Validate integrity

Hi.

I have some .tib on a harddisk which became partially corrupt. Some .tib files show validate errors, some don't.

I wonder if those who don't show these validate errors, are 100% integer? Does a successful validate mean that the .tib file is 100% correct, bit by bit?

Thanks,

 

When ATI creates a tib, it does checksum every so often and embeds the checksums in the file. When it does a "validation", it reads the file data and the checksums and recomputes the checksums to see if the checksums match -- if any checksum doesn't match, then the file is declared "corrupt."

It's possible for a file to not be corrupt and be declared corrupt by ATI, for example, if you try to validate with the bootcd and it can't properly read the drive the tib is on.

There is no comparison to the original source data but a validation can be a good indicator that no bytes have changed since the file was written.

 

If you can't validate an archive it means that TI can't read it properly for some reason. As sheiber pointed out, a successful validate means that TI can read it properly and nothing has changed in the file(s).

Do you get validate errors on the same .tib file each time or do the occur randomly? If they always appear on the same file then it sounds like you might have a bad area or areas on your disk. If it is random then it could be a disk hardware error or a bad RAM location or even a marginal surface problem where the magnetic bits are weak. Bad RAM can cause the checksums be recalculated improperly. If you have an overclocked PC or overly aggessive memory timings to boost performance, these things can cause problems with TI but not manifest themselves with other applications.

If it only happens with the TI rescue CD version then it is possible there is a Linux driver problem but this isn't too likely since the problem is intermittent.

 

If you have the opportunity to test the tib on another machine and it validates, then it's most likely a file read problem -- an inability of ATI to read the file on your hardware. If ATI is the only program you have that doesn't work wotih your hardware, you have to decide if you want to spend a lot of time making your hardware work with ATI or getting ATI to work with your hardware or just try another brand.

ATI often has probs being able to validate tibs on external drives often enough that I had to stop recommending it to clients.

Note that if ATI has trouble writing to, or reading from a drive, it's unlikely that it will be able to restore from that drive. You can also try copying the file to the source drive and see if it validates there to narrow down if its an issue related just to the one drive.

 

Why should the bootcd not be able to read the .tib file properly. Since the CRCs are embedded in the file, the bootcd validate result should be of the same value as the WinXP ATI validate function.

In other words ... if the bootcd can read the drive the .tib is on, it should be able to validate it or not. This result should not be dependent on whether the file is checked via the bootcd, or via the WinXP ATI interface.

Thanks,

 

The problem is always the same file at the same place. No radom factor.

It's a high reliability/availability server. RAMs are tested regularly (memtest86).

It happens with the bootcd as well as with the WinXP ATI interface.

Since I work on test with +2 TB arrays and backup was done onto a partition beyond this limit, I was wondering if this could have been the problem. Unfortunately, this is all testing, since there are no experience values out on the internet regarding this issue. But this is offtopic for the thread.

Anyway ... my initial question was in fact, if the validate mechanism really validates backupped data in terms of CRCs, or if it "only" checks the correct file structure of the .tib, not detecting potential data integrity errors (which would result in wrong data being restored without any notification).

Thanks,

 

The bootCD does not use windows, it uses linux and linux drivers and these might not be adequate for your hardware even though the windows version (on your harddisk) works fine (with windows drivers). But you can make a windows boot disk with ATI on it and it should be able to run on your hardware if ATI can works under windows.

 

Thanks for the info.

To make it short ... this was my issue ...

As far as I understood ... it's the CRC option. Hence, every bit is checked being checked against a CRC. If not multiple errors compensate each other, a correct validate is guarantee that the .tib file did not become corrupt after file generation.

Thanks,

 

Yes, TI writes a checksum into the archive for every 256K bytes of data which results in 4000 checksums per gigabyte. It only takes one failed checksum to have the archive declared corrupt.

 

Thanks for the replies!

 

If it can't read properly, it's not going to get the right bits and then the chcukc sum calc will come up with a diff answer -- or the read errors them selves will force ATI to decelare the file invalid.