[Nottingham] Strange ssh errors - Corrupted MAC on input.

Mike Hemstock mike at csits.net
Mon Sep 12 15:44:26 BST 2005


On Sunday 11 September 2005 14:54, Simon Amor wrote:
> I've been trying to copy data between two servers, one that I know is
> working fine and the other which was freshly installed yesterday.
> Using rsync over ssh, I'm getting the following error messages after
> transferring anywhere between 10MB and 1000MB. I have around 50GB to
> copy and don't really want to have to restart rsync every few minutes
> as that's just a waste of time.
>
> On the old server, whilst rsyncing to the new server:
>
> Received disconnect from UNKNOWN: 2: Corrupted MAC on input.
> rsync: writefd_unbuffered failed to write 4 bytes: phase
> "unknown" [sender]: Broken pipe (32)
> rsync: connection unexpectedly closed (463060 bytes received so far)
> [sender]
> rsync error: error in rsync protocol data stream (code 12) at io.c(434)
>
> The reason the subject says ssh errors not rsync errors, is that I've
> been ssh'd into the new server from home and had exactly the same
> thing appear in the middle of my session at which point it closed the
> connection on me. The "Corrupted MAC on input" message appears to be
> the important part.
>
> Both servers are Fedora Core 3 and google'ing for the error has given
> me plenty of people with the same problem, but never any solutions.
> The network port on the new server has been plugged into a new switch
> port with a new cable but still results in the same error. The main
> difference between the two is that the old one runs
> 2.6.11-1.27_FC3smp and the new one runs 2.6.12-1.1376_FC3smp
>
> Unfortunately I can't find the 2.6.11 rpms any more so can't even try
> using the same kernel on both. The servers are DELL boxes with 1GB
> and 2GB RAM (new and old) and a pair of 160GB SATA drives each. I
> tried upgrading the old server to 2.6.12-1.1372_FC3smp (the release
> before the current one) and it failed to find the drives on bootup
> which is why it's on 2.6.11 still.
>
> Any suggestions on how to resolve this - either a software fix I can
> apply or some way to prove that the hardware is faulty - would be
> greatly appreciated.
>
>     Simon

I've seen this problem before.  It turned out to be a faulty port on my 
firewall.

Mike.



More information about the Nottingham mailing list