Hello
I write patch to speed up ALTER TABLE SET NOT NULL by check existed check
constraints or indexes. Huge phase 3 with verify table data will be skipped if
table has valid check constraint cover "alteredfield IS NOT NULL" condition or
by SPI query if found index with compatible condition or
Here is new patch with check only existed valid constraints and without SPI at
all.
Thanksdiff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index d979ce2..7ab7580 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -370,6 +370,8
Thank you for review! My apologies for my errors. It seems i read developers
wiki pages not enough carefully. I will reread wiki, code style and then update
patch with all your remarks.
> The comment /* T if we added new NOT NULL constraints */ should
> probably be changed to /* T if we should
Why not just use ctrl+D shortcut? This EOF signal works both in bash, mysql,
psql, any CLI tool which I remember
Hello
I update patch and also rebase to current head. I hope now it is better aligned
with project stylediff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index dd4a8d3..36bcb3f 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@
Hello Michael
Thank you, i understand your opinion.
I really tried to find a discussion about reload recovery.conf (or walreceiver
conninfo changing and similar stuff), not about GUC.
regards, Sergei
> Having been bitten by the feature on MySQL, I think it's not a good thing.
Even in MySQL itself this feature was already removed.
Hello all
I want propose patch to make possible change primary_conninfo,
primary_slot_name, restore_command and trigger_file in recovery.conf without
restart postgresql.
Startup process will reread recovery.conf on SIGHUP.
My primary usecase is postgresql cluster with streaming replication.
Hi
I think having only long option is enough
regards, Sergei
Hello
I found SELECT pg_stat_statements_reset(NULL,NULL,s.queryid) in tests and it
pass tests, but i wonder how it works. Should not we check the NULL through
PG_ARGISNULL macro before any PG_GETARG_*? According src/include/fmgr.h
> * If function is not marked "proisstrict" in pg_proc, it must
Hello all
We already have feature to logging query parameters. If we use log_statement =
'all' we write parameters before execution and all is fine here. If we use
log_min_duration_statement statement is logged obviously after execution, but
currently we have no parameters if query was
Hello
> key.userid = GetUserId();
We can remove query id only from current user, right? Maybe better provide
optional argument for userid? Typically user with pg_read_all_stats and user
for application queries are different users.
At least it should be documented.
regards, Sergei
Hello
I tried build this patch and got error during make docs
> postgres.sgml:19626: element xref: validity error : IDREF attribute linkend
> references an unknown ID "runtime-checksumhelper-cost-limit"
> postgres.sgml:19625: element xref: validity error : IDREF attribute linkend
> references
Hello
Do you know http://atalia.postgresql.org/morgue/ repository?
> copy of postgresql-9.6-dbg_9.6.5-1.pgdg80+1_amd64.deb anywhere
Here is
http://atalia.postgresql.org/morgue/p/postgresql-9.6/postgresql-9.6-dbg_9.6.5-1.pgdg80%2b1_amd64.deb
regards, Sergei
Hello
I am surprised, but i can confirm error on versions prior 9.6: on 9.5, 9.4, 9.3
same error. On 9.6 and 10 query works correctly
All checked postgresql are x86_64 from postgresql debian repository:
http://apt.postgresql.org/pub/repos/apt/ jessie-pgdg main
postgres=# select 0.5678::numeric
Hello
>> patch -p1 gives some "Stripping trailing CRs from patch"
>> messages for me, but applied to current HEAD and builds. After
>
> Hmm. I wonder why I get that complaint so often. (It's rather
> common? or caused by the MIME format of my mail?) I'd say with
> confidence that it is because
HelloYou can start with functions ATPrepCmd and ATExecCmd in src/backend/commands/tablecmds.cAlso note, one alter table statement can have multiple different actions. Regards, Sergei11.01.2018, 18:56, "Jeremy Finzel" :Hello -I have found that in leveraging the parser code to
Hello
No, this is not bug.
https://www.postgresql.org/docs/10/static/hot-standby.html#HOT-STANDBY-CONFLICT
Message in DETAIL describe conflict reason.
