Raw structs as network packets (was: Re: svn commit: samba r20943 - in branches/SAMBA_4_0/source/cluster/ctdb/common: .)

[tridge at samba.org]

Andrew,

 > At linux.conf.au, you were discussing the need for a wireshark decoder
 > for this protocol.  How should this handle guessing the protocol,
 > without any clue as to host endian?

The first word is an operation, so you could get byte order from that
in wireshark. Not ideal, but it would work.

 > Talking with the SGI folks, they are worried that long term, when real
 > customers deploy this, they will mix architectures, and even program
 > versions.  Even if we don't support this (and I think we should at least
 > try to, for sane combinations), we should at least ensure the protocol
 > prevents it...

the problem with mixed endianness clusters isn't the ctdb protocol
itself. Making that work would be not hard. The problem is the data
that is put into the databases. Nearly all our existing tdb databases
contain binary data, and are just arrays of structures. Fixing all of
those to use encoded structures is a _much_ bigger job.

So for now I'm going to ignore this. If/when this becomes a problem
then we'd have to start by going through all uses of tdb and change
them all to use NDR or similar encoding for all structures. After that
is done we can worry about the half dozen structures in the ctdb
protocol.

Cheers, Tridge