Re: PATCH: logical_work_mem and logical streaming of large in-progress transactions

On Tue, Jan 28, 2020 at 1:55 PM Dilip Kumar <dilipbalaut(at)gmail(dot)com> wrote:
>
> On Tue, Jan 28, 2020 at 1:30 PM Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
> >
> > On Tue, Jan 28, 2020 at 11:58 AM Dilip Kumar <dilipbalaut(at)gmail(dot)com> wrote:
> > >
> > > On Tue, Jan 28, 2020 at 11:43 AM Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
> > > >
> > > > > > It seems to me that we need to add all of this new handling because
> > > > > > while taking the decision whether to stream or not we don't know
> > > > > > whether the txn has changes that can't be streamed. One idea to make
> > > > > > it work is that we identify it while decoding the WAL. I think we
> > > > > > need to set a bit in the insert/delete WAL record to identify if the
> > > > > > tuple belongs to a toast relation. This won't add any additional
> > > > > > overhead in WAL and reduce a lot of complexity in the logical decoding
> > > > > > and also decoding will be efficient. If this is feasible, then we can
> > > > > > do the same for speculative insertions.
> > > > > The Idea looks good to me. I will work on this.
> > > > >
> > > >
> > > > One more thing we can do is to identify whether the tuple belongs to
> > > > toast relation while decoding it. However, I think to do that we need
> > > > to have access to relcache at that time and that might add some
> > > > overhead as we need to do that for each tuple. Can we investigate
> > > > what it will take to do that and if it is better than setting a bit
> > > > during WAL logging.
> > >
> > > IMHO, for the catalog scan, we will have to start/stop the transaction
> > > for each change. So do you want that we should evaluate its
> > > performance?
> > >
> >
> > No, I was not thinking about each change, but at the level of ReorderBufferTXN.
> That means we will have to keep that transaction open until we decode
> the commit WAL for that ReorderBufferTXN or you have anything else in
> mind?
>

or probably till we start streaming.

> >
> > > Also, during we get the change we might not have the
> > > complete historic snapshot ready to fetch the rel cache entry.
> > >
> >
> > Before decoding each change (say DecodeInsert), we call
> > SnapBuildProcessChange. Isn't that sufficient?
> Yeah, Right, we can get some recache entry based on the base snapshot.
> And, that might be sufficient to know whether it's a toast relation or
> not.
> >
> > Even, if the above is possible, I am not sure how good is it for each
> > change we fetch rel cache entry, that is the point I was worried.
>
> We might not need to scan the catalog every time, we might get it from
> the cache itself.
>

Right, but I am not completely sure if that is better than setting a
bit in WAL record for toast tuples.

--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com