Обсуждение: PostGreSQL Replication question !!

Hello,
I have got several questions about replication with PostGre:

1 : Is the project RServ (Open source project of eRServer)still
existing ? or is there only eRServer ? I can't fin sources and help
with RServ do you know a good link ?


2 : Is there 'dynamic' free projects on Replication now ? (PG
Replicator, RServ... seems to be a little bit stopped or finished)


3 : What is the best way to make PostGre Replication (from two
different hosts on a same network)(with a free way)?
Use PG Replicator ? or PostGres-R ? RServ?


Thanks by advance
Cordially
Eric

On Fri, Feb 07, 2003 at 06:21:16AM -0800, Eric Cros wrote:
> 1 : Is the project RServ (Open source project of eRServer)still
> existing ? or is there only eRServer ? I can't fin sources and help
> with RServ do you know a good link ?

The rserv code is still under contrib/ in the 7.3.x tarballs, as far
as I know.

> 2 : Is there 'dynamic' free projects on Replication now ? (PG
> Replicator, RServ... seems to be a little bit stopped or finished)

rserv is the little cousin of eRServer, which is still a commercially
supported project.  As far as I know, however, nobody is working on
the free code.

dbmirror is also in contrib/.  It's free.  Some people have had good
luck with it, I understand.  I believe it is still being developed.
I don't know that there's a lot of development to do.  It uses the
primary key on a table, and multiple-column primary keys are not
supported AFAIK.  I'll bet you could modify the code to use some
other single-column unique-indexed not null field.  I haven't tried.

> 3 : What is the best way to make PostGre Replication (from two
> different hosts on a same network)(with a free way)?
> Use PG Replicator ? or PostGres-R ? RServ?

As far as I know, Postgres-R isn't ready for production use.
Experiments with it, and reports of success and failure, would surely
be welcome.

But is your question phrased correctly?  You say "from two different
hosts on the same network".  Does that mean you want two master
databases, with one slave shared?  Or do you mean two machines, one
master and one slave?  The former may be impossible with anything at
the moment.

A

--
----
Andrew Sullivan                         204-4141 Yonge Street
Liberty RMS                           Toronto, Ontario Canada
<andrew@libertyrms.info>                              M2P 2A8
                                         +1 416 646 3304 x110

On Mon, 17 Feb 2003 13:20:08 -0500 Andrew Sullivan <andrew@libertyrms.info> wrote:
> dbmirror is also in contrib/.  It's free.  Some people have had good
> luck with it, I understand.  I believe it is still being developed.
> I don't know that there's a lot of development to do.  It uses the
> primary key on a table, and multiple-column primary keys are not
> supported AFAIK.  I'll bet you could modify the code to use some
> other single-column unique-indexed not null field.  I haven't tried.

i'm about to start working on using dbmirror for a project. i'd say that in
its current state, it's not complete, but it can be persuaded to serve in a
useful way. it lacks conflict resolution, and it currently replicates one
master to one slave (but can be coached into something more comprehensive.)
it also makes assumptions about the initial states of the databases
involved replication.

richard
--
Richard Welty                                         rwelty@averillpark.net
Averill Park Networking                                         518-573-7592
              Unix, Linux, IP Network Engineering, Security

> i'm about to start working on using dbmirror for a project. i'd say that
in
> its current state, it's not complete, but it can be persuaded to serve in
a
> useful way. it lacks conflict resolution,

Agreed, but it's meant to be one way, so...

> and it currently replicates one
> master to one slave (but can be coached into something more
comprehensive.)

Actually it handles multi-slave easily... that's why there's a "MirrorHost"
table. I'm currently replicating out to two slaves with no modifications.

> it also makes assumptions about the initial states of the databases
> involved replication.

Yes, which makes it a true pain in the ass to bring another slave online.

Greg

Doesn't mysql have open source, free replication? If so, it wouldn't hurt to look at the code :-)

> Doesn't mysql have open source, free replication? If so, it wouldn't hurt
to look at the code :-)

1. MySQL is not under a BSD license, meaning you can't just grab their code
and put it into PostgreSQL.
2. What makes you think that one replication solution for one database would
work with a completely different database? Particularly since they do some
form of binary replication, I would think that's just not an option.
3. I believe the replication projects in the works are far more ambitious
than just copying those features.
4. There is no number 4.

Greg