Disk integrity check

[Vladimir_75]

Hi,
Here is an interesting one for a gurus.
Currently running critical 24X7 application, was discovered, that some of the latest *.exe files in the directories gone missing (shouldn't be a user or other person interaction) cause during an application build all files been pushed to a folder location, and worked perfectly up to yesterday.
Today, been noticed, after application crashed,
that "some" of the files got missing from a folder, or having an older version number, other then "all of the files" in the folder should contain,
Here is a deal, user interaction, is not the case, cause del *.*; would delete, for exmpl, all the latest ones, not the particular ones, there is no other way, users could delete 1000+ files, and system is running off-line (no internet or network connect), nor applications can delete those
So, I suspect, there should be a bad block(s) on the drive, but really do not know, how to check for this.
Sorry, I'm a newbie in OpenVMS system administration, so I'd appreciate any help

Thanks a lot.

[Robert_Boyd]

Vladimir,

There are a couple of approaches I might suggest for investigating what is happening on the disk with the files that are missing.

The first is an integrity check as you suggest. The standard way to do this is to login as SYSTEM and do


$ analyze/disk/repair :/lock/noconfirm

Then the course of action from there will depend on the messages (or none) generated by this utility.

Another way to do the same thing is to install the latest version of the DFU freeware utility and use the command

$ dfu verify :/fix/lock/direct

Another approach is to assume that there may be something on the system that is accidentally or otherwise deleting files or removing directory links to files. In this case it would be useful to place alarm access control lists on the directories for WRITE access and possibly the files involved for DELETE access. You will also need to enable the correct audit logging to signal any such access to OPCOM and/or the audit log. You would do well to review the Guide to Security to fully understand how to do this and the impact it might have on the system.

You may want to do the basic integrity check of the disk and report back what you find.


Robert

[Vladimir_75]

Let me check that commands and I get back to you tomorrow. Thanks a lot again.

[Antoniov.]

Vladimir,
welcome to VMS forum!

Because you have a 24x7, I guess you cannot halt it so be careful when you type ANALYZE/DISK with /REPAIR and /LOCK options.
In running system, /REPAIR may display unexistent errors due to buffer workload of system.
/LOCK option is not avaiable on V6.x of VMS. This qualifier lock disk while analyze is running to prevent wrong messages.
Use help for more details:
$ HELP ANALYZE/DISK

Antonio Vigliotti

[Vladimir_75]

Thanks a lot, I'll check those out as well.

[Andy Bustamante]

At a less intrusive level use

$ show error

and see if the suspect disk is listed. If the error count is steadily climbing you'll need to replace the disk as soon as possible.

What version of VMS and platform are you running?

[Veli Körkkö]

What can you tell us about the environment in generally (like VMS version etc) and about the application specifically?

I do remember one particular case where files got deleted mysteriously after patch upgrade. We
had (I think, after all it is close to ten years ago) VAX/VMS V6.2, ALL-IN-1 V2.4 (maybe).

ALL-IN-1 has forms and their functions typically in form libraries called typed .FLB and for performance these libraries plus other stuff were sort of precompiled to have an ...FLC. In those precompiled libraries all scripts apparently had function names or such replaced with sort of goto to a table index.

Unfortunaly with installed patch the ALL-IN-1 internal function list changed with DELETE_FILE moving upward to somewhere else
and whilst during patch installation all standard precompiled libraries were precompilied again, the add on application specific ones were not.

Hence first user of one particular menu entry
deleted a few key files, all subsequent users getting error "File not found"

_veli

[Vladimir_75]

Hi,
I've checked the diskj using analyze command as well as show error an it didn't show any errors.
This is certainly not a software issue, cause 2 VMS boxes (OpenVMS 8.0) are dual redundant, and I'd expect to have the same or similar effect on another one, but this is not the case.

[Antoniov.]

If you didn't see any error, I suspect somebody deleted files.
May you disable deleting files of your user?
If this is possible without trouble, you can remove delete flag from files; obviously, no user can be privilegiated user.

You can also enable audit on delete files, but warning, in this way auditing file may growth up. If you want to enable auditing on delete type
$ SET AUDIT /AUDIT-
/ENABLE=(ACCESS=(SUCCESS:DELETE))-
/CLASS=FILE
Then you can analyze deleting with
$ ANALYZE/AUDIT/SINCE=

Antonio Vigliotti

[Veli Körkkö]

Are you sure about the VMS version? You specified V8.0 but that did existed only for Itanium/integrity machines and even then it was pretty much restricted evaluation version.

I would rather suspect somekind of software or operator/user issue rather than hardware issue.

I would expect the disk filesystem corrupted rather than losing files from here and there.

_veli

[Vladimir_75]

Hi,
No, this is certainly a version 8 of OpenVMS.
About software, got debugging flag on, so no messages about file operations logged, and operator error:
person cannot go and delete particular files in the directory (for exmpl. 14 files missing from 178 in the directory total). No wildcard on files can be used (different names) so he'd go and do:
del a;
del b;
del c;
for exmpl... etc...

I really do not believe it, plus our operators does not have a training in OpenVMS console command, only a GUI interface.

[Steven_101]

This might be an ugly way to track down whats happening, but consider a batch job to sho user, sho que/all/batch and directory listing every 1/2 hr for 1 day. This might help you determine when and how the files are getting deleted. If its the same files all the time then someone could have a batch job doing it or some cleanup