Does rdist/ssh verify that file copied OK (uncorrupted)? - Unix

This is a discussion on Does rdist/ssh verify that file copied OK (uncorrupted)? - Unix ; We maintain copies of some files using rdist with ssh as the transport. I have been asked about the integrity of the copies, i.e. are we sure they are identical. Does either rdist or ssh do any checking to verify ...

+ Reply to Thread
Results 1 to 2 of 2

Thread: Does rdist/ssh verify that file copied OK (uncorrupted)?

  1. Does rdist/ssh verify that file copied OK (uncorrupted)?

    We maintain copies of some files using rdist with ssh as the
    transport. I have been asked about the integrity of the copies, i.e.
    are we sure they are identical. Does either rdist or ssh do any
    checking to verify that the copy has been OK i.e. no corruption
    occurred etc?

    M

  2. Re: Does rdist/ssh verify that file copied OK (uncorrupted)?

    Michael B. Trausch wrote:
    > SSH won't permit corruption at the protocol level because if a
    > packet is sent and it is corrupted, SSH detects this and rejects the
    > packet and requires that it be re-sent. Essentially, SSH protects
    > the data that goes across the wire, and so it's not susceptible to
    > corruption over-the-wire like a raw bit-stream a l?? telnet, rsh, or
    > rcp would be.


    Well, telnet, rsh and rcp rely on TCP's use of the Internet Checksum.
    No it isn't foolproof but it is stronger than one might read into the
    paragraph above. I wouldn't call them raw bitstreams

    I've not delved deeply into SSH - just to be pedantic and paranoid is
    there no mode in which ssh does not put at least a hash on the data
    blocks?

    > If you really need to verify it at both sides of the wire, you can
    > always take a hash of the files on both ends of the connection and
    > verify that they are the same. md5 is good enough for file hashes
    > for such a purpose, and is widely available, so you should be able
    > to run either "md5" or "md5sum" at the terminal of either box to get
    > it to work. You can then compare the outputs and, if identical, the
    > files are identical.


    To a very good statistical certainty.

    rick jones
    --
    oxymoron n, commuter in a gas-guzzling luxury SUV with an American flag
    these opinions are mine, all mine; HP might not want them anyway...
    feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...

+ Reply to Thread