[OpenAFS] Doubts about OpenAFS implementation in a company

Harald Barth haba@kth.se
Wed, 18 May 2011 17:46:51 +0200 (CEST)

I doubt if I would have any RW volues at the place with the 10/1 link
as one of the directions is so much more painful to move data over,
unless I can guarantee that the data in that volume is _exclusively_
used by that office.

> - every office has it's own db and fileserver (Debian 5/6)

How many are there in total? I would only have 3 "main" db servers
that sync with each others and the others at the weakest locations as
clones (see email by Hartmut).

Do you see that fragments of UDP packets are dropped? In that case the
server has to resend all fragments unless RX_MAX_FRAG=1 (change in
source code). Someone who knows more about rx than I do can help
me with the diagnosis here.