> Application of a vacuum cleanup record from WAL conflicts with queries
> accessing the target page on the standby, whether or not the data to
Hello
> select into b from my_insert('from func atx');
You missed select something into b. For example,
select ret into b from my_insert('from func atx') as ret;
Using scalar function in from is not bug.
Silent assigning NULL for variables in "into" not matches same in "select"... I
think
Hello
I think limit wal in replication slots is useful in some cases. But first time
i was confused with proposed terminology secured/insecured/broken/unknown state.
patch -p1 gives some "Stripping trailing CRs from patch" messages for me, but
applied to current HEAD and builds. After little
Hello all
I would like to continue work on new recovery api proposed in thread [1]. We
have some form of consensus but thread has been inactive for a long time and i
hope i can help.
I start from last published patch [2] and make some changes:
- updated to current HEAD
- made the patch pass
Hello
> Changed the status as per last email of Tom Lane.
I have no idea what can i change.
I received no feedback how i can print parameters in other cases. Should i mark
CF as rejected?
regards, Sergei
Hello
> As mentioned by others, it sounds like we could have an option to try
> contacting the primary before running restore_commnad
Why about primary?
If we have restore_command on slave (or during point in time recovery) - we
force using XLOG_FROM_ARCHIVE, even if XLOG_FROM_PG_WAL source can
Hello
I think one restart is acceptable for such feature. I doubt user want often
disable-enable checksums. In many cases checksums will be enabled one time
during all cluster life. We need more downtimes for minor updates (4/year) and
changes in config PGC_POSTMASTER (max_connections or
Hi
> This doesn't test the consequences of the restart being skipped, nor
> does it review on a code level the correctness.
I check not only one stuff during review. I look code too: bgworker
checksumhelper.c registered with:
> bgw.bgw_start_time = BgWorkerStart_RecoveryFinished;
And then
Hi
> They fail over to a secondary to do maintenance on a primary.
But this is not problem even in current patch state. We can restart replica
before failover and it works. I tested this behavior during my review.
We can:
- call pg_enable_data_checksums() on master
- wait change data_checksums
Hello
> Yes, a bit more verbosity would be nice for that. Using the term
> "anti-wraparound" directly in the logs makes the most sense?
I used "(to prevent wraparound)" to looks like reporting in pg_stat_activity.
As far i can see currently we not using "anti-wraparound" for user messages.
On
Hello
I notice latest patch version (v06) not applied after
da6f3e45ddb68ab3161076e120e7c32cfd46d1db commit in april. We have some
consensus with design of this feature?
I will mark CF entry as waiting on author
regards, Sergei
The following review has been posted through the commitfest application:
make installcheck-world: tested, failed
Implements feature: not tested
Spec compliant: not tested
Documentation:tested, failed
Hello
As i wrote few weeks ago i can not build documentation due
Hello
Thank you for review!
Well, i can miss some cases. I'm not sure about overall design of this patch.
Is acceptable add errdetail_params to statement_timeout ereport in such way?
After shutdown signal we must be in aborted state, so we mustn't call
user-defined I/O functions. (quotation
Hello
Currently log_autovacuum_min_duration log message has no difference between
regular autovacuum and to prevent wraparound autovacuum. There are important
differences, for example, backend can automatically cancel regular autovacuum,
but not anti-wraparound. I think it is useful indicate
ith actual start background worker only at recovery end
(or promote). I think better using DEBUG ereport in postmaster and LOG in
checksumhelper.
regards, Sergei
25.07.2018, 12:35, "Daniel Gustafsson" :
>> On 24 Jul 2018, at 11:05, Sergei Kornilov wrote:
>>
>> The
Hello
23.07.2018, 17:08, "Tom Lane" :
> Sergei Kornilov writes:
>> Please test with logging command tag %i in log_line_prefix. Extended
>> protocol has three different messages, each can be canceled by timeout. But
>> here is completely no parameters in
Hello
Current patch moves recovery.conf settings into GUC system:
- if startup process found recovery.conf - it throw error
- recovery mode is turned on if new special file recovery.signal found
- standby_mode setting was removed. Standby mode can be enabled if startup
found new special file
Hi
> This parameter is defined in postgresql.conf
Huh, i believe it be in future.
