Shell Programming and Scripting

Hello all!

I'm a recent college grad now working for a contractor.

I've been tasked with writing a program/script that will verify that something was copied correctly.

This is how the system is used -

The user will get a distribution tape (a 4mm tape DDS3). On this tape are a bunch of files.

The user will take this tape ( all data on tape is in a tar(ball) ) - and un-tar/copy it to the disk (hard drive).

At this point my program/script should be able to verify that what was transferred from the tape is IDENTICAL to what is NOW copied to the hard drive.

So - bascially - what I'm trying to do is write a program that will check that data transmission occured without transmission errors.

I thought of using diff and cmp to compare the files but I don't know of a way to check that an un-tar operation occured without transmission errors.

Is there a way to check an un-tar operation to make sure that the data that was transferred from the tape is INDENTICAL to what is copied to the hard drive?

This command allows me to check differences between two directorys:

diff -r directory_one directory_two

How can this be used to compare what's on a tape and what's on the hard drive?

Does anyone have a better idea?
Any tips on how to start this script?

Thanks for your help!

I did a checksum program for ftp transmissions using Java MD5, but perl has an MD5 too.

I wrote one program to capture the value of checksum and write it to a text file. Then sent text file along with ftp file.

Once file arrived, I used another program to compare checksum value in text file with computed checksum at ftp site.

If checksum values are equal, you're good to go.

Using checksum is the best way to verify sent ftp files and verify files for backup.

MD5 is great, but a CRC is pretty good too. And you probably have the program cksum that can compute CRC's for you. cksum outputs two numbers: a byte count and a crc-32. I capture both numbers and check them both.

Either way, this must start before the tape is made. On the source system, you need to do a cksum on some files and send the ouput to a text file. Then you tar up your files together with this text file. Once you untar the files on the destination system, you rerun cksum and compare the results to the text file.

Thanks for the quick responses...

This is the main problem that I'm having. I understand the whole concept. Just not sure how to implement it.

The 4mm tape is created before hand. Therefore I can't create a checksum on the source data.
(actually since I haven't seen the original source code for this process I haven't been able to verify if this is true)

I can only verify the destination data.

Here's an example:

The user has a 4mm tape. On this tape is some data. The user will transfer (un tar) the data to the hard drive. The user will then use this data to create other copies onto other hard drives.

I want to verify that what's on the tape is EXACTLY what is NOW on the hard drive.

(as a sidenote - is there some kind of option that can be used within the tar command itself to verify that transmission occured error free???)

I just did a quick experiment with the cksum command.

I did:

cksum /dev/tape/tape0
It returned: some number another number /dev/tape/tape0

I was thinking I could then un-tar the contents of the tape to the hard drive and then do a cksum on the directory.

For example: cksum /temp_directory
BUT - when the command is used on the temp directory another number is returned!!! Not the same cksum value. Obviously something is happening that I'm not aware of.


Anyone have any insight to this problem?f

You're not really going to be able to do that. About as close as you can come would be to untar the tape twice and compare both disk copies.

I'm still thinking about this one!

Would any of you happen to know if the tar command ITSELF has some kind of checksum algorithm that is used when something is un-tarred?

This would be for Tru64 unix.

If the above is true then I really don't need to do a checksum routine on the tar operation, correct?

Tar does not checksum stuff. However dds drives employ a crc and if the crc is wrong, they attempt to use a ecc to correct the data. If that fails, a read error is declared.

See this link.

And this link.