Thread: clustering and/or failover?

clustering and/or failover?

From
"Robert J. Sanford, Jr."
Date:
thinking in the long-term for my project...

i'm looking at trying to set up two database servers with
some sort of clustering and/or failover that are talking
to a SAN. is this even feasible using postgres?

using ms sql server as an example (because that is what
my office is currently using) you can set up two database
servers that will check each other's heartbeat to see if
they are both up. if one goes down then the other takes
over responding to requests. you can also set up a
cluster that will have both machines responding to
requests concurrently.

with the data set up on a fibre channel raid box you
don't have to worry about replication, you just have to
make sure that the data is properly locked by each of
the servers (in the clustering example) so that no two
machines are updating the same data at the same time.

i am fairly certain that postgres cannot handle the
first example (heartbeats) but am wondering if it can
handle the second example where two instances are
hitting the same physical data store.

many thanks!

rjsjr


Re: clustering and/or failover?

From
Tom Lane
Date:
"Robert J. Sanford, Jr." <rsanford@nolimitsystems.com> writes:
> ... two instances are
> hitting the same physical data store.

Nope.  Don't even *think* about trying that.

            regards, tom lane

RE: clustering and/or failover?

From
"Robert J. Sanford, Jr."
Date:
okay, the clustering option is out, what about a failover
option?

rjsjr

> -----Original Message-----
> From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
> Sent: Monday, August 20, 2001 7:57 PM
> To: Robert J. Sanford, Jr.
> Cc: pgsql-general@postgresql.org
> Subject: Re: [GENERAL] clustering and/or failover?
>
>
> "Robert J. Sanford, Jr." <rsanford@nolimitsystems.com> writes:
> > ... two instances are
> > hitting the same physical data store.
>
> Nope.  Don't even *think* about trying that.
>
>             regards, tom lane
>

Re: clustering and/or failover?

From
"Jim Buttafuoco"
Date:
Robert,

I have been running a "heartbeat" PG database cluster for a year now
using a shared SCSI bus.  Yes the heartbeat software has to be careful
about when to mount the disk and on what system.   But this seems to
work well (I did a lot of testing before putting into production.)  This
is a pure master/standby cluster config.

Jim


> thinking in the long-term for my project...
>
> i'm looking at trying to set up two database servers with
> some sort of clustering and/or failover that are talking
> to a SAN. is this even feasible using postgres?
>
> using ms sql server as an example (because that is what
> my office is currently using) you can set up two database
> servers that will check each other's heartbeat to see if
> they are both up. if one goes down then the other takes
> over responding to requests. you can also set up a
> cluster that will have both machines responding to
> requests concurrently.
>
> with the data set up on a fibre channel raid box you
> don't have to worry about replication, you just have to
> make sure that the data is properly locked by each of
> the servers (in the clustering example) so that no two
> machines are updating the same data at the same time.
>
> i am fairly certain that postgres cannot handle the
> first example (heartbeats) but am wondering if it can
> handle the second example where two instances are
> hitting the same physical data store.
>
> many thanks!
>
> rjsjr
>
>
> ---------------------------(end of
broadcast)---------------------------
> TIP 2: you can get off all lists at once with the unregister command
>     (send "unregister YourEmailAddressHere" to
majordomo@postgresql.org)
>
>



RE: clustering and/or failover?

From
"Robert J. Sanford, Jr."
Date:
what software are you running for this? where can i find it?

thanks!

