pgpool2.git
2 years agoFix PREPARE in multi-statement case.
Tatsuo Ishii [Wed, 12 Jul 2023 07:35:46 +0000 (16:35 +0900)]
Fix PREPARE in multi-statement case.

If multi-statement query includes PREPARE in the second or latter
position, and subsequent bind message uses the prepared statement, it
fails with "unable to bind" error because the prepared statement is not
saved in sent messages.

To fix this if such a case found after parsing the statement, create a
query context for the named statement and add it to the sent message
list.
Discussion: https://www.pgpool.net/pipermail/pgpool-general/2023-July/008931.html

For this new regression test 079..multi_prepare is added.

2 years agoDoc: Enhance SSH public key authentication setting section in "8.2. Pgpool-II + Watch...
Bo Peng [Tue, 11 Jul 2023 06:10:20 +0000 (15:10 +0900)]
Doc: Enhance SSH public key authentication setting section in "8.2. Pgpool-II + Watchdog Setup Example".

2 years agoFix pgproto to work with bind message using params.
Tatsuo Ishii [Sun, 9 Jul 2023 10:09:26 +0000 (19:09 +0900)]
Fix pgproto to work with bind message using params.

Previously pgproto can only process bind messages without params.

2 years agoFix typo in log message.
Bo Peng [Wed, 5 Jul 2023 08:08:04 +0000 (17:08 +0900)]
Fix typo in log message.

2 years agoDoc: fix load balance mode chapter.
Tatsuo Ishii [Thu, 29 Jun 2023 02:27:31 +0000 (11:27 +0900)]
Doc: fix load balance mode chapter.

In some places "streaming replication mode" was written as "native
replication mode".  Also enhance description regarding additional
requirements for load balancing in the streaming replication mode.
Add some indexes.

2 years agoDowngrade log message.
Tatsuo Ishii [Thu, 29 Jun 2023 01:51:14 +0000 (10:51 +0900)]
Downgrade log message.

It is pointed out in https://www.pgpool.net/mantisbt/view.php?id=806
that there are log entries in the log file:

LOG: pool_pending_message_set_flush_request: msg: Parse

The messages were for debugging and the log level should have been a
DEBUG*. So I changed the log level from LOG to DEBUG5.

2 years agoDoc: fix load balance mode chapter.
Tatsuo Ishii [Wed, 28 Jun 2023 02:56:29 +0000 (11:56 +0900)]
Doc: fix load balance mode chapter.

In some places "streaming replication mode" was written as "native
replication mode".  Also enhance description regarding additional
requirements for load balancing in the streaming replication mode.
Add some indexes.

2 years agoDoc: enhance online recovery documentation.
Bo Peng [Tue, 27 Jun 2023 02:46:17 +0000 (11:46 +0900)]
Doc: enhance online recovery documentation.

Mention that 2nd stage of online recovery does not work properly only for multiple pgpool nodes without watchdog enabled.

2 years agoFix missing CTE SEARCH and CYCLE rewrites.
Bo Peng [Fri, 23 Jun 2023 08:05:53 +0000 (17:05 +0900)]
Fix missing CTE SEARCH and CYCLE rewrites.

In native replication mode, Pgpool-II need to rewrite Date/Time functions to timestamp in WRITE queries.
CTE SEARCH and CYCLE rewrites were missing.

2 years agoRemoved duplicate pcp_listen_address setting in src/sample/pgpool.conf.sample-stream.
Bo Peng [Mon, 19 Jun 2023 04:19:00 +0000 (13:19 +0900)]
Removed duplicate pcp_listen_address setting in src/sample/pgpool.conf.sample-stream.

2 years agoPrevent query cache update under shared lock.
Tatsuo Ishii [Wed, 14 Jun 2023 02:17:38 +0000 (11:17 +0900)]
Prevent query cache update under shared lock.

In https://www.pgpool.net/mantisbt/view.php?id=795 it was pointed out
that expired query cache entry can be modified under shared lock. This
could cause shared memory corruption used by query cache. In order to
fix this, we temporarily release the shared lock and then acquire an
exclusive lock before modifying the cache entry. This could create a
window and we need to get the cache entry meta data again to make sure
that the meta data has not been modified by someone else.

Back-patch to V4.4 stable where shared locking for query cache was
introduced.

2 years agoAdd schema qualification to some system catalog inquiry functions.
Tatsuo Ishii [Mon, 12 Jun 2023 00:46:47 +0000 (09:46 +0900)]
Add schema qualification to some system catalog inquiry functions.

There were a few places where schema qualification were not used while
issuing system catalog inquiry.

Discussion: https://www.pgpool.net/pipermail/pgpool-hackers/2023-June/004346.html

2 years agoFix delay_threshold_by_time and prefer_lower_standby_delay.
Tatsuo Ishii [Mon, 5 Jun 2023 11:18:36 +0000 (20:18 +0900)]
Fix delay_threshold_by_time and prefer_lower_standby_delay.

