pg.ddx.io pgpool-committers@postgresql.org mailing list archivehelp / color / mirror / Atom feed
pgpool: Fix segfault when a watchdog sync changes the primary under run 6+ messages / 1 participants [nested] [flat]
* pgpool: Fix segfault when a watchdog sync changes the primary under run @ 2026-11-13 01:14 Tatsuo Ishii <ishii@postgresql.org> 0 siblings, 0 replies; 6+ messages in thread From: Tatsuo Ishii @ 2026-11-13 01:14 UTC (permalink / raw) To: pgpool-committers@lists.postgresql.org Fix segfault when a watchdog sync changes the primary under running children. sync_backend_from_watchdog() applies a failover performed by the leader watchdog node: it updates the backend status, Req_info->primary_node_id and Req_info->main_node_id, and only then kills and re-forks the children. Unlike failover() it did not set Req_info->switching, so a child processing a query meanwhile could get the new primary node from pool_virtual_main_db_node_id(). If its session had no connection to that node, for example because the node was down when the session started, the child dereferenced a NULL connection slot (MAIN(backend)->fd in read_packets_and_process()). Set Req_info->switching while syncing, as failover() does, so children wait for the sync to finish or to be killed. Requests registered while switching is set do not wake up the main process, so raise the failover interrupt when the sync ends with requests queued. Author: Emond Papegaaij <emond.papegaaij@gmail.com> Reported-by: Claude Reviewed-by: Tatsuo Ishii <ishii@postgresql.org> Backpatch-through: v4.3 Branch ------ V4_3_STABLE Details ------- https://git.postgresql.org/gitweb?p=pgpool2.git;a=commitdiff;h=aec897b7261c27854f78bc3cc778f0d51585b... Modified Files -------------- src/main/pgpool_main.c | 45 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 45 insertions(+) ^ permalink raw reply [nested|flat] 6+ messages in thread
* pgpool: Fix segfault when a watchdog sync changes the primary under run @ 2026-11-13 01:14 Tatsuo Ishii <ishii@postgresql.org> 0 siblings, 0 replies; 6+ messages in thread From: Tatsuo Ishii @ 2026-11-13 01:14 UTC (permalink / raw) To: pgpool-committers@lists.postgresql.org Fix segfault when a watchdog sync changes the primary under running children. sync_backend_from_watchdog() applies a failover performed by the leader watchdog node: it updates the backend status, Req_info->primary_node_id and Req_info->main_node_id, and only then kills and re-forks the children. Unlike failover() it did not set Req_info->switching, so a child processing a query meanwhile could get the new primary node from pool_virtual_main_db_node_id(). If its session had no connection to that node, for example because the node was down when the session started, the child dereferenced a NULL connection slot (MAIN(backend)->fd in read_packets_and_process()). Set Req_info->switching while syncing, as failover() does, so children wait for the sync to finish or to be killed. Requests registered while switching is set do not wake up the main process, so raise the failover interrupt when the sync ends with requests queued. Author: Emond Papegaaij <emond.papegaaij@gmail.com> Reported-by: Claude Reviewed-by: Tatsuo Ishii <ishii@postgresql.org> Backpatch-through: v4.3 Branch ------ V4_4_STABLE Details ------- https://git.postgresql.org/gitweb?p=pgpool2.git;a=commitdiff;h=4bb2e86b2a56fdd24d3bb4fabf8703efc939f... Modified Files -------------- src/main/pgpool_main.c | 45 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 45 insertions(+) ^ permalink raw reply [nested|flat] 6+ messages in thread
* pgpool: Fix segfault when a watchdog sync changes the primary under run @ 2026-11-13 01:14 Tatsuo Ishii <ishii@postgresql.org> 0 siblings, 0 replies; 6+ messages in thread From: Tatsuo Ishii @ 2026-11-13 01:14 UTC (permalink / raw) To: pgpool-committers@lists.postgresql.org Fix segfault when a watchdog sync changes the primary under running children. sync_backend_from_watchdog() applies a failover performed by the leader watchdog node: it updates the backend status, Req_info->primary_node_id and Req_info->main_node_id, and only then kills and re-forks the children. Unlike failover() it did not set Req_info->switching, so a child processing a query meanwhile could get the new primary node from pool_virtual_main_db_node_id(). If its session had no connection to that node, for example because the node was down when the session started, the child dereferenced a NULL connection slot (MAIN(backend)->fd in read_packets_and_process()). Set Req_info->switching while syncing, as failover() does, so children wait for the sync to finish or to be killed. Requests registered while switching is set do not wake up the main process, so raise the failover interrupt when the sync ends with requests queued. Author: Emond Papegaaij <emond.papegaaij@gmail.com> Reported-by: Claude Reviewed-by: Tatsuo Ishii <ishii@postgresql.org> Backpatch-through: v4.3 Branch ------ V4_5_STABLE Details ------- https://git.postgresql.org/gitweb?p=pgpool2.git;a=commitdiff;h=2db6100bd69fa42496106a3af07e4b7ea784b... Modified Files -------------- src/main/pgpool_main.c | 45 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 45 insertions(+) ^ permalink raw reply [nested|flat] 6+ messages in thread
