Re: Wrong HINT during database recovery when occur a minimal wal.

From: Kyotaro Horiguchi <horikyota(dot)ntt(at)gmail(dot)com>
To: lchch1990(at)sina(dot)cn
Cc: pgsql-hackers(at)postgresql(dot)org
Subject: Re: Wrong HINT during database recovery when occur a minimal wal.
Date: 2021-01-15 12:22:58
Message-ID: 20210115.212258.2262134479731399431.horikyota.ntt@gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

At Fri, 15 Jan 2021 17:04:19 +0800, "lchch1990(at)sina(dot)cn" <lchch1990(at)sina(dot)cn> wrote in
>
> >Mmm. Maybe something's missing. If you took the base-backup using
> >pg_basebackup, that means max_wal_senders > 0 on the primary. If you
> >lowered wal_level in the backup (or replica) then started it, You
> >would get something like this.
> >| FATAL: WAL streaming (max_wal_senders > 0) requires wal_level "replica" or "logical".
> >If you changed max_wal_senders to zero, you would get the following instead.
> >| FATAL: hot standby is not possible because max_wal_senders = 0 is a lower setting than on the primary server (its value was 2)
> Then mark hot_standby off and continue try lowered wal_level.
> And do recovery from the basebackup, then you will see the FATAL.
>
> >So I couldn't reproduce the situation.
> >Anyways.
>
> >> My question is that what's the mean of [set wal_level to "replica" on the primary] in
> >> HINT describe, I can't think over a case to solve this FATAL by set wal_level, I can
> >> solve it by turn off hot_standby only.
> >>
> >> Do you think we can do this code change?
> >> --- a/src/backend/access/transam/xlog.c
> >> +++ b/src/backend/access/transam/xlog.c
> >> @@ -6300,7 +6300,7 @@ CheckRequiredParameterValues(void)
> >> if (ControlFile->wal_level < WAL_LEVEL_REPLICA)
> >> ereport(ERROR,
> >> (errmsg("hot standby is not possible because wal_level was not set to \"replica\" or higher on the primary server"),
> >> - errhint("Either set wal_level to \"replica\" on the primary, or turn off hot_standby here.")));
> >> + errhint("You should turn off hot_standby here.")));
>
> >Since it's obvious that the change in a primary cannot be propagted by
> >taking a backup or starting replication, the first sentence reads to
> >me as "you should retake a base-backup from a primary where wal_level
> >is replica or higher". So *I* don't think it needs a fix.
> I think this HINT is want to guide users to finish this recovery, and the first guide is
> invalid in my opinion.

I think it's also important to suggest to the users how they can turn
on hot_standby on their standby. So, perhaps-a-bit-verbose hint would
be like this.

"Either start this standby from base backup taken after setting
wal_level to \"replica\" on the primary, or turn off hot_standby
here."

This this make sense?

regards.

--
Kyotaro Horiguchi
NTT Open Source Software Center

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Ashutosh Bapat 2021-01-15 12:23:03 Re: Determine parallel-safety of partition relations for Inserts
Previous Message Bharath Rupireddy 2021-01-15 11:55:44 Re: Logical Replication - behavior of ALTER PUBLICATION .. DROP TABLE and ALTER SUBSCRIPTION .. REFRESH PUBLICATION