On Wed, 17 Dec 2025 at 00:54, Noah Misch <[email protected]> wrote: > > On Tue, Dec 16, 2025 at 05:22:36PM +0530, vignesh C wrote: > > On Tue, 16 Dec 2025 at 00:00, Noah Misch <[email protected]> wrote: > > > On Mon, Dec 15, 2025 at 11:35:35PM +0530, vignesh C wrote: > > > > This issue has started failing after commit: > > > > commit 0decd5e89db9f5edb9b27351082f0d74aae7a9b6 > > > > Sort dump objects independent of OIDs, for the 7 holdout object types. > > > > That makes sense. Thanks. Do you have commands we could add to > > > src/test/regress/sql/subscription.sql to cover this code? > > > > This dumping of subscription relation is specific to upgrading to > > preserve the subscription relation. So I felt we will not be able to > > add tests to subscription.sql, instead how about adding one more table > > to 004_subscription.pl where subscription upgrade tests are verified > > like the attached patch. > > That's a good way to test it. > > > --- a/src/bin/pg_dump/pg_dump_sort.c > > +++ b/src/bin/pg_dump/pg_dump_sort.c > > @@ -454,6 +454,20 @@ DOTypeNameCompare(const void *p1, const void *p2) > > if (cmpval != 0) > > return cmpval; > > } > > + else if (obj1->objType == DO_SUBSCRIPTION_REL) > > + { > > + SubRelInfo *srobj1 = *(SubRelInfo *const *) p1; > > + SubRelInfo *srobj2 = *(SubRelInfo *const *) p2; > > + > > + /* Sort by schema name (subscription name was already > > considered) */ > > Let's change the values getSubscriptionRelations() stores in > SubrRelInfo.obj.namespace and SubrRelInfo.obj.name to be more like > DO_PUBLICATION_REL: > > pubrinfo[j].dobj.namespace = tbinfo->dobj.namespace; > pubrinfo[j].dobj.name = tbinfo->dobj.name; > pubrinfo[j].publication = pubinfo; > > DO_SUBSCRIPTION_REL (new in v17) is gratuitously different: > > subrinfo[i].dobj.name = pg_strdup(subinfo->dobj.name); > subrinfo[i].tblinfo = tblinfo; > subrinfo[i].subinfo = subinfo;
Modified > What do you think? This would change sort order from (subname, rel) to (rel, > subname). Historically, we've avoided churning dump order, in case folks diff > dump output over time. Although grouping by publication and subscription would be cleaner, publication relations have been ordered first by relation and then by publication. Retaining the same ordering for subscription relations preserves consistency. > I bet that practice is less common for binary dumps. > Since DO_SUBSCRIPTION_REL is only for binary dumps, let's just change it. > > > + cmpval = strcmp(srobj1->tblinfo->dobj.namespace->dobj.name, > > + > > srobj2->tblinfo->dobj.namespace->dobj.name); > > The subrinfo change will make this comparison go away. If it had stayed, it > should be ready for namespace==NULL like pgTypeNameCompare() is. (I think > pgTypeNameCompare() could drop that defense, because the getTypes() call to > findNamespace() will pg_fatal() on absence. Until it does drop that defense, > the rest of pg_dump_sort.c should handle namespace==NULL.) > > > --- a/src/bin/pg_upgrade/t/004_subscription.pl > > +++ b/src/bin/pg_upgrade/t/004_subscription.pl > > > -# Wait till the table tab_upgraded1 reaches 'ready' state > > +# Wait till the tables tab_upgraded and tab_upgraded1 reaches 'ready' state > > s/reaches/reach/ there. Modified > To find other text needing edits, I searched this file for tab_upgraded. The > following comment still implies "all tables" encompasses just two: > > # ------------------------------------------------------ > # Check that pg_upgrade is successful when all tables are in ready or in > # init state (tab_upgraded1 table is in ready state and tab_upgraded2 table is > # in init state) along with retaining the replication origin's remote lsn, > # subscription's running status, failover option, and retain_dead_tuples > # option. > # ------------------------------------------------------ Updated the comments The attached v3 version patch has the changes for the same. Regards, Vignesh
From 5b3c16301be655f8205f3f5308696aa785375493 Mon Sep 17 00:00:00 2001 From: Vignesh C <[email protected]> Date: Mon, 15 Dec 2025 23:27:40 +0530 Subject: [PATCH v3] Fix pg_dump crash for DO_SUBSCRIPTION_REL sorting MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit pg_dump did not fully order DO_SUBSCRIPTION_REL objects. When multiple subscription–relation entries belonged to the same subscription, the comparison fell through to the assertion path and crashed. Fix this by extending the comparison to order such entries by the referenced table's schema name and table name. --- src/bin/pg_dump/pg_dump.c | 8 ++++---- src/bin/pg_dump/pg_dump_sort.c | 10 +++++++++ src/bin/pg_upgrade/t/004_subscription.pl | 26 +++++++++++++++--------- 3 files changed, 30 insertions(+), 14 deletions(-) diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c index 24ad201af2f..0f3f41c10b4 100644 --- a/src/bin/pg_dump/pg_dump.c +++ b/src/bin/pg_dump/pg_dump.c @@ -5386,7 +5386,9 @@ getSubscriptionRelations(Archive *fout) subrinfo[i].dobj.catId.tableoid = relid; subrinfo[i].dobj.catId.oid = cur_srsubid; AssignDumpId(&subrinfo[i].dobj); - subrinfo[i].dobj.name = pg_strdup(subinfo->dobj.name); + subrinfo[i].dobj.namespace = tblinfo->dobj.namespace; + subrinfo[i].dobj.name = tblinfo->dobj.name; + subrinfo[i].subinfo = subinfo; subrinfo[i].tblinfo = tblinfo; subrinfo[i].srsubstate = PQgetvalue(res, i, i_srsubstate)[0]; if (PQgetisnull(res, i, i_srsublsn)) @@ -5394,8 +5396,6 @@ getSubscriptionRelations(Archive *fout) else subrinfo[i].srsublsn = pg_strdup(PQgetvalue(res, i, i_srsublsn)); - subrinfo[i].subinfo = subinfo; - /* Decide whether we want to dump it */ selectDumpableObject(&(subrinfo[i].dobj), fout); } @@ -5438,7 +5438,7 @@ dumpSubscriptionTable(Archive *fout, const SubRelInfo *subrinfo) "\n-- For binary upgrade, must preserve the subscriber table.\n"); appendPQExpBufferStr(query, "SELECT pg_catalog.binary_upgrade_add_sub_rel_state("); - appendStringLiteralAH(query, subrinfo->dobj.name, fout); + appendStringLiteralAH(query, subinfo->dobj.name, fout); appendPQExpBuffer(query, ", %u, '%c'", subrinfo->tblinfo->dobj.catId.oid, diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c index 164c76e0864..661d933e8aa 100644 --- a/src/bin/pg_dump/pg_dump_sort.c +++ b/src/bin/pg_dump/pg_dump_sort.c @@ -454,6 +454,16 @@ DOTypeNameCompare(const void *p1, const void *p2) if (cmpval != 0) return cmpval; } + else if (obj1->objType == DO_SUBSCRIPTION_REL) + { + SubRelInfo *srobj1 = *(SubRelInfo *const *) p1; + SubRelInfo *srobj2 = *(SubRelInfo *const *) p2; + + /* Sort by subscription name, since (namespace, name) match the rel */ + cmpval = strcmp(srobj1->subinfo->dobj.name, srobj2->subinfo->dobj.name); + if (cmpval != 0) + return cmpval; + } /* * Shouldn't get here except after catalog corruption, but if we do, sort diff --git a/src/bin/pg_upgrade/t/004_subscription.pl