Re: "Corrupted MAC on input"
From: Matt Howard (securityfocus_at_hattmoward.org)
To: firstname.lastname@example.org Date: Fri, 2 Jan 2004 20:12:42 -0500
-----BEGIN PGP SIGNED MESSAGE-----
The MAC in a ssh packet is similar to a CRC checksum. I usually see errors
like this crop up on systems with corruption problems in the NIC buffers.
The packet was received OK and checksummed OK, but when the OS reads the data
out for it's own use, the data is corrupt. If you're using an Intel 10/100
nic, I would try switching to the other network driver ( eepro100 vs. e100 )
I've had this issue with Intel's driver, specifically. Anyway, this is not a
definitive answer, just what I surmised from investigation. I'd like to see
an update on the issue, to see if we find a common thread.
Matt Howard <mhoward__hattmoward.org> http://hattmoward.org
PGP: 2364 B8D1 199C 4DB6 9E23 2409 C10B EC85 29B9 91EE
On Wednesday 31 December 2003 4:35 am, Keith Edmunds wrote:
> I am trying to backup a 2.5Gb dataset over an ssh connection, and I am
> seeing this error after around 1Gb of the transfer:
> Received disconnect from [destination IP address]: 2: Corrupted MAC on
> I have searched the archives and Googled for this, but other than a
> mention of a possible problem with Linksys routers I have found nothing
> to help. There is no Linksys router involved in this setup, but there is
> a cable modem (Pace: Model DITV-2000, Part No: 541-4408400).
> The backup software is rdiff-backup, but this merely uses ssh as a
> transport. I don't believe that rdiff-backup is contributing to this
> problem, but I'm happy to take any steps necessary to prove or disprove
> My question is: what can I do to provide more information on this
> Thanks (and happy new year!)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)
-----END PGP SIGNATURE-----