RE: Disable WAL logging to speed up data loading

Hi, Movead


Thanks for your comments.
> I read the patch and have two points:
> 
> 1. I do basebackup for database then switch wal level from logical to none to
> logical and of cause I archive the wal segments. Next I do PITR base on the
> basebackup, as a result it success startup with a waring said maybe data
> missed.
This applies to wal_level=minimal and is going to be addressed by
another thread [1].

> 
> Because the 'none' level is to bulkload data, do you think it's good that we still
> support recover from a 'none' wal level.
> 
> 2. I just mark wal_level as 'none' but fail to startup, it success after I drop the
> publication and it's subscription,mark max_wal_senders as 0, drop replicate slot.
> I think it worth to write how we can startup a 'none' wal level database in
> document .
The documentation [2] touches this aspect.
It says "Also, wal_level must be set to replica or
higher to allow replication slots to be used." already.
It is documented and applies to wal_level=minimal also
because 'src/backend/replication/slot.c' has code that throws the error you got as below.

    else if (wal_level < WAL_LEVEL_REPLICA)
        ereport(FATAL,
                (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
                 errmsg("physical replication slot \"%s\" exists, but wal_level < replica",
                        NameStr(cp.slotdata.name)),
                 errhint("Change wal_level to be replica or higher.")));

[1] -
https://www.postgresql.org/message-id/OSBPR01MB4888CBE1DA08818FD2D90ED8EDF90%40OSBPR01MB4888.jpnprd01.prod.outlook.com
[2] - https://www.postgresql.org/docs/13/runtime-config-replication.html

Best Regards,
    Takamichi Osumi