They were broken since delay_threshold_by_time was introduced in 4.4.

- delay_threshold_by_time was not checked in where_to_send. This broke
  load balancing when replication is delayed.

- select_load_balancing_node was broken if both
  delay_threshold_by_time and prefer_lower_standby_delay were enabled.

In order to fix the issue, where_to_send and
select_load_balancing_node are fixed.  Also add test case for
delay_threshold_by_time are added to 033.prefer_lower_standby_delay.

Discussion: https://www.pgpool.net/pipermail/pgpool-general/2023-June/008864.html

2 years agoFix the wrong variable names in replication_mode_recovery_2nd_stage.sample sample...
Bo Peng [Tue, 30 May 2023 13:17:33 +0000 (22:17 +0900)]
Fix the wrong variable names in replication_mode_recovery_2nd_stage.sample sample script.

2 years agoTest: fix occasional 069.memory_leak_extended test failure.
Tatsuo Ishii [Mon, 22 May 2023 07:27:40 +0000 (16:27 +0900)]
Test: fix occasional 069.memory_leak_extended test failure.

It turned out that reason of the occasional test failure is, pgpool
child process is gone before running ps command after pgbench
finishes. The cause is a kind mismatch FATAL error, "DISCARD ALL
cannot be executed within a pipeline". To fix this, run pgbench in
background and get the process size before pgbench finishes.

Discussion: https://www.pgpool.net/pipermail/pgpool-hackers/2023-May/004338.html

2 years agoPrepare 4.4.3 V4_4_3 V4_4_3_RPM
Bo Peng [Wed, 17 May 2023 12:22:45 +0000 (21:22 +0900)]
Prepare 4.4.3

2 years agoDoc: Bump version to 4.4.3
Bo Peng [Wed, 17 May 2023 12:15:56 +0000 (21:15 +0900)]
Doc: Bump version to 4.4.3

2 years agoDoc: modify release notes.
Bo Peng [Wed, 17 May 2023 12:11:20 +0000 (21:11 +0900)]
Doc: modify release notes.

2 years agoDoc: add release notes.
Bo Peng [Wed, 17 May 2023 11:48:49 +0000 (20:48 +0900)]
Doc: add release notes.

2 years agoTest: stabilize some tests.
Tatsuo Ishii [Fri, 12 May 2023 08:16:01 +0000 (17:16 +0900)]
Test: stabilize some tests.

069.memory_leak_extended, 070.memory_leak_extended_memqcache and
073.pg_terminate_backend uses "sleep 1" to confirm pgpool starting up,
which makes the tests unstable because there's no guarantee that pgpool
becomes ready within 1 second. Use wait_for_pgpool_startup instead to
stabilize the tests.

2 years agoTest: unbreak 033.prefer_lower_standby_delay.
Tatsuo Ishii [Tue, 2 May 2023 00:38:44 +0000 (09:38 +0900)]
Test: unbreak 033.prefer_lower_standby_delay.

Commit "7a28bbb1 Fix 033.prefer_lower_standby_delay" broke
033.prefer_lower_standby_delay for PostgreSQL 13 or before because
they do not have pg_get_wal_replay_pause_state().

2 years agoFix prefer_lower_delay_standby bug.
Tatsuo Ishii [Sun, 30 Apr 2023 06:41:02 +0000 (15:41 +0900)]
Fix prefer_lower_delay_standby bug.

When client connects to pgpool, one of standbys are chosen as the load
balancing node.  If standby delay exceeds delay_threshold while the
session continues, prefer_lower_delay_standby will choose the least
delay standby node as the new load balancing node and set the target
backend to the node. Unfortunately the decision was made *before* the
checking that SELECT query includes writing function etc., pgpool
happily sends SELECT which cannot be executed on standby.  To fix
this, prefer_lower_delay_standby treatment is moved after the writing
function etc. check.

033.prefer_lower_standby_delay regression test is modified to include
the case above. Also I have done some refactoring:

- Remove unnecessary while loop for each clustering mode because the
  test is only useful for streaming replication mode.

- Add checking wal_replay_pause is actually executed.

Bug reported by: https://www.pgpool.net/mantisbt/view.php?id=798
along with suggested fix.

2 years agoTest: fix regress.sh to show correct number of total tests.
Tatsuo Ishii [Sat, 29 Apr 2023 12:44:59 +0000 (21:44 +0900)]
Test: fix regress.sh to show correct number of total tests.

Previously it ignored the number of timed out tests. As a result total
number of tests showed incorrect number because num-total tests was
calculated as num-ok + num-failed.

This is an oversight when timeout was introduced in 6688332da.

2 years agoDoc: add explanation for wd_priority.
Chen Ningwei [Tue, 25 Apr 2023 01:34:26 +0000 (10:34 +0900)]
Doc: add explanation for wd_priority.