* pgpool: Fix segfault when a watchdog sync changes the primary under run @ 2026-11-13 01:14 Tatsuo Ishii <ishii@postgresql.org> 0 siblings, 0 replies; 6+ messages in thread From: Tatsuo Ishii @ 2026-11-13 01:14 UTC (permalink / raw) To: pgpool-committers@lists.postgresql.org Fix segfault when a watchdog sync changes the primary under running children. sync_backend_from_watchdog() applies a failover performed by the leader watchdog node: it updates the backend status, Req_info->primary_node_id and Req_info->main_node_id, and only then kills and re-forks the children. Unlike failover() it did not set Req_info->switching, so a child processing a query meanwhile could get the new primary node from pool_virtual_main_db_node_id(). If its session had no connection to that node, for example because the node was down when the session started, the child dereferenced a NULL connection slot (MAIN(backend)->fd in read_packets_and_process()). Set Req_info->switching while syncing, as failover() does, so children wait for the sync to finish or to be killed. Requests registered while switching is set do not wake up the main process, so raise the failover interrupt when the sync ends with requests queued. Author: Emond Papegaaij <emond.papegaaij@gmail.com> Reported-by: Claude Reviewed-by: Tatsuo Ishii <ishii@postgresql.org> Backpatch-through: v4.3 Branch ------ V4_6_STABLE Details ------- https://git.postgresql.org/gitweb?p=pgpool2.git;a=commitdiff;h=223545c39e2a8426a159b7eddd29f465509e4... Modified Files -------------- src/main/pgpool_main.c | 45 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 45 insertions(+) ^ permalink raw reply [nested|flat] 6+ messages in thread
* pgpool: Fix segfault when a watchdog sync changes the primary under run @ 2026-11-13 01:14 Tatsuo Ishii <ishii@postgresql.org> 0 siblings, 0 replies; 6+ messages in thread From: Tatsuo Ishii @ 2026-11-13 01:14 UTC (permalink / raw) To: pgpool-committers@lists.postgresql.org Fix segfault when a watchdog sync changes the primary under running children. sync_backend_from_watchdog() applies a failover performed by the leader watchdog node: it updates the backend status, Req_info->primary_node_id and Req_info->main_node_id, and only then kills and re-forks the children. Unlike failover() it did not set Req_info->switching, so a child processing a query meanwhile could get the new primary node from pool_virtual_main_db_node_id(). If its session had no connection to that node, for example because the node was down when the session started, the child dereferenced a NULL connection slot (MAIN(backend)->fd in read_packets_and_process()). Set Req_info->switching while syncing, as failover() does, so children wait for the sync to finish or to be killed. Requests registered while switching is set do not wake up the main process, so raise the failover interrupt when the sync ends with requests queued. Author: Emond Papegaaij <emond.papegaaij@gmail.com> Reported-by: Claude Reviewed-by: Tatsuo Ishii <ishii@postgresql.org> Backpatch-through: v4.3 Branch ------ V4_7_STABLE Details ------- https://git.postgresql.org/gitweb?p=pgpool2.git;a=commitdiff;h=908ac2ea76ba741d16758068aa76b7b59a36c... Modified Files -------------- src/main/pgpool_main.c | 45 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 45 insertions(+) ^ permalink raw reply [nested|flat] 6+ messages in thread
* pgpool: Fix segfault when a watchdog sync changes the primary under run @ 2026-11-13 01:14 Tatsuo Ishii <ishii@postgresql.org> 0 siblings, 0 replies; 6+ messages in thread From: Tatsuo Ishii @ 2026-11-13 01:14 UTC (permalink / raw) To: pgpool-committers@lists.postgresql.org Fix segfault when a watchdog sync changes the primary under running children. sync_backend_from_watchdog() applies a failover performed by the leader watchdog node: it updates the backend status, Req_info->primary_node_id and Req_info->main_node_id, and only then kills and re-forks the children. Unlike failover() it did not set Req_info->switching, so a child processing a query meanwhile could get the new primary node from pool_virtual_main_db_node_id(). If its session had no connection to that node, for example because the node was down when the session started, the child dereferenced a NULL connection slot (MAIN(backend)->fd in read_packets_and_process()). Set Req_info->switching while syncing, as failover() does, so children wait for the sync to finish or to be killed. Requests registered while switching is set do not wake up the main process, so raise the failover interrupt when the sync ends with requests queued. Author: Emond Papegaaij <emond.papegaaij@gmail.com> Reported-by: Claude Reviewed-by: Tatsuo Ishii <ishii@postgresql.org> Backpatch-through: v4.3 Branch ------ master Details ------- https://git.postgresql.org/gitweb?p=pgpool2.git;a=commitdiff;h=7e6bf8326526451224a182ab182d53f2597a7... Modified Files -------------- src/main/pgpool_main.c | 45 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 45 insertions(+) ^ permalink raw reply [nested|flat] 6+ messages in thread
end of thread, other threads:[~2026-11-13 01:14 UTC | newest] Thread overview: 6+ messages (download: mbox mbox.gz follow: Atom feed) -- links below jump to the message on this page -- 2026-11-13 01:14 pgpool: Fix segfault when a watchdog sync changes the primary under run Tatsuo Ishii <ishii@postgresql.org> 2026-11-13 01:14 pgpool: Fix segfault when a watchdog sync changes the primary under run Tatsuo Ishii <ishii@postgresql.org> 2026-11-13 01:14 pgpool: Fix segfault when a watchdog sync changes the primary under run Tatsuo Ishii <ishii@postgresql.org> 2026-11-13 01:14 pgpool: Fix segfault when a watchdog sync changes the primary under run Tatsuo Ishii <ishii@postgresql.org> 2026-11-13 01:14 pgpool: Fix segfault when a watchdog sync changes the primary under run Tatsuo Ishii <ishii@postgresql.org> 2026-11-13 01:14 pgpool: Fix segfault when a watchdog sync changes the primary under run Tatsuo Ishii <ishii@postgresql.org>
This inbox is served by DDX for PostgreSQL; see mirroring instructions for how to clone and mirror all data and code used for this inbox