Problem with removing LDA device

[Piet Timmers_1]

ls.

I have a device LDA59: , not mounted, not allocated.

$ LD DISCONNECT LDA59: /ABORT gives
%LD-F-DEVASSIGN, Device has channels assigned

I have tried everything I know, even
$ANA/SYS
SDA> SHOW PROC ALL/CHANN

I found no process with channels open to LDA59:

Help.

Greetings,

Piet

I know, a reboot will solve this??!!??
(but this is an OpenVMS node, not windows...)

[Karl Rohwedder]

Perhaps on another node in a cluster?
LD SHOW /ALL looks o.k.?

regards Kalle

[Volker Halle]

Piet,

did you happen to use LD CONNECT/REPLACE ?

What does $ LD SHOW LDA59: report ?

Volker.

[Piet Timmers_1]

$ LD SHOW/ALL tells me that the device LDA59: is connected to a DSK file.

$ SHOW DEVI LD says status Online

The node is part of a cluster, but the LD driver is only started on the current node.

Greetings.

[Piet Timmers_1]

/REPLACE is not used, the LD device is connected to a DSK file.

$ MOUNT LDA59: label gives
device already allocated to another user.

$ MOUNT/CLUSTER LDA59: label, the device is mounted and can be used.

Frustrating.

[Volker Halle]

Piet,

$ ld help error devassign

talks about a possible LD trace operation, that may stil be active ?!

Volker.

[Hein van den Heuvel]

Ayup an 'allocate' will also increase the reference count, giving 'assigned channels'.

Your last reply confirms this.

So find the owning process and either shoot it, or issue a DEASSIGN LDA59: and proceed.

[Ik heb dit 'probleem' / misverstand ook gehad, en dat was het advies van Jur]

Met vriendelijke groetjes,

Hein.

[Piet Timmers_1]

$ LD TRACE/STOP
%LD-F-CONTTRACENOTACT, Continuous tracing is not active for device LDA59:


$ pipe sh log | sear sys$pipe lda59
%SEARCH-I-NOMATCHES, no strings matched

Sorry, both did not work.

ps. the sh log command i tried also with SYSMAN to execute this on all nodes in the cluster.

[Hein van den Heuvel]

Hmm,
Why do you think SHOW LOG would show it?

You still have not shown us a simple $SHOW DEV LDA59:/FULL.
Is there anything in the "Owner process ID"?
If there is, then it is allocated and can not be destroyed as the reference count is non-zero.

Hein.

[Volker Halle]

Hein,


Why do you think SHOW LOG would show it?


Because you wrote 'DEASSIGN LDA59:' to get rid of it ?!

Volker.

[Piet Timmers_1]

Bij deze:

sh dev lda59/full

Disk $1$LDA59: (NODE), device type Foreign disk type 1, is online, file-
oriented device, shareable.

Error count 0 Operations completed 671
Owner process "" Owner UIC [AP_HTTPD,APACHE$WWW]
Owner process ID 00000000 Dev Prot S:RWPL,O:RWPL,G:R,W
Reference count 2 Default buffer size 512
Total blocks 10000 Sectors per track 10
Total cylinders 100 Tracks per cylinder 10
Allocation class 1

[Volker Halle]

Piet,

this is the problem:

Reference count 2 <<<

After connecting a LD device to a container file, the LD device reference count is 0 (with OpenVMS V8.2).

Whether it's possible to find existing references or whether some bug had let to not decrementing the reference count, may be hard to find out.

I'm now left with a LD device in 'Mounted dismount alloc', allocated to a non-existing PID after playing around with devices spooled to a LD: device. After all, it's only software ;-(

Volker.

[Ian Miller.]

Volker,
try patching the UCB with xdelta to clear the owner PID :-)

You could allocate the device, patch the owner count to 1 then deallocate - as long as you don't mind risking a crash :-)

[John Gillings]

Piet,

Is LD now part of OpenVMS? If so, please log a formal report. Anything that leaves an orphan reference count is, by definition, a bug and needs to be fixed.

Even if it isn't supported by HP, I'm sure there must be a feedback address somewhere in the docs.

[Jur van der Burg]

The increased refcount prevents the driver from removing the device.

May I ask why you used /ABORT on the diconnect? (I should probably remove that qualifier as it makes no sense nowadays).

It could be an issue with the abort handling. Is it possible to force a crashdump and send me a pointer?

Thanks,

Jur (LDdriver author).

(Don't bother to log a call with HP for this, there's probably nobody with a clue about this driver. And my support is way faster :-))