2 years agoDoc: add restriction regarding PREPARE/EXECUTE/DEALLOCATE.
Tatsuo Ishii [Wed, 19 Apr 2023 10:41:21 +0000 (19:41 +0900)]
Doc: add restriction regarding PREPARE/EXECUTE/DEALLOCATE.

2 years agoUpdate src/Makefile.in.
Bo Peng [Wed, 19 Apr 2023 08:15:09 +0000 (17:15 +0900)]
Update src/Makefile.in.

2 years agoAdd src/utils/psqlscan.c.
Tatsuo Ishii [Wed, 19 Apr 2023 07:56:24 +0000 (16:56 +0900)]
Add src/utils/psqlscan.c.

V4.4 or older branches still keep generated files in git repository. I
forgot this in commit:
https://git.postgresql.org/gitweb/?p=pgpool2.git;a=commit;h=77f1e64d715c25d24830623f705877030d34e768

2 years agoDoc: add caution to use -D option with pgpool.
Tatsuo Ishii [Wed, 19 Apr 2023 03:40:48 +0000 (12:40 +0900)]
Doc: add caution to use -D option with pgpool.

2 years agoDoc: remove unnecessary restriction regarding multi-statement.
Tatsuo Ishii [Sun, 9 Apr 2023 04:11:12 +0000 (13:11 +0900)]
Doc: remove unnecessary restriction regarding multi-statement.

This applies to master branch only for now because this needs
commit:
https://git.postgresql.org/gitweb/?p=pgpool2.git;a=commit;h=48da8715bf403965507eef0321c0ab10054ac71c

If we decide to back port this, we will apply this change to stable
branches.

2 years agoJudge multi statement query using psqlscan.
Tatsuo Ishii [Sat, 25 Mar 2023 07:21:27 +0000 (16:21 +0900)]
Judge multi statement query using psqlscan.

Psqlscan is a module in the PostgreSQL source tree. It is essentially
subset of PostgreSQL SQL scanner but it is specialized for detecting
the end of each SQL statement. Therefore we can count the number of
SQL statements in a query string by using it.
Note that we use psqlscan only when query string is large.

Because psqlscan is not designed as an external library, it is hard to
call it from outside of PostgreSQL source tree. So I decided to import
psqlscan source code. This module consists of multiple files.  Program
sources are deployed in src/utils directory and header files are
deployed in src/include/utils directory.

psqlscan module was originally designed for frontend programs and uses
malloc directly. So I changed them so that it calls palloc and
friends. Additionally pgstrcasecmp.c and pqexpbuffer.c are also
imported. They are used by psqlscan.

The example usage of psqlscan can be found in
multi_statement_query():src/protocol/pool_proto_modules.c.

Discussion:
https://www.pgpool.net/pipermail/pgpool-hackers/2023-February/004291.html
Discussion: https://www.pgpool.net/pipermail/pgpool-hackers/2023-April/004320.html

2 years agoTest: increase timeout in 077.invalid_failover_node test.
Tatsuo Ishii [Sun, 16 Apr 2023 02:31:49 +0000 (11:31 +0900)]
Test: increase timeout in 077.invalid_failover_node test.

Increase timeout for pcp_promote_node from 2*30 = 60 sec to 3*30 = 90
sec.  It seems the test fails just because too small timeout value.

2 years agoMore schema qualification added to system function.
Tatsuo Ishii [Sun, 16 Apr 2023 01:45:08 +0000 (10:45 +0900)]
More schema qualification added to system function.

2 years agoTest: skip the test if there's no test.sh.
Tatsuo Ishii [Sat, 15 Apr 2023 07:26:23 +0000 (16:26 +0900)]
Test: skip the test if there's no test.sh.

This is useful when developers want to run the test on git repository
because the check suppresses false positive errors. There could newer
branch's test directories remain without any contents and regress.sh
reports wrong failures.

2 years agoDoc: Enhancing installation documentation.
Bo Peng [Fri, 14 Apr 2023 04:00:59 +0000 (13:00 +0900)]
Doc: Enhancing installation documentation.
- add detailed decriptions of packages
- mention that  pgpool-II-pgXX-extensions needs to be installed on PostgreSQL servers

2 years agoDoc: add "Backing up PostgreSQL database" section to "Server Setup and Operation...
Tatsuo Ishii [Thu, 13 Apr 2023 00:25:09 +0000 (09:25 +0900)]
Doc: add "Backing up PostgreSQL database" section to "Server Setup and Operation" chapter.

2 years agoDoc: remove the configuration of "- D" start OPTS from "8.2. Pgpool-II + Watchdog...
Bo Peng [Thu, 13 Apr 2023 07:29:18 +0000 (16:29 +0900)]
Doc: remove the configuration of "- D" start OPTS from "8.2. Pgpool-II + Watchdog Setup Example".

2 years agoFix occasional 005.jdbc test failure.
Tatsuo Ishii [Thu, 6 Apr 2023 03:43:06 +0000 (12:43 +0900)]
Fix occasional 005.jdbc test failure.