b/src/bin/pg_upgrade/t/004_subscription.pl index 77387be0f9d..fdb9806713e 100644 --- a/src/bin/pg_upgrade/t/004_subscription.pl +++ b/src/bin/pg_upgrade/t/004_subscription.pl @@ -250,22 +250,25 @@ rmtree($new_sub->data_dir . "/pg_upgrade_output.d"); # Verify that the upgrade should be successful with tables in 'ready'/'init' # state along with retaining the replication origin's remote lsn, # subscription's running status, failover option, and retain_dead_tuples -# option. +# option. Use multiple tables to verify deterministic pg_dump ordering +# of subscription relations during --binary-upgrade. $publisher->safe_psql( 'postgres', qq[ + CREATE TABLE tab_upgraded(id int); CREATE TABLE tab_upgraded1(id int); - CREATE PUBLICATION regress_pub4 FOR TABLE tab_upgraded1; + CREATE PUBLICATION regress_pub4 FOR TABLE tab_upgraded, tab_upgraded1; ]); $old_sub->safe_psql( 'postgres', qq[ + CREATE TABLE tab_upgraded(id int); CREATE TABLE tab_upgraded1(id int); CREATE SUBSCRIPTION regress_sub4 CONNECTION '$connstr' PUBLICATION regress_pub4 WITH (failover = true, retain_dead_tuples = true); ]); -# Wait till the table tab_upgraded1 reaches 'ready' state +# Wait till the tables tab_upgraded and tab_upgraded1 reach 'ready' state my $synced_query = - "SELECT count(1) = 1 FROM pg_subscription_rel WHERE srsubstate = 'r'"; + "SELECT count(1) = 2 FROM pg_subscription_rel WHERE srsubstate = 'r'"; $old_sub->poll_query_until('postgres', $synced_query) or die "Timed out while waiting for the table to reach ready state"; @@ -303,6 +306,8 @@ my $remote_lsn = $old_sub->safe_psql('postgres', # Have the subscription in disabled state before upgrade $old_sub->safe_psql('postgres', "ALTER SUBSCRIPTION regress_sub5 DISABLE"); +my $tab_upgraded_oid = $old_sub->safe_psql('postgres', + "SELECT oid FROM pg_class WHERE relname = 'tab_upgraded'"); my $tab_upgraded1_oid = $old_sub->safe_psql('postgres', "SELECT oid FROM pg_class WHERE relname = 'tab_upgraded1'"); my $tab_upgraded2_oid = $old_sub->safe_psql('postgres', @@ -317,10 +322,10 @@ $new_sub->append_conf('postgresql.conf', # ------------------------------------------------------ # Check that pg_upgrade is successful when all tables are in ready or in -# init state (tab_upgraded1 table is in ready state and tab_upgraded2 table is -# in init state) along with retaining the replication origin's remote lsn, -# subscription's running status, failover option, and retain_dead_tuples -# option. +# init state (tab_upgraded and tab_upgraded1 tables are in ready state and +# tab_upgraded2 table is in init state) along with retaining the replication +# origin's remote lsn, subscription's running status, failover option, and +# retain_dead_tuples option. # ------------------------------------------------------ command_ok( [ @@ -369,9 +374,10 @@ regress_sub5|f|f|f), # Subscription relations should be preserved $result = $new_sub->safe_psql('postgres', "SELECT srrelid, srsubstate FROM pg_subscription_rel ORDER BY srrelid"); -is( $result, qq($tab_upgraded1_oid|r +is( $result, qq($tab_upgraded_oid|r +$tab_upgraded1_oid|r $tab_upgraded2_oid|i), - "there should be 2 rows in pg_subscription_rel(representing tab_upgraded1 and tab_upgraded2)" + "there should be 3 rows in pg_subscription_rel(representing tab_upgraded, tab_upgraded1 and tab_upgraded2)" ); # The replication origin's remote_lsn should be preserved -- 2.43.0
