Wednesday, September 24, 2008

Re: [HACKERS] Hot Standby Design

On Wednesday 24 September 2008 03:27:44 Simon Riggs wrote:
> On Wed, 2008-09-24 at 00:30 -0400, Robert Treat wrote:
> > here are some scattered thoughts i had while reading it :
> Thanks for your comments.
> It is very detailed, so further feedback is going to be very beneficial
> in making this all work in the way we hope and expect.
> > * "However, some WAL redo actions will be for DDL actions. These DDL
> > actions are repeating actions that have already committed on the primary
> > node, so they must not fail on the standby node. These DDL locks take
> > priority and will automatically cancel any read-only transactions that
> > get in their way."
> >
> > Some people will want the option to stack-up ddl transactions behind
> > long-running queries (one of the main use cases of a hot slave is
> > reporting stye and other long running queries)
> Scheduling tools can help here. Run set of queries on Standby, then when
> complete run DDL on Primary and have its changes filter through.

true... i am just reminded of Oracle 8's log replay, where you had to stop
replay to run any queries... given that was a usefull feature, I suspect
we'll hear complaints about it not going that way. I think one could argue
that we might be able to provide such an option in the future, if not in the
first release.

> > * Actions not allowed on Standby are:
> > DML - Insert, Update, Delete, COPY FROM, Truncate
> >
> > copy from suprised me a bit, since it is something i could see people
> > wanting to do... did you mean COPY TO in this case?
> I checked...
> COPY TO is allowed, since it extracts data. So pg_dump will run.
> COPY FROM will not be allowed since it loads data. So pg_restore will
> not run.

ah, of course... the late hour had me thinking backwards. pg_dump on the
clone will be a big plus.

> > * Looking for suggestions about what monitoring capability will be
> > required.
> >
> > one place to start might be to think about which checks in check_nagios
> > might still apply. Possibly also looking to systems like slony for some
> > guidence... for example
> check_pgsql will work, but its very simple.
> check_postgres will also work, though many some actions could give
> different or confusing results. e.g. last vacuum time will not be
> maintained for example, since no vacuum occurs on the standby.


> slony won't run on the standby either, so those checks won't work
> either.
> > everyone will want some way to check how far the lag
> > is on a stnadby machine.
> Agreed

right... my thought with slony was, what do people monitor on thier slony
slaves? (there is actually a nagios script for that iirc)

> The Standby is a Clone and not a Slave. A Slave is a separate database
> that is forced to duplicate the actions of the Master. The Standby is an
> exact copy, in every detail that matters.

This is an interesting clarification.

Robert Treat
Build A Brighter LAMP :: Linux Apache {middleware} PostgreSQL

Sent via pgsql-hackers mailing list (
To make changes to your subscription:

No comments: