Raw structs as network packets
alisonw at sgi.com
Tue Jan 30 05:28:16 GMT 2007
Sorry for the slow reply, i've been on vacation post-linuxconf.
tridge at samba.org wrote:
> abartlet at samba.org wrote:
> > 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.
This is true, but from what i understand of the ctdb proposal you're
initially only planning on ctdb-ifying databases that are transient.
When the cluster membership changes, all the locking and whatnot will be
renegotiated over the network anyway. If the network comms part is
architecture-independent, then it shouldn't matter how individual nodes
are storing their data.
With regard to the non-transient databases like user databases and so
forth, i suspect a useful implementation of clustered Samba would be
best handled by requiring shared user and password backends (NIS, LDAP,
ADS) anyway. I'm not sure how many other non-transient databases would
need to be shared by different nodes in this instance?
While we're on the topic, is there a list of All Possible TDBs that get
created and a quick oneliner description of what they do?
More information about the samba-technical