rjsjr

> -----Original Message-----
> From: pgsql-general-owner@postgresql.org
> [mailto:pgsql-general-owner@postgresql.org]On Behalf Of Jim
> Buttafuoco
> Sent: Tuesday, August 21, 2001 6:53 AM
> To: Robert J. Sanford, Jr.; pgsql-general@postgresql.org
> Subject: Re: [GENERAL] clustering and/or failover?
>
>
> Robert,
>
> I have been running a "heartbeat" PG database cluster for
> a year now using a shared SCSI bus.  Yes the heartbeat
> software has to be careful about when to mount the disk
> and on what system.   But this seems to work well (I did
> a lot of testing before putting into production.)  This
> is a pure master/standby cluster config.
>
> Jim
>
>
> > thinking in the long-term for my project...
> >
> > i'm looking at trying to set up two database servers with
> > some sort of clustering and/or failover that are talking
> > to a SAN. is this even feasible using postgres?
> >
> > using ms sql server as an example (because that is what
> > my office is currently using) you can set up two database
> > servers that will check each other's heartbeat to see if
> > they are both up. if one goes down then the other takes
> > over responding to requests. you can also set up a
> > cluster that will have both machines responding to
> > requests concurrently.
> >
> > with the data set up on a fibre channel raid box you
> > don't have to worry about replication, you just have to
> > make sure that the data is properly locked by each of
> > the servers (in the clustering example) so that no two
> > machines are updating the same data at the same time.
> >
> > i am fairly certain that postgres cannot handle the
> > first example (heartbeats) but am wondering if it can
> > handle the second example where two instances are
> > hitting the same physical data store.
> >
> > many thanks!
> >
> > rjsjr
> >
> >
> > ---------------------------(end of
> broadcast)---------------------------
> > TIP 2: you can get off all lists at once with the
> unregister command
> >     (send "unregister YourEmailAddressHere" to
> majordomo@postgresql.org)
> >
> >
>
>
>
> ---------------------------(end of
> broadcast)---------------------------
> TIP 5: Have you checked our extensive FAQ?
>
> http://www.postgresql.org/users-lounge/docs/faq.html
>

RE: clustering and/or failover?

From
"Jim Buttafuoco"
Date:
http://linux-ha.org/ look for heartbeat.


> what software are you running for this? where can i find it?
>
> thanks!
>
> rjsjr
>
> > -----Original Message-----
> > From: pgsql-general-owner@postgresql.org
> > [mailto:pgsql-general-owner@postgresql.org]On Behalf Of Jim
> > Buttafuoco
> > Sent: Tuesday, August 21, 2001 6:53 AM
> > To: Robert J. Sanford, Jr.; pgsql-general@postgresql.org
> > Subject: Re: [GENERAL] clustering and/or failover?
> >
> >
> > Robert,
> >
> > I have been running a "heartbeat" PG database cluster for
> > a year now using a shared SCSI bus.  Yes the heartbeat
> > software has to be careful about when to mount the disk
> > and on what system.   But this seems to work well (I did
> > a lot of testing before putting into production.)  This
> > is a pure master/standby cluster config.
> >
> > Jim
> >
> >
> > > thinking in the long-term for my project...
> > >
> > > i'm looking at trying to set up two database servers with
> > > some sort of clustering and/or failover that are talking
> > > to a SAN. is this even feasible using postgres?
> > >
> > > using ms sql server as an example (because that is what
> > > my office is currently using) you can set up two database
> > > servers that will check each other's heartbeat to see if
> > > they are both up. if one goes down then the other takes
> > > over responding to requests. you can also set up a
> > > cluster that will have both machines responding to
> > > requests concurrently.
> > >
> > > with the data set up on a fibre channel raid box you
> > > don't have to worry about replication, you just have to
> > > make sure that the data is properly locked by each of
> > > the servers (in the clustering example) so that no two
> > > machines are updating the same data at the same time.
> > >
> > > i am fairly certain that postgres cannot handle the
> > > first example (heartbeats) but am wondering if it can
> > > handle the second example where two instances are
> > > hitting the same physical data store.
> > >
> > > many thanks!
> > >
> > > rjsjr
> > >
> > >
> > > ---------------------------(end of
> > broadcast)---------------------------
> > > TIP 2: you can get off all lists at once with the
> > unregister command
> > >     (send "unregister YourEmailAddressHere" to
> > majordomo@postgresql.org)
> > >
> > >
> >
> >
> >
> > ---------------------------(end of
> > broadcast)---------------------------
> > TIP 5: Have you checked our extensive FAQ?
> >
> > http://www.postgresql.org/users-lounge/docs/faq.html
> >
>
>