Re: Metadata replication in tabled

[Pete Zaitcev <zaitcev@xxxxxxxxxx>]

On Fri, 25 Jun 2010 03:26:05 -0400, Jeff Garzik <jeff@xxxxxxxxxx> wrote:
> On 06/24/2010 08:31 PM, Pete Zaitcev wrote:

> > We essentially have a tabled that can really be considered as replicated.
> > Before, it was only data replication, which was great and all but
> > useless against disk failues in the tabled's database. I think it's
> > a major treshold for tabled.
> 
> er, huh?  In addition to data replication, we already have metadata 
> replication via db4 repmgr in tabled.git, which ensures metadata db 
> integrity in the case of disk or tabled node failure.

> Your work here, moving to the base replication API, eliminates several 
> obstacles on the path to making all tabled nodes support PUT/DELETE. 
> But it is not true to say that metadata replication did not exist prior 
> to this patch.

I think it did not work in practice. We had the code, e.g. set all the
right flags for DB operations, cleaned up all calls to make them DB_THREAD
compatible, provided all the right callbacks for the state change.
But when I tried to demonstrate the slave replication and fail-over,
it turned out that a few small parts were simply forgotten.
Most importantly, tabled would not boot right away into the Slave.
There was something in cldu.c that precluded it. Admittedly it's
possible that I broke it at some point in the past. I fixed that up
first, before running into this /etc/hosts thing.

-- Pete
--
To unsubscribe from this list: send the line "unsubscribe hail-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html