The direct cause of the error is:
2023-02-22 08:51:47.705: PostgreSQL JDBC Driver pid 12420: LOG:  Parse: Error or notice message from backend: : DB node id: 0 backend pid: 12488 statement: "COMMIT" message: "prepared statement "S_1" already exists"

Actually the root of the error is this:
 2023-02-22 08:51:45.242: PostgreSQL JDBC Driver pid 12420: LOG:  pool_send_and_wait: Error or notice message from backend: : DB node id: 0 backend pid: 12488 statement: "DISCARD ALL" message: "DISCARD ALL cannot be executed within a pipeline"

"DISCARD ALL" was generated by pgpool (reset_query_list) to discard
some objects including prepared statements created in the
session. Since DISCARD ALL failed, the prepared statement S_1 was not
removed. Thus the next session failed because S_1 already existed.

To fix this, new global boolean flag reset_query_error is
introduced. The flag is set inside pool_send_and_wait() when a reset
query executed by SimpleQuery() results in ERROR. If the flag is true,
backend_cleanup() discards the backend connection so that any objects,
including named statement, corresponding to the session is discarded

Discussion: https://www.pgpool.net/pipermail/pgpool-hackers/2023-February/004293.html

2 years agoDoc: fix restrictions section.
Tatsuo Ishii [Sun, 9 Apr 2023 02:36:30 +0000 (11:36 +0900)]
Doc: fix restrictions section.

- Add mention about SCRAM-SHA-256 authentication
- Add mention about snapshot isolation mode

2 years agoDoc: add index for online recovery.
Tatsuo Ishii [Mon, 3 Apr 2023 23:21:28 +0000 (08:21 +0900)]
Doc: add index for online recovery.

2 years agoDoc: add explanation when pg node status is shown as "unknown".
Tatsuo Ishii [Mon, 3 Apr 2023 21:42:27 +0000 (06:42 +0900)]
Doc: add explanation when pg node status is shown as "unknown".

2 years agoDoc: add explanation when pg node status is shown as "unknown".
Tatsuo Ishii [Mon, 3 Apr 2023 12:17:36 +0000 (21:17 +0900)]
Doc: add explanation when pg node status is shown as "unknown".

2 years agoMore schema qualification fix.
Tatsuo Ishii [Thu, 30 Mar 2023 00:15:42 +0000 (09:15 +0900)]
More schema qualification fix.

This is a follow up to:
https://git.postgresql.org/gitweb/?p=pgpool2.git;a=commit;h=49ca4800d1a804c1f6ef5807d9fbeeef85888fb6

Back patch to all supported branches: 4.4 to 4.0

2 years agoChange pgpool_setup to append some parameters to the end of the configuration file.
Bo Peng [Tue, 28 Mar 2023 13:48:40 +0000 (22:48 +0900)]
Change pgpool_setup to append some parameters to the end of the configuration file.

2 years agoUse schema qualification for internal queries.
Tatsuo Ishii [Tue, 28 Mar 2023 01:55:59 +0000 (10:55 +0900)]
Use schema qualification for internal queries.

Some of objects such as function and cast did not use "pg_catalog."
schema qualification.  This does not lead to immediate security
concern but using the schema qualification is always good practice.

Not that for this I had to increase some buffer length:

- POOL_RELCACHE.query was changed from 1024 to 1500.
- query buffer size in pool_search_relcache was changed from 1024 to 1500.

Back patch to all supported branches: 4.4 to 4.0

2 years agoDoc: fix typo in the description of backend_application_name.
Tatsuo Ishii [Tue, 28 Mar 2023 00:49:08 +0000 (09:49 +0900)]
Doc: fix typo in the description of backend_application_name.

2 years agoFix compile error on systems using musl libc.
Tatsuo Ishii [Wed, 22 Mar 2023 11:53:38 +0000 (20:53 +0900)]
Fix compile error on systems using musl libc.

Patch provided by leimaohui.
https://www.pgpool.net/mantisbt/view.php?id=790

2 years agoDoc: mention that the target node to promote must be up and running.
Tatsuo Ishii [Wed, 22 Mar 2023 05:15:20 +0000 (14:15 +0900)]
Doc: mention that the target node to promote must be up and running.

Back patch to 4.3 in which the -n option of pcp_promote_node was
introduced.

2 years agoDoc: mention that AES256 support requires --with-openssl option.
Tatsuo Ishii [Wed, 22 Feb 2023 01:20:15 +0000 (10:20 +0900)]
Doc: mention that AES256 support requires --with-openssl option.

2 years agoDoc: add index for "AES256".
Tatsuo Ishii [Tue, 21 Feb 2023 08:25:20 +0000 (17:25 +0900)]
Doc: add index for "AES256".

Also fix previous commit for adding index ".pcppass" and "PCPPASSFILE".

2 years agoDoc: explicitly stat that it is -w option of pcp command is needed to use .pcppass.
Tatsuo Ishii [Tue, 21 Feb 2023 05:49:02 +0000 (14:49 +0900)]
Doc: explicitly stat that it is -w option of pcp command is needed to use .pcppass.

