Shell Programming and Scripting

Hey guys.

I'm knocking up a script, a part of which will be copying some pretty large files to some remote servers. With the size of the files I need to make sure that they don't lose any data / corrupt on the way (it's unlikely, but these are business critical).

Obviously I could use the exit code to confirm the copy completed successfully but I assume that it wouldn't know if any data was lost.

The option I've thought of is to do a checksum on the file being copied, and do a remote checksum once the copy is completed and compare the two. But was just wondering if anyone could think of a neater way of doing it.

Thanks.

Of course it would! Actually this is what the so-called "flow control" of the network layer is for - make sure that no packet is lost or at least make sure such a loss is noticed.

If you do some "scp /file/here user@host:/file/there" and it comes back with RC=0 you are guaranteed that the transmission was: 1) successful; 2) complete.


If you are really paranoid (which is, btw., a fitting attitude for a systems administrator and there is absolutely nothing wrong about it) you could of course check the checksums of both the local file and the remote copy.

There is a freeway utility called "md5sum" (download here), which is included in many Linux distributions. Perhaps it can be compiled for most other Unix systems too. It should be well suited for your purpose.

I hope this helps.

bakunin

Ah right, thanks for that. I assumed the return code would be that the copy command had completed successfully, not that the data copied was duplicated (if that makes sense).

Might do a basic cksum check as well just to be sure. But sounds like that'll be enough.

Thanks,