On Wed, Dec 25, 2019 at 8:01 AM Michael Paquier <mich...@paquier.xyz> wrote:
> On Tue, Dec 24, 2019 at 05:29:25PM +0530, Prabhat Sahu wrote: > > While performing below operations with Master-Slave configuration, Slave > is > > crashed. > > Below are the steps to reproduce: > > > > -- create a Slave using pg_basebackup and start: > > ./pg_basebackup -v -R -D d2 -p 55510 > > mkdir /home/centos/ts1 > > > > -- Session 1(Master): > > ./psql postgres -p 55510 > > > > CREATE TABLESPACE ts1 location '/home/centos/ts1'; > > Your mistake is here. Both primary and standby are on the same host, > so CREATE TABLESPACE would point to a path that overlap for both > clusters as the tablespace path is registered the WAL replayed, > leading to various weird behaviors. What you need to do instead is to > create the tablespace before taking the base backup, and then take the > base backup using pg_basebackup's --tablespace-mapping. Thanks Michael for pointing it out, I have re-tested the scenario with "--tablespace-mapping=OLDDIR=NEWDIR" option of pg_basebackup, and now its working fine. But I think, instead of the crash, a proper error message would be better. > -- > Michael > -- With Regards, Prabhat Kumar Sahu Skype ID: prabhat.sahu1984 EnterpriseDB Software India Pvt. Ltd. The Postgres Database Company