Also add .pcppass and PCPPASSFILE to index.

Backpatch-through: master and 4.4 to 4.0.

2 years agoFix that show pool_version shows row description twice.
Tatsuo Ishii [Sat, 18 Feb 2023 07:34:14 +0000 (16:34 +0900)]
Fix that show pool_version shows row description twice.

test=# show pool_version;
 pool_version
--------------
(0 rows)

     pool_version
-----------------------
 4.3.5 (tamahomeboshi)
(1 row)

2 years agoDoc: fix "2.8 Creating insert lock table" section.
Tatsuo Ishii [Wed, 15 Feb 2023 11:13:03 +0000 (20:13 +0900)]
Doc: fix "2.8 Creating insert lock table" section.

It only refereed to the native replication where it should have refereed
to the snapshot isolation mode.  Also enhance some xreflabels for
clustering mode.

2 years agoAllow to use multiple statements extensively.
Tatsuo Ishii [Sun, 12 Feb 2023 10:59:00 +0000 (19:59 +0900)]
Allow to use multiple statements extensively.

This commit tries to eliminate pgpool's long standing limitations
regarding multiple statements (multi-statements).

Previously

BEGIN;SELECT;
SAVEPOINT foo;

will fail in streaming replication mode because "BEGIN" was sent to
the primar node, but "SAVEPOINT" will be sent to both the primary and
standbys, and standbys will complain "SAVEPOINT can only be used in
transaction blocks".

Basic idea to solve the problem is, tracking explicit transactions
started by multi-statement queries so that all commands including
PREPARE, EXECUTE, DEALLOCATE, SAVEPOINT and COMMIT/ROLLBACK are sent
to the primary node in streaming replication mode or logical
replication mode.  In native replication or snapshot isolation mode,
those queries are sent to all of the backend nodes.

For this purpose new member: is_tx_started_by_multi_statement is added
to session context and also support functions are added.

extern bool is_tx_started_by_multi_statement_query(void);
extern void set_tx_started_by_multi_statement_query(void);
extern void unset_tx_started_by_multi_statement_query(void);

Discussion: https://www.pgpool.net/pipermail/pgpool-hackers/2023-February/004287.html
Back-patch-through: 4.2 as backport to 4.1 and before looks difficult

2 years agoFix multiple query cache bug.
Tatsuo Ishii [Sun, 5 Feb 2023 09:56:05 +0000 (18:56 +0900)]
Fix multiple query cache bug.

1) pool_add_item_shmem_cache() calls pool_init_cache_block() when
   there's no free cache item hash table entry. But this is
   unnecessary since pool_reuse_block() is already called from
   pool_add_item_shmem_cache(). This is actually harmless because the
   second pool_init_cache_block() call just set the same data as the
   first call of pool_init_cache_block(). It's just a waste of CPU
   cycle.

2) The cache blocks are supposed to be initialized while Pgpool-II
   starts up but actually not. Each cache block has the free space
   length in the block header after initialization. Since the free
   space length is not set, pool_get_block() fails to find a cache
   block which has enough free space, and it calls pool_reuse_block(),
   which is actually unnecessary (you will see something like
   "pool_reuse_block: blockid: 0" in pgpool log). Since
   pool_reuse_block() returns a free block anyway, this is just a
   waste of CPU cycle but better to fix.

Discussion: https://www.pgpool.net/pipermail/pgpool-hackers/2023-January/004259.html
Backpatch-through: 4.0

2 years agoFix sr worker to not send wrong query to standby server in corner case.
Tatsuo Ishii [Sun, 5 Feb 2023 07:13:15 +0000 (16:13 +0900)]
Fix sr worker to not send wrong query to standby server in corner case.

When ALWAYS_PRIMARY flag is set, PRIMARY_NODE_ID macro returns node
id, rather than -1 even if the primary is down. This confuses the test
if a node is primary or not, because PRIMARY_NODE_ID macro returns
main node id.  In this case streaming replication delay check worker
sends "SELECT pg_current_wal_lsn()" or "SELECT
pg_current_xlog_location()" depending on PostgreSQL's version to
standby which of course raises an error.

To fix this, test the primary node is down or not, and if it's down,
skip the replication delay loop. If primary is down, there's no point
to perform streaming replication delay checking anyway.

Discussion: https://www.pgpool.net/pipermail/pgpool-hackers/2023-February/004279.html
Backpatch-through: 4.0

2 years agoFix comment mistakes.
Tatsuo Ishii [Thu, 2 Feb 2023 10:34:00 +0000 (19:34 +0900)]
Fix comment mistakes.

2 years agoFix comment mistakes.
Tatsuo Ishii [Thu, 2 Feb 2023 05:29:33 +0000 (14:29 +0900)]
Fix comment mistakes.

2 years agoDoc: enhance show pool_cache manual.
Tatsuo Ishii [Wed, 1 Feb 2023 13:39:41 +0000 (22:39 +0900)]
Doc: enhance show pool_cache manual.