Currently it is recovery.conf parameter, and yes - it can be set (or changed)
only at database start.
regards, Sergei
Hello
> Is there any reason this cannot be changed via a signal? Is it a general
> lack of infrastructure or is it there significant problem we want to ensure
> never happens?
Just make possible reload recovery.conf is not what the pgsql-hackers community
want. My patch with exactly this
Hello
> Currently pg_stat_statements_reset() and other stats reset functions like
> pg_stat_reset() can be executed only by superusers.
> But why did you allow pg_read_all_stats role to execute that function,
> by default? That change looks strange to me.
This is not behavior change, only fix
06.07.2018, 22:35, "Robert Haas" :
> On Fri, Jul 6, 2018 at 1:26 PM, Fujii Masao wrote:
>> Hmm... so pg_stat_statements_reset() is allowed to be executed by
>> pg_read_all_stats role while other stats reset functions like
>> pg_stat_reset() can be executed only by superusers. Which looks
>>
The following review has been posted through the commitfest application:
make installcheck-world: tested, passed
Implements feature: tested, passed
Spec compliant: tested, passed
Documentation:tested, passed
I reviewed and tested patch one more times, including check
Hello Stephen
I changed the suggested things and added some regression tests. Also i wrote
few words to the documentation. New patch is attached.
Thank you for feedback!
regards, Sergeidiff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index 286c7a8..cf2c56f
Hello
> You should be able to use an event trigger that raises a message when
> table_rewrite is hit, to notify the test driver that a rewrite happens.
Here is no table rewrite, only verify. So here EventTriggerTableRewrite is not
called. Or i missed something?
> ISTM that depending on DEBUG
Hello
> Do you actually need test output proving that this code path was taken
> rather than the default one? Seems like looking at the code coverage
> report might be enough.
I not known. In v4 i use DEBUG1 message and do not check code path in tests at
all: by full table scan or by
Hello
My patch does not apply after commit 5748f3a0aa7cf78ac6979010273bd9d50869bb8e.
Here is update to current master. Not null constraint is immutable too, so here
is no changes in PartConstraintImpliedByRelConstraint excepts rename and
comments fix.
In this patch version i also revert tests
Hello all!
Summarizing, how i must update tests?
We want test real skip of verify phase or only result correctness?
I can verify what i do not break alter table by tests in v4 of my patch. But
here is no check code path. So my feature may be broken in future without test
fail. alter table will
Hello, Ildar
Thanks. I looked ATTACH PARTITION tests and found such checks. If no one is
against i will just use in my patch INFO level instead DEBUG1, similarly ATTACH
PARTITION code. Updated patch attached.
Or i can rewrite tests to use DEBUG1 level.
regards, Sergeidiff --git
Hello
See for example contrib pg_stat_statements extension. Setting
pg_stat_statements.max defined as PGC_POSTMASTER
regards, Sergei
Hi
I noticed new merge conflict, updated version attached.
regards, Sergeidiff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index bd22627..db98a98 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -215,8 +215,15 @@ WITH (
Hello
I notice my patch does not apply again. Here is update to current HEAD
regards, Sergeidiff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index bd22627..db98a98 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -215,8
peter.eisentr...@2ndquadrant.com>:
> On 11/29/17 10:52, Sergei Kornilov wrote:
>> My target problem of adding NOT NULL to big relation without long downtime
>> can be done with ADD CONSTRAINT NOT VALID, VALIDATE it in second
>> transaction, then SET NOT NULL by my patch and drop un
Hello
Attached updated patch follows recent Reorganize partitioning code commit.
regards, Sergeidiff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index bd22627..db98a98 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -215,8
Hello
Something was wrong? Attached file is empty.
regards, Sergei
Hello
I can not find "Reports" in bottom any page of CF app...
Such stats covers only reviews marked in CF app? Through "Comment"->"Review"
buttons? I'm afraid this statistics will be inaccurate for new users (like me).
Wiki page https://wiki.postgresql.org/wiki/Reviewing_a_Patch say
> Send
Hello
thank you for review!
> Adding check constraint will also force the full table scan. So I think
> it would be better to rephrased it as follows:
Agree. I updated docs in new attached patch slightly different
> Regarding regression tests it may be useful to set client_min_messages
> to
Hello
Thank you for update
It may be better to use NULL as the default value at sql level.
