WWID correlation, and the
device automatically begins to function correctly.

This command is useful in two different cases:

o In one case, the drive might malfunction and need to be
replaced immediately without rebooting the system. If this
happens, the drive is physically replaced with a new drive,
and the command SYSMAN IO REPLACE_WWID $2$MGAn is issued
clusterwide. The /WWID qualifier is not appropriate in this
case, because the new WWID is automatically detected using
information stored in the device's data structures.

o In the other case, the drive might malfunction and not
be replaced until after the system has been shut down or
rebooted. The device name no longer appears in the SHOW DEVICE
display because the device failed to configure during the
reboot.

The configuration failure occurred either because the
broken drive did not respond, or because the new drive has
a different WWID from the one SYSMAN IO AUTOCONFIGURE expected
at boot time. Therefore, in this situation, in which the
device name is in SYS$DEVICES.DAT but not in the SHOW DEVICE
display, use the /WWID qualifier to define the new devnam-WWID
correlation.

Follow these steps clusterwide:

1. Execute the SYSMAN IO LIST_WWID command to display the new
WWID.

2. Use the command SYSMAN IO REPLACE_WWID $2$MGAn/WWID=new_
wwid to define the new correlation.

3. Use the SYSMAN IO AUTOCONFIGURE command to configure the
device.

When you use the SYSMAN IO LIST_WWID command, keep in mind that:

o You must set the replacement device to the same SCSI target ID
as the original device.

o You must stop all activity on the device before issuing the
SYSMAN IO REPLACE_WWID command.

o The command requires CMKRNL privilege and applies only to FC
tapes behind an NSR or MDR.


--

And yes, the whole WWID stuff in that description is rather murky. With(out) your current command sequence, I'd guess that you have missed the invocation of the IO AUTOCONFIGURE auto-configuration command in SYSMAN.

[Lucie]

Thanks Hoff! Yes I can put in a ticket but sometimes ITRC is faster with a solution! :)
The connection via NSR (external)

The drive was replaced several month ago and a IO REPLACE was not issued.
New drive was added in the san and there was a io auto issued.

my command sequence was
IO replace $2$mga6 (received error)
IO LIST (received output, see first thread)
IO replace $2$mga6 /wwid=02000008:500E-09E0-0018-325B
(wwid received from the IO LIST.)

returned to sysman prompt.

The wwid information is in the file, but when issuing a IO list - indicates drive is not configured.
I know if you don't follow certain sequence, things can get out of whack. I think the system is half configured because an IO AUTO was issued before an IO REPLACE.

I followed your sequence:

1. Execute the SYSMAN IO LIST_WWID command to display the new WWID.

2. Use the command SYSMAN IO REPLACE_WWID $2$MGAn/WWID=new_wwid to define the new correlation.

3. Use the SYSMAN IO AUTOCONFIGURE command to configure the device.

and it worked!

SYSMAN> IO LIST_WWID

%SYSMAN-I-OUTPUT, command execution on node N058V7
On port _FGA0:, the following tape WWIDs are not yet configured:

Target 15, LUN 6, COMPAQ SDLT320
WWID=02000008:500E-09E0-0018-325B

On port _FGC0:, the following tape WWIDs are not yet configured:

Target 15, LUN 6, COMPAQ SDLT320
WWID=02000008:500E-09E0-0018-325B

SYSMAN> IO REPLACE_WWID $2$mga6 /wwid=02000008:500E-09E0-0018-325B
SYSMAN> io auto
SYSMAN> IO LIST_WWID

%SYSMAN-I-OUTPUT, command execution on node N058V7
SYSMAN>
N058V7 has access to the tape drive fiber

Thanks Hoff! It was faster than calling!!

[Hoff · Accepted solution]

Credit where due: that sequence was directly copied from the SYSMAN utility's HELP text for the specified command.

[Lucie]

Thanks Hoff!
I will hunt down the manual!

[Lucie]

Thanks for the help!

[John Gillings]

Lucie,

Although ITRC may sometimes be faster than logging a case, please make sure you always log a case against your contract as well. If you get an answer here first, that's great - the HP tech won't mind closing the case as solved, and you can get a second (official) opinion.

Although ITRC is provided by HP, HP technical support folk are NOT encouraged to answer questions here (indeed, it's somewhat the opposite). The suits running the show would LOVE customers to stop logging calls to costly call centres, and rely on "free" channels like this. As call volumes go down, they get more metrics they get to justify their actions, and ITRC gets more ex-HP engineers doing corporate charity work ;-)