Обсуждение: [HACKERS] wal_level > WAL_LEVEL_LOGICAL
As per my understabding, current postgres server supports only three values for wal_level i.e. 'minimal' , 'replica' or 'logical'. But following error message brought to notice that there are various code spots that try to look for wal_level >= WAL_LEVEL_LOGICAL: select * from pg_create_logical_replication_slot('regression_slot', 'test_decoding'); ERROR: logical decoding requires wal_level >= logical The code locations that look for/expect wal_level > WAL_LEVEL_LOGICAL are: heapam.c 7690 * This is only used in wal_level >= WAL_LEVEL_LOGICAL, and only for cataloglogical.c 83 errmsg("logical decoding requires wal_level >= logical")));standby.c LogStandbySnapshot 950 if (wal_level >= WAL_LEVEL_LOGICAL)xlog.h XLogLogicalInfoActive 162 #define XLogLogicalInfoActive() (wal_level >= WAL_LEVEL_LOGICAL) Since postgres does not allow wal_level > WAL_LEVEL_LOGICAL, the above code locations should be modified like: s/>=/= Thoughts/Suggestions? Regards, Neha
On Mon, May 22, 2017 at 9:08 AM, Neha Khatri <nehakhatri5@gmail.com> wrote: > As per my understabding, current postgres server supports only three > values for wal_level i.e. 'minimal' , 'replica' or 'logical'. But > following error message brought to notice that there are various code > spots that try to look for wal_level >= WAL_LEVEL_LOGICAL: I suspect that this was intended as future-proofing. I think it's actually very reasonable to write the internal tests that way, but it does seem strange that it's crept into the user-visible error messages. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company
On Wed, 24 May 2017 at 10:29 pm, Robert Haas <robertmhaas@gmail.com> wrote:
On Mon, May 22, 2017 at 9:08 AM, Neha Khatri <nehakhatri5@gmail.com> wrote:
> As per my understabding, current postgres server supports only three
> values for wal_level i.e. 'minimal' , 'replica' or 'logical'. But
> following error message brought to notice that there are various code
> spots that try to look for wal_level >= WAL_LEVEL_LOGICAL:
I suspect that this was intended as future-proofing. I think it's
actually very reasonable to write the internal tests that way,
Agreed. Share the same thought and also started another thread just for the user visible error message improvement [1]. In that thread the error message is perceived to be correct.
but it
does seem strange that it's crept into the user-visible error
messages.
Yep, this seems useful for developer but not the end user.
Regards,
Neha
--
Cheers,
Neha
Neha