> ereport(LOG, (errmsg("userid %u, dbid %u, queryid %ld does not exist",
> userid, dbid, queryid)));
I think LOG level is not useful here. In common case this is server log only.
How about WARNING? Or
Hello
I complete new version of this patch. I've attached it.
In this version i remove proposed recovery_target_type/recovery_target_value
and implement set of current settings: recovery_target (immediate),
recovery_target_name, recovery_target_time, recovery_target_xid,
recovery_target_lsn
Hi
Maybe you have some event trigger? Please check pg_event_trigger system view:
https://www.postgresql.org/docs/11/static/catalog-pg-event-trigger.html
regards, Sergei
Hi
> I could get behind fixing
> it to always throw the error, but that's not what Sergei was hoping for.
On the contrary i think it is reasonable way. It is much better to have error
on input value instead of finding wrong value during table processing. We
always reject this value for jsonb
Hi
We have some json regression tests in
src/test/regress/expected/json_encoding_1.out with \u symbol
select json '{ "a": "null \u escape" }' as not_unescaped;
not_unescaped
{ "a": "null \u escape" }
(1 row)
select json '{ "a":
Hi
24.10.2018, 17:40, "David G. Johnston" :
> On Wed, Oct 24, 2018 at 7:25 AM Sergei Kornilov wrote:
>
>> DETAIL: \u cannot be converted to text.
>>
>> Well, requested text type can not have \u byte. But seems strange: we
>> test json t
Hi
> OK, so the problem is COPY.
>
> Which means we have an issue with restore. We need to be able to pg_dump a
> table with generated columns, then restore it afterwards. More generally, we
> need to be able to handle data that has already been generated - the
> "generate" idea should apply
Hi
I applied this patch on top 2fe42baf7c1ad96b5f9eb898161e258315298351 commit and
found a bug while adding STORED column:
postgres=# create table test(i int);
CREATE TABLE
postgres=# insert into test values (1),(2);
INSERT 0 2
postgres=# alter table test add column gen_stored integer GENERATED
Hi
> patch --dry-run --ignore-whitespace -p 0 -F 5 <
> v5-0001-Generated-columns.patch
>
> and varied both -p and -F paramaters to no avail. Am I doing it wrong?
I am able apply patch by command
patch -p1 < v5-0001-Generated-columns.patch
or by "git apply v5-0001-Generated-columns.patch", but
Hi
I attached new version of this patch due merge conflict with pg_promote
function.
regards, Sergeidiff --git a/contrib/pg_standby/pg_standby.c b/contrib/pg_standby/pg_standby.c
index ee1fbd7..946239c 100644
--- a/contrib/pg_standby/pg_standby.c
+++ b/contrib/pg_standby/pg_standby.c
@@ -611,7
Hi
At least this is documented behavior:
> When the option list is surrounded by parentheses, the options can be written
> in any order. Without parentheses, options must be specified in exactly the
> order shown above.
https://www.postgresql.org/docs/current/static/sql-vacuum.html
Previously
Hello
We have some functions for directory listing, e.g. pg_ls_dir, pg_ls_logdir,
pg_ls_waldir. All of them call pg_ls_dir_files function
(src/backend/utils/adt/genfile.c) and i have question about per file error in
this function.
> /* Get the file info */
>
doesn't seem to
>> have any clear purpose. So, I don't see much value in breaking
>> compatibility.
>>
>> Does anyone else have an opinion on this matter?
>
> This was proposed by Sergei Kornilov in
> https://postgr.es/m/3368121530260...@web21g.yandex.ru saying th
Hello
Last published patch index-skip-fallback-v2 applied and builds cleanly.
I found reproducible crash due assert failure: FailedAssertion("!(numCols >
0)", File: "pathnode.c", Line: 2795)
> create table tablename (i int primary key);
> select distinct i from tablename where i = 1;
Query is
Hello
Thank you! Here is patch update addressing your comments.
> - useless whitespace change in xlog.c
Oops, did not notice. Fixed.