Add a table to explain each items shown by show pool_cache.

2 years agoRemove unnecessary macro.
Tatsuo Ishii [Tue, 31 Jan 2023 11:36:29 +0000 (20:36 +0900)]
Remove unnecessary macro.

The macro "DUAL_MODE" is no longer used anywhere.
Moreover, if it is used, it causes a compile error.

2 years agoFix kind mimatch error with DEALLOCATE
Tatsuo Ishii [Tue, 31 Jan 2023 09:49:48 +0000 (18:49 +0900)]
Fix kind mimatch error with DEALLOCATE

When conditions below are all met:
- streaming replication mode
- load balance node is other than primary
- PREPARE is used in a multi-statement query

Kind mimatch error occurs.

For DEALLOCATE pool_where_to_send() sets the nodes to be sent to all
backend if pgpool failed to find a prepared statement previously
received. For example with "SELECT 1\;PREPARE foo;", pgpool ignores
"PREPARE" part and just sends the whole multi-statement query to
primary. So primary actually has the prepared statement "foo" but
pgpool thinks that there's no prepared statement named "foo". And
pgpool sends DEALLOCATE to both primary and standby, then a kind
mismatch error raised. Fix is, just sending DEALLOCATE to primary node
in this case if pgpool is in streaming replication mode.  Same thing
can be said to EXECUTE too. I fixed this and merge similar treatment
with EXECUTE into where_to_send_deallocate() to make the code simpler.

I also found another bug: in replication mode or SI mode, pgpool needs
to send multi-statement query to all backend because the
multi-statement query maybe a write query. However pgpool sends to
main node only in this case.

Test cases are added to 071..execute_and_deallocate.

Backpatch-through: 4.0

Problem reported in:
https://www.pgpool.net/mantisbt/view.php?id=780

2 years agoObtain stronger lock while commiting shared relcache.
Tatsuo Ishii [Mon, 30 Jan 2023 02:02:22 +0000 (11:02 +0900)]
Obtain stronger lock while commiting shared relcache.

Previously pool_search_relcache() obtained only shared lock, which is
not safe enough to call pool_catalog_commit_cache() because it
registers new cache entry. Unfortunately our locking system does not
allow to escalate a shared lock to an exclusive lock. So we release
the shared lock then acquire the exclusive lock before calling
pool_catalog_commit_cache(). There's a window between them and we may
end up with an effort to register duplicate cache entry. But
underlying infrastructure of the query system will reject it and
should be safe.

Back-patch through 4.4 where the shared locking of the query cache
system was introduced.

2 years agoTest: refactor 071.execute_and_deallocate/test.sh.
Tatsuo Ishii [Sat, 28 Jan 2023 03:17:23 +0000 (12:17 +0900)]
Test: refactor 071.execute_and_deallocate/test.sh.

Previously the test was performed on only streaming replication mode.
Now the test covers native replication mode, snapshot isolation mode
and raw mode. Note that since case 6 test tries to test load balancing
on node 1, the case is not applied to raw mode.

2 years agoTest: fix indentation for further refactoring.
Tatsuo Ishii [Sat, 28 Jan 2023 02:37:18 +0000 (11:37 +0900)]
Test: fix indentation for further refactoring.

2 years agoDoc: fix typo in the description of read_only_function_list.
Tatsuo Ishii [Wed, 25 Jan 2023 05:45:34 +0000 (14:45 +0900)]
Doc: fix typo in the description of read_only_function_list.

2 years agoFix compile error of regression test
Takuma Hoshiai [Tue, 24 Jan 2023 15:46:46 +0000 (00:46 +0900)]
Fix compile error of regression test

The error occurred by timestamp test using gcc10.

2 years agoPrepare 4.4.2 V4_4_2 V4_4_2_RPM
Bo Peng [Sat, 21 Jan 2023 15:00:41 +0000 (00:00 +0900)]
Prepare 4.4.2

2 years agoDoc: Bump version to 4.4.2
Bo Peng [Sat, 21 Jan 2023 15:00:30 +0000 (00:00 +0900)]
Doc: Bump version to 4.4.2

2 years agoDoc: update copy right.
Bo Peng [Sat, 21 Jan 2023 15:00:17 +0000 (00:00 +0900)]
Doc: update copy right.

2 years agoDoc: Add release notes.
Bo Peng [Sat, 21 Jan 2023 15:00:04 +0000 (00:00 +0900)]
Doc: Add release notes.

2 years agoChange the default value for wd_lifecheck_password to empty string as documented.
Bo Peng [Sat, 21 Jan 2023 14:54:35 +0000 (23:54 +0900)]
Change the default value for wd_lifecheck_password to empty string as documented.

2 years agoDo not expose wd_lifecheck_password in show pool_status command.
Bo Peng [Sat, 21 Jan 2023 14:54:24 +0000 (23:54 +0900)]
Do not expose wd_lifecheck_password in show pool_status command.

