A few questions on audit volume?

[ldmwndletsm]

I ran an `audit volume fix=no` on several tapes as a test. No discrepancies were reported, but I have several questions:

1. Does the audit actually read each file on the tape, just as a recover operation would, or does it instead read some type of metadata on the tape to compare with the database?

In other words, if you were restoring data to the backup server from the same tape, and the tape drive was local (no network) then aside from the time to write to disk would the speed to read the data from tape be the same as the audit's reading of the data? Is that part of the operation the same?

2. What happens if you audit volumeA after you move the data to volumeB?

I would assume that if no newer data was written to volumeA (read-only) then TSM is no longer tracking any data for tape A. So a `q content volumeA` would report nothing. Data is still physically on the tape, of course, but TSM no longer knows about it being there, only now volumeB. But I wonder about an audit? Would it:

A. Fail immediately?

B. It would try to read the tape (surely it could read it), but then it would report an error for each file since that file would not be listed in the DB for volumeA?

C. It would run silently and then end with the usual summary report wherein everything would then be reported as missing or some such thing?

I meant to try this after data was moved to another tape, but the volume was returned to the scratch pool (resuse delay for the stgpool expired) and recycled and then written to before I thought about it.

 

[marclant]

1. Does the audit actually read each file on the tape, just as a recover operation would, or does it instead read some type of metadata on the tape to compare with the database?

It reads the data. It can't validate the integrity of the data, but it can validate that the data is readable from the tape, meaning it's not damaged.

2. What happens if you audit volumeA after you move the data to volumeB?

None of the above. D Because A is empty is empty, the audit finishes successfully in a matter of seconds.

 

[ldmwndletsm]

Thank you But I'm a bit confused about D. The tape is not physically empty since the data was copied, not wiped, so it's still physically on the tape, just as it was, never mind the fact that TSM no longer lists that data in its database for tape A. Does the move command write a double end of file mark to the beginning of the tape once it's done? If so, most tape drive firmware will not read past that, so then I could see that an audit would just stop, nothing to read. It's done. Otherwise, why would moving the data to another tape prevent the audit from still reading the data on the source tape? Does it even load the tape in a drive? Or does it check with the database, and the DB tells it don't even bother, there's nothing on that tape? Something like that?

 

[marclant]

That doesn't matter. During the audit, the server is looking for what it knows exists on the tape, and after the move data, it knows there's nothing on the tape. As far as the server is concerned, the tape is empty, regardless what is physically on the tape.

 

[marclant]

Basically, if you do a Q CONTENT TAPE1, then do AUDIT VOLUME TAPE1, the server will look to make sure that it can read all the files listed in Q CONTENT. So if Q CONTENT is empty, the audit doesn't need to read anything.

 

[ldmwndletsm]

Okay, I guess that makes sense. I was thinking that *maybe* the audit was like a scan wherein it would read everything on the tape that's physically there. So does it even load the tape? Seems it would be unnecessary except perhaps it would need to verify that that is in fact the correct tape so it would have to load it to read the label? One it's read the label, and its internal q content reports nothing (pretty quick), it's done?

Does it matter if you run: `q content volume1` and then `audit volume volume1` VERSUS just: `audit volume1`?

 

[marclant]

So does it even load the tape?

Not sure, but probably not since there's no files that it needs to read.

Does it matter if you run: `q content volume1` and then `audit volume volume1` VERSUS just: `audit volume1`?

You don't need to, it was an example to tell you in easy terms what the server is looking for on the tape.

 

[ldmwndletsm]

Right, got it. Thanks, again for all your help

 

[rodelgad]

It reads the data. It can't validate the integrity of the data, but it can validate that the data is readable from the tape, meaning it's not damaged.

None of the above. D Because A is empty is empty, the audit finishes successfully in a matter of seconds.

Hi Marclant,

It reads the data. It can't validate the integrity of the data, but it can validate that the data is readable from the tape, meaning it's not damaged.

None of the above. D Because A is empty is empty, the audit finishes successfully in a matter of seconds.

Hi Marclant,
To verify this point, the Audit Command does the verification at the object level; therefore, it can indicate which object is available on a particular volume and checks objects against the database. Is this correct? and if so, would that be considered a check for data integrity?

 

[marclant]

Hi Marclant,

Hi Marclant,
To verify this point, the Audit Command does the verification at the object level; therefore, it can indicate which object is available on a particular volume and checks objects against the database. Is this correct? and if so, would that be considered a check for data integrity?

Only if the storage pool is set with CRCDATA=YES. Otherwise, the server has no info to compare if the file is the same as the original, it can only check if it can read the file from start to finish, but not if it's unchanged. The default for CRCDATA is NO.

 

[Rosario]

Does the CRCData command have to be used only when a storage pool is created or can the CRCDATA be activated to yes before an AUDIT command is run on a storage pool?

 

[marclant]

It has to be enabled before data is backed up. So if you turn it on today, backups before today would not have CRCDATA stored with it, so an audit would not be able to verify the CRCDATA. All backups done after today would and the audit could validate the CRCDATA on those in the future.