Re: Scriptable way to validate a pg_dump restore ?

On Tue, 30 Jan 2024, 05:02 Adrian Klaver, <adrian(dot)klaver(at)aklaver(dot)com> wrote:

> On 1/29/24 11:35, Shaheed Haque wrote:
> >
> >
> > On Tue, 30 Jan 2024, 00:27 Adrian Klaver, <adrian(dot)klaver(at)aklaver(dot)com
> > <mailto:adrian(dot)klaver(at)aklaver(dot)com>> wrote:
> >
> > On 1/29/24 10:12, Shaheed Haque wrote:
> >
> >
> > >
> > > Yes. But I was under the impression that the initial copy of
> logical
> > > replication was the same?
> > >
> >
> > Are you taking about the copy_data option to WITH?
> >
> > If so yes and no.
> >
> > Yes as it uses COPY to transfer the data.
> >
> >
> > Yes, this is what I meant.
> >
> > No as what COPY transfers can be affected by WHERE clauses on the
> > publisher. Also if you have cascading publishers/subscriptions the
> > 'original' data maybe upstream of the publisher you are comparing to.
> >
> >
> > Good points, understood. For the next bit, let's assume neither of these
> > are in play.
> >
> > Finally logical replication is generally not static so there is the
> > issue of determining a point in time for the check.
> >
> >
> > Indeed. I currently have a static source db but would eventually like to
> > eliminate the implied downtime. What I'd like to provide my user is some
>
> Implied downtime of what?
>

Oh, forgot to say: the downtime of my Django based app. Not anything at the
PG level.

> > indication of progress initially during the copy_data phase, and for the
> > future, of the anticipated incremental convergence.
> >
> > And, as per my other note, I would ideally like to be able to do this
> > using only a connection to one db.
> >
> > I was assuming that logical replication needed "something" similar
> > internally, and was hoping the LSNs were that "something".
>
> I'm going to say up front I am no expert on the internals of logical
> replication. Will point you at:
>
> https://www.postgresql.org/docs/current/protocol-message-formats.html
>
> A quick look at that indicates to me it is more involved then you think.
>
> >
> > Thanks, Shaheed
> >
> >
> > --
> > Adrian Klaver
> > adrian(dot)klaver(at)aklaver(dot)com <mailto:adrian(dot)klaver(at)aklaver(dot)com>
> >
>
> --
> Adrian Klaver
> adrian(dot)klaver(at)aklaver(dot)com
>
>