Re: [HACKERS] Check that streaming replica received all data after master shutdown

Hi,

> On Wed, Jan 14, 2015 at 2:11 AM, Heikki Linnakangas <hlinnakangas(at)vmware(dot)com
> > wrote:
>
> > On 01/13/2015 12:11 PM, Vladimir Borodin wrote:
> >
> >>
> >> 05 янв. 2015 г., в 18:15, Vladimir Borodin <root(at)simply(dot)name> написал(а):
> >>
> >> Hi all.
> >>>
> >>> I have a simple script for planned switchover of PostgreSQL (9.3 and
> >>> 9.4) master to one of its replicas. This script checks a lot of things
> >>> before doing it and one of them is that all data from master has been
> >>> received by replica that is going to be promoted. Right now the check is
> >>> done like below:
> >>>
> >>> On the master:
> >>>
> >>> postgres(at)pgtest03d ~ $ psql -t -A -c 'select
> >>> pg_current_xlog_location();'
> >>> 0/33000090
> >>> postgres(at)pgtest03d ~ $ /usr/pgsql-9.3/bin/pg_ctl stop -m fast
> >>> waiting for server to shut down.... done
> >>> server stopped
> >>> postgres(at)pgtest03d ~ $ /usr/pgsql-9.3/bin/pg_controldata | head
> >>> pg_control version number: 937
> >>> Catalog version number: 201306121
> >>> Database system identifier: 6061800518091528182
> >>> Database cluster state: shut down
> >>> pg_control last modified: Mon 05 Jan 2015 06:47:57 PM MSK
> >>> Latest checkpoint location: 0/34000028
> >>> Prior checkpoint location: 0/33000028
> >>> Latest checkpoint's REDO location: 0/34000028
> >>> Latest checkpoint's REDO WAL file: 0000001B0000000000000034
> >>> Latest checkpoint's TimeLineID: 27
> >>> postgres(at)pgtest03d ~ $
> >>>
> >>> On the replica (after shutdown of master):
> >>>
> >>> postgres(at)pgtest03g ~ $ psql -t -A -c "select
> >>> pg_xlog_location_diff(pg_last_xlog_replay_location(), '0/34000028');"
> >>> 104
> >>> postgres(at)pgtest03g ~ $
> >>>
> >>> These 104 bytes seems to be the size of shutdown checkpoint record (as I
> >>> can understand from pg_xlogdump output).
> >>>
> >>> postgres(at)pgtest03g ~/9.3/data/pg_xlog $ /usr/pgsql-9.3/bin/pg_xlogdump
> >>> -s 0/33000090 -t 27
> >>> rmgr: XLOG len (rec/tot): 0/ 32, tx: 0, lsn:
> >>> 0/33000090, prev 0/33000028, bkp: 0000, desc: xlog switch
> >>> rmgr: XLOG len (rec/tot): 72/ 104, tx: 0, lsn:
> >>> 0/34000028, prev 0/33000090, bkp: 0000, desc: checkpoint: redo 0/34000028;
> >>> tli 27; prev tli 27; fpw true; xid 0/6010; oid 54128; multi 1; offset 0;
> >>> oldest xid 1799 in DB 1; oldest multi 1 in DB 1; oldest running xid 0;
> >>> shutdown
> >>> pg_xlogdump: FATAL: error in WAL record at 0/34000028: record with zero
> >>> length at 0/34000090
> >>>
> >>> postgres(at)pgtest03g ~/9.3/data/pg_xlog $
> >>>
> >>> I’m not sure that these 104 bytes will always be 104 bytes to have a
> >>> strict equality while checking. Could it change in the future? Or is there
> >>> a better way to understand that streaming replica received all data after
> >>> master shutdown? The check that pg_xlog_location_diff returns 104 bytes
> >>> seems a bit strange.
> >>>
> >>
> > Don't rely on it being 104 bytes. It can vary across versions, and across
> > different architectures.
> >
> > You could simply check that the standby's pg_last_xlog_replay_location() >
> > master's "Latest checkpoint location", and not care about the exact
> > difference.
> >
>
> I believe there were some changes made in v9.3 which will wait for pending
> WALs to be replica​ted before a fast and smart shutdown (of master) can
> close the replication connection.
>
> http://git.postgresql.org/pg/commitdiff/985bd7d49726c9f178558491d31a570d47340459

I don't understand the relation between it and 104 bytes, it says
that the change is backpatched up to 9.1. Since it assures all
xlog records to be transferred if no trouble happens. Relying on
the mechanism, you don't need to check that if master is known to
have gracefully shut down and had no trouble around the
environment. Judging from that you want this check, I suppose
you're not guaranteed not to have trouble or not trusting the
mechanism itself.

Given the condition, as Alvaro said upthread, verifying that the
last record is a shutdown checkpoint should raise a lot the
chance for the all record being received except for the exteme
case such that the master have upped and downed while replication
connection cannot be made. For the case, I think there's no means
to confirm that by standby alone, you should at least compare the
next LSN to the last xlog record with the old master by any
means. Or doing any sanity check of the database on the standby
utilizing the nature of the data instead?

regards,

--
Kyotaro Horiguchi
NTT Open Source Software Center