Postgres Pro
Downloads
Home   >   mailing lists

Re: [HACKERS] logical decoding of two-phase transactions - Mailing list pgsql-hackers

From Ajin Cherian
Subject Re: [HACKERS] logical decoding of two-phase transactions
Date
Msg-id CAFPTHDaP1A8WVrOmS_Cx+SmLuvtguq7KOp=zbrSMqHi4BOSsiA@mail.gmail.com
Whole thread Raw
In response to Re: [HACKERS] logical decoding of two-phase transactions  (Amit Kapila <amit.kapila16@gmail.com>)
Responses Re: [HACKERS] logical decoding of two-phase transactions
List pgsql-hackers
Tree view


On Sat, Mar 20, 2021 at 1:35 AM Amit Kapila <amit.kapila16@gmail.com> wrote:
On Fri, Mar 19, 2021 at 5:03 AM Ajin Cherian <itsajin@gmail.com> wrote:
>
> Missed the patch - 0001, resending.
>


@@ -538,10 +550,21 @@ CreateDecodingContext(XLogRecPtr start_lsn,
..
+ /* Set two_phase_at LSN only if it hasn't already been set. */
+ if (ctx->twophase && !MyReplicationSlot->data.two_phase_at)
+ {
+ MyReplicationSlot->data.two_phase_at = start_lsn;
+ slot->data.two_phase = true;
+ ReplicationSlotMarkDirty();
+ ReplicationSlotSave();
+ SnapBuildSetTwoPhaseAt(ctx->snapshot_builder, start_lsn);
+ }

What if the walsender or apply worker restarts after setting
two_phase_at/two_phase here and updating the two_phase state in
pg_subscription? Won't we need to set SnapBuildSetTwoPhaseAt after
restart as well?

After a restart,   two_phase_at will be set by calling  AllocateSnapshotBuilder with two_phase_at

@@ -207,7 +207,7 @@ StartupDecodingContext(List *output_plugin_options,
  ctx->reorder = ReorderBufferAllocate();
  ctx->snapshot_builder =
  AllocateSnapshotBuilder(ctx->reorder, xmin_horizon, start_lsn,
- need_full_snapshot, slot->data.initial_consistent_point);
+ need_full_snapshot, slot->data.two_phase_at);

and then in  AllocateSnapshotBuilder:

@@ -309,7 +306,7 @@ AllocateSnapshotBuilder(ReorderBuffer *reorder,
  builder->initial_xmin_horizon = xmin_horizon;
  builder->start_decoding_at = start_lsn;
  builder->building_full_snapshot = need_full_snapshot;
- builder->initial_consistent_point = initial_consistent_point;
+ builder->two_phase_at = two_phase_at;

regards,
Ajin Cherian
Fujitsu Australia

pgsql-hackers by date:

Previous
From: Hannu Krosing
Date:
Subject: Re: shared memory stats: high level design decisions: consistency, dropping
Next
From: Peter Smith
Date:
Subject: Re: [HACKERS] logical decoding of two-phase transactions