topic Re: Can everyone post an backup error message? in Operating System - OpenVMS

Can everyone post an backup error message?

Victor Mendham — Thu, 22 Apr 2004 11:37:11 GMT

I am trying to collect a list of backup error messages, so that a monitoring script can be created.

So far I have..
%SYSTEM-F-ACCVIO
%SYSTEM-F-MEDOFL, medium is offline
%SYSTEM-F-DEVNOTMOUNT, device is not mounted

Anyone have any others errors or other type backup messages, like next tape needed, or please mount tape etc...

Re: Can everyone post an backup error message?

Martin P.J. Zinser — Thu, 22 Apr 2004 11:55:02 GMT

Hello Victor,

how about the following

%BACKUP-E-FATALERR, fatal error on _$2$MGA3:[]xxxxxxxx.BCK;
-SYSTEM-F-PARITY, parity error
%BACKUP-I-OPERSPEC
%BACKUP-I-OPERASSIST, operator assistance has been requested
%BACKUP-I-OPREPLY, operator reply is "QUIT"
%BACKUP-F-ABORT, operator requested abort on fatal error

Fortunatly does not happen often, but just had one of these yesterday ;-)

Greetings, Martin

Re: Can everyone post an backup error message?

Victor Mendham — Thu, 22 Apr 2004 12:41:17 GMT

Thanks, I have given you points, Thanks again.

Re: Can everyone post an backup error message?

Mohamed K Ahmed — Thu, 22 Apr 2004 12:57:45 GMT

Some more

%BACKUP-W-EXLABEER, volume label processing failed because volume TAPE4 is out of order, Volume label TAPE1 was expected specify option (QUIT, NEW tape, OVERWRITE tape, USE loaded tape)

%BACKUP-F-WRITEERRS, excessive error rate writing VOL3

%BACKUP-I-SPECIFY, specify option (CONTINUE, RESTART, QUIT)

%BACKUP-W-MOUNTERR, volume 1 on _SODAK$MUA0 was not mounted because
its label does not match the one requested

Mohamed

Re: Can everyone post an backup error message?

Jan van den Ende — Thu, 22 Apr 2004 14:36:10 GMT

Well, Victor

there _MAY_ be a more generic approach;

There is a file SYS$HELP:MSGHLP$LIBRARY:MSGHLP$DATA

This is the file from which HELP/MESSAGE takes it's info.

I haven't found a wildcard HELP command that takes out the BACKUP related messages (not been seeking very thorough though), but I trust that now you know WHERE to look, you will be able to extract the relevant data.
As far as I know, this SHOULD also be the closest you will get to completeness.

Well, I'll leave you to do the dirty work yourself!

Success

Jan

Re: Can everyone post an backup error message?

Mark Hopkins_5 — Thu, 22 Apr 2004 15:18:56 GMT

You can use help to get all the errors with
a backup facility -

$ help/message/facility=backup

$ help help/message ! for other options

Mark Hopkins

Re: Can everyone post an backup error message?

Victor Mendham — Thu, 22 Apr 2004 15:26:42 GMT

I am only concerned with Fatal type errors, not warnings.

This monitoring will page at 02:00, so I would rather not get paged on a warning item like - doing a verify and the file is different. But rather on a fatal, where the backup actually died.

I can script a simple search batch to run during the day for warning messages from logs.

Is there a way to use the help/message for fatal errors?

Vic...

Re: Can everyone post an backup error message?

John Gillings — Thu, 22 Apr 2004 16:07:25 GMT

Vic,

If you want to determine the severity of the completion status, just look at the severity bits - they're the low 3 bits of the status value (see Chapter 6 of "OpenVMS Calling Standard" for complete details):

$ some-command ! could be BACKUP
$ STATUS=$STATUS
$ SEVERITY=STATUS.AND.7

There is also the DCL symbol $SEVERITY, but it and $STATUS get set after every command, so if you want to capture the full status you need to save $STATUS immediately after the command of interest.

