File identification problem.

[Hoff]

It appears distinctly possible somebody has a DCL symbol for the DIRECTORY command, and have gotten that wrong somehow. Post the following:

$ SHOW SYMBOL DIR

And with the Q and all, issue and post the output from the following commands:

$ DIRECTORYQ SY0:[CUP.LIVE.DAT]ITLEXT_25083.NORMAL
$ DIRECTORYQ SY0:[CUP.LIVE.DAT]ITLEXT_25083.NORMAL;1

[Dave Deacon-Holmes]

We have resolved this issue.
A system reboot untwisted it's knickers, so all post-reboot files were behaving as normal.
To fix the 'corrupted' files, I copied them to the same area with a generation of ;10, and purged them, which fixed it.

Many thanks for your time and suggestions guys, much appreciated as always.
Dave.

[Hoff]

This reeks of a file system or directory cache or volume corruption.

[John McL]

How did you identify that certain files were still corrupt after the reboot?

What exactly was the problem?

I have two reasons for asking
(a) As Hoff says, you may still have problems that will continue to play havoc with your file systems
(b) An important use of this forum is a resource to search when looking for answers to one's own problems. Recording what the problem was, no matter if it was self-inflicted, might help others avoid or rapidly resolve a similar problem in future.

[John Gillings]

Dave,

>A system reboot untwisted it's knickers, so
>all post-reboot files were behaving as
>normal.

Unlike some operating systems, OpenVMS tends to be completely deterministic, so reboots typically DO NOT change behaviour.

The 3R "solutions" (Restart, Reboot, Reinstall) don't have a place in the OpenVMS world, which is one reason that some of us inhabit it.

If you DO find a significant change in behaviour after a reboot (such as you describe), I'd strongly suspect something else is going on, and probably has NOT been fixed, just covered up. Now that you've rebooted, any diagnostic information has been lost.

For the future, if you see similar symptoms, particularly the same file or disk, I'd suggest you call in an expert to do some proper diagnosis. Alternatively, rather than just reboot, force a crash so the dump can be analysed.

[Dave Deacon-Holmes]

You are right, the reboot didn't fix anything.
The entire data area had to be rebuilt in order to receive uncorrupted files.
Still haven't got around to what broke them in the first place.

[GuentherF]

Dave,

"The entire data area had to be rebuilt" tells nothing to those who want to help you.

If you can explain a bit what that means in VMS commands/environment.

Is this node in a VMScluster?

/Guenther

[GuentherF]

And btw. this is neither a problem with OpenVMS file identification nor are these corrupted files.

It seems to be a problem with your "dil" command (I doubt you made a typo). There could be a command procedure activated for "dil" who knows. That is all I could figure out from the information you provided.

/Guenther

[P Muralidhar Kini]

Hi Dave,

>> OPS2-VISLIV $ dil SY0:[CUP.LIVE.DAT]ITLEXT_25083.NORMAL
>>
>> Directory SY0:[CUP.LIVE.DAT]
>>
>> ITLEXT_25083.NORMAL;1 2/48 6-JAN-2011 13:39:49.54
>>
>> Total of 1 file, 2/48 blocks.
>>
>> OPS2-VISLIV $ dil SY0:[CUP.LIVE.DAT]ITLEXT_25083.NORMAL;1
>> %DIRECT-W-NOFILES, no files found

This is very strange indeed.

For any file, there would be entry in two places.
One in the INDEXF.SYS file and another in the directory file in which the file
resides. If you only issue a DIR command, the filename would be read from the
directory itself. In case you also issue qualifiers such as /DATE or /SIZE and
so on... the data corresponding to this would have to be read from the
INDEXF.SYS file. As you are able to get these details above, means that the
file has entries in both the INDEXF.SYS file and directory.

From the data available, its hard to say what the root cause of the problem is.
If problem reoccurs, the data to collect for analysis would be -

* What "dil" points to ?

* In memory Data
- System crash Dump

* On disk data
- DIR/FILE
- DUMP/HEADER/ALLOC
- DUMP/DIR !to get a better formatted output

The above data would help in analysis from file system point of view.

Regards,
Murali

[Dave Deacon-Holmes]

'dil' is a symbol set up inhouse;
dil == "directory/size=all/date"