> - I think we can drop logRecoveryParameters(). Users can now inspect
> those parameters using the normal GUC mechanisms. (They were all DEBUG2
> anyway, so it's
Hello
> This patch went through the last tree commit fests without any noticeable
> activity
Well, last two CF. During march commitfest patch has activity and was marked as
ready for committer. But at end of july CF was no further activity and patch
was reverted back to "need review" with
Hello
thank you for review!
>> - if present both standby.signal and recovery.signal - we use standby mode
>> (this design from previous thread)
>
> Didn't find the discussion on that and don't understand the reasons, but
> seems OK and easy to change if we don't like it.
I meant this was
Hello, Robert
> My first question was whether TWO of them were dead code ... isn't an
> aggressive vacuum to prevent wraparound, and a vacuum to prevent
> wraparound aggressive?
Maybe i am wrong, aggressive autovacuum was your commit.
Message split was in b55509332f50f998b6e8b3830a51c5b9d8f666aa
Hi
> An autovacuum can't be just aggressive; it's either anti-wraparound or normal.
But autovacuum _can_ be aggressive and not anti-wraparound.
I build current master and can see 3 different line types:
2018-09-24 23:47:31.500 MSK 27939 @ from [vxid:4/272032 txid:0] [] LOG:
automatic
Hello
> I don't want a situation like this:
> CREATE INDEX CONCURRENTLY ...
> DROP INDEX CONCURRENTLY ...
> REINDEX INDEX (CONCURRENTLY) ...
>
> All three should be the same, and my suggestion is to add the
> parenthesized version to CREATE and DROP and not add the unparenthesized
>
Hello
> This allows the state of walrcv and startup to diverge, they could e.g. have
> different configuration, due to differently time config reloads.
So we have exactly same problem with, for example, hot_standby_feedback, right?
We change hot_standby_feedback value, reload it and we can have
Hello
> Do we no longer need static version of conninfo/slotname in
> walreceiver.c? We can detect a change of the variables without
> them (as you did in the previous version.).
Depends on this discussion:
https://www.postgresql.org/message-id/20181212053042.gk17...@paquier.xyz
If walreceiver
Hello
I am +1 for recovery_target_timeline=latest by default. This is common case in
my opinion. And first release without recovery.conf is reasonable time for
change default value.
But i doubt if we can ignore recovery_target_timeline in standby and always use
latest in standby. I have no
oops, forgot attach patch
11.12.2018, 13:14, "Sergei Kornilov" :
> Hello
>
> After some thinking i can rewrite this patch in another way.
>
> This is better or worse?
>
> regards, Sergeidiff --git a/doc/src/sgml/config.sgml b/doc/src/sgml/config.sgml
index 4a7121
Hello
After some thinking i can rewrite this patch in another way.
This is better or worse?
regards, Sergei
Hello
thank you for review!
> but I think that this patch should document clearly that if
> primary_conninfo or primary_slot_name are changed then the WAL receiver
> is stopped immediately.
Good idea, will change.
> /*
> - * replication slot name; is also used for walreceiver to connect
Hi
Not sure i can be reviewer since this was initially my proposal.
I vote +1 for this patch. Code looks good and i think we have no reason to
leave RequestXLogStreaming as-is.
regards, Sergei
Hi
> I would recommend waiting for the conclusion of other thread before
> moving on with this one.
Agreed.
I mark this patch as Waiting on Author. Not quite correct for waiting another
discussion, but most suitable.
> We could also consider using
> the show hook function of a parameter to
Hello
> OK. Rebased again, and committed (although I forgot to include a link
> to this discussion - sorry about that).
Seems we erroneously moved this thread to next CF:
https://commitfest.postgresql.org/21/1842/
Can you close this entry?
regards, Sergei
Hello
> My preference is for NULL to mean *all* so this is my favorite option,
> except that the first query should reset everything.
I am +1 for this option. NULL treat as "any" and
pg_stat_statements_reset(NULL,NULL,NULL) to reset everything
We can use named notation?
Hello
I am +1 to have separate header with path names.
Another examples are:
- pg_ctl with many hardcoded files such as "promote" (PROMOTE_SIGNAL_FILE),
"logrotate" (LOGROTATE_SIGNAL_FILE), postgresql.conf, PG_VERSION, backup_label,
postmaster.pid
- pg_basebackup with currently hardcoded
Hello
Thank you for working on this patch!