2 years agoTest: tweak 028.watchdog_enable_consensus_with_half_votes.
Tatsuo Ishii [Sat, 14 Jan 2023 11:10:06 +0000 (20:10 +0900)]
Test: tweak 028.watchdog_enable_consensus_with_half_votes.

Sometimes buildfarm fails by:

bind on socket failed with error "Address already in use".

This happens after shutting down all pgpools then start them again. So
wait for 5 seconds between them instead of 1 second.  Let's see if
things are going to get better.

2 years agoAdd patch files to EXTRA_DIST.
Bo Peng [Mon, 16 Jan 2023 13:32:40 +0000 (22:32 +0900)]
Add patch files to EXTRA_DIST.

2 years agoAdd pgpool_adm--1.5.sql and pgpool_adm--1.4--1.5.sql.
Bo Peng [Mon, 16 Jan 2023 13:15:55 +0000 (22:15 +0900)]
Add pgpool_adm--1.5.sql and pgpool_adm--1.4--1.5.sql.

2 years agoFix pgpool_adm: pcp_pool_status.
Tatsuo Ishii [Mon, 16 Jan 2023 09:45:59 +0000 (18:45 +0900)]
Fix pgpool_adm: pcp_pool_status.

If the function is used with PostgreSQL 15, it fails with:

ERROR:  table-function protocol for value-per-call mode was not followed

pcp_pool_status actually returns set of record but it was declared as
"returns record".  This sloppiness had been tolerated until 14 but 15
becomes more strict. To fix this we need to change the return type of
pcp_pool_status from "returns record" to "returns setof record".
So the pgpool_adm extension version is bumped up.

2 years agoRemove pgpool-II-head.patch from SPEC file.
Bo Peng [Mon, 16 Jan 2023 08:27:50 +0000 (17:27 +0900)]
Remove pgpool-II-head.patch from SPEC file.

2 years agoFix integer overflow in streaming replication check.
Tatsuo Ishii [Fri, 13 Jan 2023 00:21:59 +0000 (09:21 +0900)]
Fix integer overflow in streaming replication check.

When delay_threshold_by_time is enabled, following query is
periodically sent to the primary node to obtain the replication delay
in micro seconds.

SELECT application_name, state, sync_state,(EXTRACT(EPOCH FROM
replay_lag)*1000000)::integer FROM pg_stat_replication";

Unfortunately the result from EXTRACT::integer overflows if replay_lag
is bigger than 2147483647/1000000 (that is about 35 minutes and 47
seconds). Fix is, changing "integer" to "bigint". By this, the result
will not overflow unless it becomes bigger than 292,472 years.
Discussion: https://www.pgpool.net/pipermail/pgpool-general/2023-January/008600.html
Backpatch-through: 4.4

2 years agoFix Uninitialized variables pointed out by Coverity
Takuma Hoshiai [Mon, 9 Jan 2023 18:08:13 +0000 (03:08 +0900)]
Fix Uninitialized variables pointed out by Coverity

Fix Uninitialized variables pointed out by Coverity.

2 years agoFix comment.
Tatsuo Ishii [Sat, 7 Jan 2023 04:27:34 +0000 (13:27 +0900)]
Fix comment.

sizeof(POOL_CACHE_ITEM_HEADER) is 16 bytes, not 24 bytes.

2 years agoAdd comment to query cache module header file.
Tatsuo Ishii [Thu, 5 Jan 2023 07:50:43 +0000 (16:50 +0900)]
Add comment to query cache module header file.

Some struct size added to explain required shared memory size.

2 years agoDoc: enhance "Configurations to use shared memory" chapter.
Tatsuo Ishii [Fri, 6 Jan 2023 01:37:21 +0000 (10:37 +0900)]
Doc: enhance "Configurations to use shared memory" chapter.

Enhance the description to try to explain what the parameters actually
mean. Also fix wrong number, i.e. the size of data management area is
64 bytes, not 48 bytes.

2 years agoDoc: enhance explanation of delay_threshold_by_time.
Tatsuo Ishii [Tue, 3 Jan 2023 09:05:03 +0000 (18:05 +0900)]
Doc: enhance explanation of delay_threshold_by_time.

Explicitly mention that in order to show delay_threshold_by_time,
backend_application_name is correctly set and match the
application_name in primary_conninfo.

Suggestion by zam bak.
Discussion: https://www.pgpool.net/pipermail/pgpool-general/2022-December/008584.html

2 years agoFix compiler warning.
Tatsuo Ishii [Sat, 24 Dec 2022 03:41:01 +0000 (12:41 +0900)]
Fix compiler warning.

pgpool-regclass.c needed to include "utils/varlena.h".  Also sort out
the header files order (except postgres.h which needs to be appear at
the top among PostgreSQL header files).

