Re: [BUG] "FailedAssertion" reported when streaming in logical replication

On Mon, Apr 26, 2021 at 7:52 PM Dilip Kumar <dilipbalaut(at)gmail(dot)com> wrote:
>
> On Mon, Apr 26, 2021 at 6:59 PM Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
> >
> > On Mon, Apr 26, 2021 at 5:55 PM Dilip Kumar <dilipbalaut(at)gmail(dot)com> wrote:
> > >
> > > I am able to reproduce this and I think I have done the initial investigation.
> > >
> > > The cause of the issue is that, this transaction has only one change
> > > and that change is XLOG_HEAP2_NEW_CID, which is added through
> > > SnapBuildProcessNewCid. Basically, when we add any changes through
> > > SnapBuildProcessChange we set the base snapshot but when we add
> > > SnapBuildProcessNewCid this we don't set the base snapshot, because
> > > there is nothing to be done for this change. Now, this transaction is
> > > identified as the biggest transaction with non -partial changes, and
> > > now in ReorderBufferStreamTXN, it will return immediately because the
> > > base_snapshot is NULL.
> > >
> >
> > Your analysis sounds correct to me.
> >
>
> Thanks, I have attached a patch to fix this.
>

Can't we use 'txns_by_base_snapshot_lsn' list for this purpose? It is
ensured in ReorderBufferSetBaseSnapshot that we always assign
base_snapshot to a top-level transaction if the current is a known
subxact. I think that will be true because we always form xid-subxid
relation before processing each record in
LogicalDecodingProcessRecord.

Few other minor comments:
1. I think we can update the comments atop function ReorderBufferLargestTopTXN.
2. minor typo in comments atop ReorderBufferLargestTopTXN "...There is
a scope of optimization here such that we can select the largest
transaction which has complete changes...". In this 'complete' should
be incomplete. This is not related to this patch but I think we can
fix it along with this because anyway we are going to change
surrounding comments.

--
With Regards,
Amit Kapila.