I perform some tests and think behavior with partition tables is slightly
inconsistent.
postgres=# reindex table measurement;
WARNING: REINDEX of partitioned tables is not yet implemented, skipping
"measurement"
NOTICE: table "measurement" has no
Hello
I review code and documentation and i have few notes. Did you register this
patch in CF app?
I found one error in phase 4. Simple reproducer:
> create table test (i int);
> create index this_is_very_large_exactly_maxnamelen_index_name_wink_wink_wink
> on test (i);
> create index
Hi
> Ah, I feared that some compiler would not be smart enough to be silent
> about that. I hope it does not emit a warning with this patch?
Yes, with this change build is clean. Thank you
PS: currently i use old gcc (Debian 4.9.2-10+deb8u1) 4.9.2
regards, Sergei
Hello
I can not build current HEAD cleanly. I got warning:
> postgres.c: In function ‘check_log_duration’:
> postgres.c:2254:17: warning: ‘in_sample’ may be used uninitialized in this
> function [-Wmaybe-uninitialized]
> if ((exceeded && in_sample) || log_duration)
Should not we have such
HelloThank you! Regards, Sergei
Hello all
Now we integrate recovery.conf into GUC system. So i would like to start
discussion to make primary_conninfo and primary_slot_name PGC_SIGHUP.
My work-in-progress patch attached.
I think we have no futher reason to have a copy of conninfo and slot name in
WalRcvData, so i propose
Hi
>> I think we have no futher reason to have a copy of conninfo and slot name
>> in WalRcvData, so i propose remove these fields from
>> pg_stat_get_wal_receiver() (and pg_stat_wal_receiver view). This data can be
>> accesible now via regular GUC commands.
>
> Is that wise? It seems
Hi
> The buildfarm is unhappy after this one:
> https://buildfarm.postgresql.org/cgi-bin/show_history.pl?nm=culicidae=HEAD
>
> If I use -DEXEC_BACKEND when compiling the tests complain about a
> timeout in 003_recovery_targets. Here is what the buildfarm reports, I
> can see the failure by myself
Hello
>> - recovery_target = immediate was replaced with recovery_target_immediate
>> bool GUC
>
> Why?
Due this comment:
https://www.postgresql.org/message-id/20181126172118.GY3415%40tamriel.snowman.net
> I've not been following this very closely, but seems like
> recovery_target_string is a
Hello
> I would think we'd have the different GUCs and then the check functions
> would only validate that they're valid inputs and then when we get to
> the point where we're starting to do recovery we check and make sure
> we've been given a sane overall configuration- which means that only
>
Hi
> The attached seems to be the simplest way to fix this. (Needs
> documentation updates, test updates, error message refinement.)
Thank you!
I add documentation change, tests and rephrased error message.
regards, Sergeidiff --git a/doc/src/sgml/config.sgml b/doc/src/sgml/config.sgml
index
Hello
Updated patch attached:
- added testcase to verify database does not start with multiple recovery
targets
- recovery_target = immediate was replaced with recovery_target_immediate bool
GUC
thank you!
regards, Sergeidiff --git a/doc/src/sgml/config.sgml b/doc/src/sgml/config.sgml
index
Hi
> What is the reason for allowing multiple recovery_target_* settings and
> taking the last one? Could we change this aspect to make this behave
> better?
Correct response to user configuration, similar to old recovery.conf logic or
other GUC - we allow redefine with rule "latest win".
I
Hi
>> Hmm... I considered SIGHUP processing was in fast loop and therefore
>> shutdown should be fast. But i recheck code and found a possible long loop
>> without processing SIGHUP (in case we receive new data faster than writes to
>> disk). Ok, i will revert back.
>> How about write to
Hello
> I think I would have done 'if (targetSettingsCount != 1)' here.
Streaming replication does not have any recovery_target.
Well, i can check StandbyModeRequested too and allow 0 recovery_target only for
standby mode.
>> -# Multiple targets
>> -# Last entry has priority (note that an
Hi
> Why don't we put recovery entries into postgresql.recovery.conf or such?
> And overwrite rather than append?
Appending to postgressql.auto.conf instead of writing new hardcoded file was
proposed here:
1 - 100 of 298 matches
Mail list logo