Severity values are:

0 = Warning
1 = Success
2 = Error
3 = Informational
4 = Fatal
5,6,7 undefined

Using severity is a much more reliable way of checking for fatal conditions than attempting to enumerate all errors (the list is potentially large and subject to change).

However, if you really want to do that, there are several ways. The simplest is to consult a source listing. But, if you don't have access to one, you can use the structure of condition values to help. A status value has several fields.

bit 0 is success/failure
bits 0-2 are severity
bits 3-15 are the message number
bits 16-27 are the facility number
bits 28-31 are control flags

So, looking at the status value for any BACKUP error, you can determine the facility code for BACKUP is %X00A30000. You can use this to determine all FATAL errors for BACKUP using F$MESSAGE:

$ msg=%XA30004 ! First possible fatal error
$ loop: WRITE SYS$OUTPUT F$MESSAGE(msg)
$ msg=msg+8
$ IF msg.LT.%XA3FFFF THEN GOTO loop

Pipe the output into SEARCH to remove all the "NOMSG" values. I get 923 possibilities on OpenVMS V7.2-2 - you won't get all of them in practice, but hopefully it illustrates why enumeration is not a good solution to your problem!

Re: Can everyone post an backup error message?

Antoniov. — Fri, 23 Apr 2004 02:47:34 GMT

Hi Victor,
%X107501F4 %INIT-F-PARITY, parity error
%X1075008C %INIT-F-DRVERR, fatal drive error
%X00720254 %MOUNT-F-VOLINV, volume is not software enabled

HTH
@Antoniov

Re: Can everyone post an backup error message?

Willem Grooters — Fri, 23 Apr 2004 03:59:11 GMT

Vic,

I would follow John suggestion. As he said - and like you may deduct from Antonio's last posting - scanning an enumeration is no good idea. Just severity would be enough to start with. However: KEEP the original status and it's meaning, you will want to know what happens. If possible - put it on the pager.
If the proces is critical, I would suggest to be informed not only in case of problems, but also if the job
- hasn't started at all
- isn't finished within a reasoable time
- has run successfully (Well, not really required but just for your ease-of-mind)
But you may have this implemented already ;-)
Willem

Re: Can everyone post an backup error message?

Jan van den Ende — Fri, 23 Apr 2004 08:45:41 GMT

Well Victor,

if you are really EXCLUSIVELY interested in FATAL errors, then your original idea might not be that bad, but (as shown by Antoniov) you can not limit yourself to BACKUP errors.

$ SEARCH "%", "-F" /EXACT/MATCH=AND

But then, is the idea of Willem not preferable?

Incluse $ ON SEVERE_ERROR THEN

Then you can even control at which parts of the backup-procedures warrant alarming and which don't.
At our site is , we rarely get paged, but IF it is needed, we ARE called.

Jan

Re: Can everyone post an backup error message?

Victor Mendham — Fri, 23 Apr 2004 09:41:27 GMT

Many Thanks to all. This is what I will probably be asking BMC Patrol to monitor.
Some of these will overlap, but that is ok.

%SYSTEM-F-ACCVIO
%SYSTEM-F-MEDOFL, medium is offline
%SYSTEM-F-DEVNOTMOUNT, device is not mounted
%BACKUP-E-FATALERR, fatal error
%SYSTEM-F-PARITY, parity error
%BACKUP-I-OPERSPEC
%BACKUP-I-OPERASSIST, operator assistance
%BACKUP-I-OPREPLY, operator reply is "QUIT"
%BACKUP-F-ABORT, operator requested abort
%BACKUP-W-EXLABEER, volume
%BACKUP-F-WRITEERRS,
%BACKUP-I-SPECIFY
%INIT-F-PARITY, parity error
%INIT-F-DRVERR, fatal drive error
%MOUNT-F-VOLINV, volume is not software

Note, I already have monitoring setup inside the cmd file, which works fairly well to e-mail as well as local mail on the server. What I will be doing here is searching the logfile afterwards & sending to a central alert system.