The compiler warning was reported by Florian Weimer.
https://www.pgpool.net/pipermail/pgpool-hackers/2022-December/004241.html

2 years agoPrepare 4.4.1 V4_4_1 V4_4_1_RPM
Bo Peng [Thu, 22 Dec 2022 06:47:48 +0000 (15:47 +0900)]
Prepare 4.4.1

2 years agoDoc: Bump doc version to 4.4.1
Bo Peng [Thu, 22 Dec 2022 06:39:24 +0000 (15:39 +0900)]
Doc: Bump doc version to 4.4.1

2 years agoDoc: update repo package version.
Bo Peng [Thu, 22 Dec 2022 06:37:25 +0000 (15:37 +0900)]
Doc: update repo package version.

2 years agoDoc: add release note.
Bo Peng [Thu, 22 Dec 2022 06:11:16 +0000 (15:11 +0900)]
Doc: add release note.

2 years agoFix time stamp rewrite bug.
Tatsuo Ishii [Tue, 20 Dec 2022 10:06:22 +0000 (19:06 +0900)]
Fix time stamp rewrite bug.

In native replication/snapshot isolation mode, any write query
including timestamp/date/time data are rewritten so that all
PostgreSQL servers accept same timestamp etc. value. From 4.4 outfuncs
module which is used to generate rewritten query was broken for
boolean data. In a parse tree constant data is represented as
"A_Const". 4.4 updated the module by using PostgreSQL 15's outfuncs
module. Starting from PostgreSQL 15 A_Const handles more data type
including boolean. Unfortunately the pgpool's outfuncs module did not
adopt the change. As a result boolean constant was ignored and turned
into empty string in the rewritten query string, which caused syntax
errors.

This commit fixes the issue. Also modify _outA_Const() so that it uses
_out* functions to handle other data types to save a few lines of
code.

Per report from Michiel van Leening.
Discussion: at: https://www.pgpool.net/pipermail/pgpool-general/2022-December/008581.html
Back-patch to: 4.4.

2 years agoFix for :[pgpool-hackers: 4227] Issue with failover_require_consensus
Muhammad Usama [Mon, 19 Dec 2022 19:39:32 +0000 (00:39 +0500)]
Fix for :[pgpool-hackers: 4227] Issue with failover_require_consensus

The fix is to dynamically set the failover command timeout based on the maximum
value of health check parameters across the watchdog cluster.

Reviewed and tested by Tatsuo Ishii

2 years agoDoc: enhance the performance section.
Tatsuo Ishii [Mon, 19 Dec 2022 07:04:39 +0000 (16:04 +0900)]
Doc: enhance the performance section.

Mention about dynamic process management.

2 years agoDoc: fix Japanese documentation to sync with English docs.
Tatsuo Ishii [Mon, 19 Dec 2022 04:50:33 +0000 (13:50 +0900)]
Doc: fix Japanese documentation to sync with English docs.

2 years agoAllow to define PGPOOLDIR uing environment variable.
Bo Peng [Sat, 17 Dec 2022 00:57:37 +0000 (09:57 +0900)]
Allow to define PGPOOLDIR uing environment variable.

per bug766.

2 years agoDowngrade streaming replication debugging message.
Tatsuo Ishii [Fri, 16 Dec 2022 09:49:02 +0000 (18:49 +0900)]
Downgrade streaming replication debugging message.

The log line "standby_delay: 54" was actually for debugging and should
have been down graded.  So I have downgraded to DEBUG1 from LOG. Also
tweaked the message to be saner.

2 years agoDoc: fix typo in Aurora example.
Tatsuo Ishii [Wed, 14 Dec 2022 08:38:29 +0000 (17:38 +0900)]
Doc: fix typo in Aurora example.

2 years agoDoc: enhance dynamic process management document.
Tatsuo Ishii [Mon, 12 Dec 2022 06:56:22 +0000 (15:56 +0900)]
Doc: enhance dynamic process management document.

2 years agoDowngrade "we have hit the ceiling" log.
Tatsuo Ishii [Mon, 12 Dec 2022 06:20:55 +0000 (15:20 +0900)]
Downgrade "we have hit the ceiling" log.

Under dynamic process management, when connection request exceeds
num_init_children, pgpool keeps on logging every 2 seconds "we have
hit the ceiling", which is flooding log file. So downgrade the log
level from LOG to DEBUG5.

2 years agoDoc: fix typo.
Tatsuo Ishii [Mon, 12 Dec 2022 05:48:06 +0000 (14:48 +0900)]
Doc: fix typo.

2 years agoFix typo in comments.
Tatsuo Ishii [Sat, 10 Dec 2022 06:23:30 +0000 (15:23 +0900)]
Fix typo in comments.

2 years agoPrepare 4.4.0 V4_4_0 V4_4_0_RPM
Bo Peng [Tue, 6 Dec 2022 01:58:11 +0000 (10:58 +0900)]
Prepare 4.4.0

2 years agoDOc: update doc version.
Bo Peng [Tue, 6 Dec 2022 01:57:23 +0000 (10:57 +0900)]
DOc: update doc version.