Re: BUG #17023: wal_log_hints not configured even if it on

Yes
I just enabled the streaming replication among 2 postgres instance and I
just enabled the option wal_log_hints at the initial level of the streaming
replication flow. So both master and stand by are configured with
wal_log_hints as on. The streaming is going good . All the DB data
replication is happening successfully.

These are the config options I have enabled in master before connecting to
the slave

wal_log_hints = on
restart_command ='cp /path to the directory/%f %p'
archive_command='.......some command'
promote_trigger_file='/tmp/switch.txt'
recovery_target_timeline='latest'

After enabled the slave the same options are enabled in slave also.
Slave also running with wal_log_hints = on and all options....

Now coming to the manual fail over case I have converted the slave as
master by using promote_trigger_file it was done successfully. The slave is
accepting the read and write also. And now I am coverting the old master as
slave for that I have done everything in new master and now by using
pg_rewind I am syncing the data time lines but that command is giving the
error like wal_log_hints should be in on . But I have enabled this option
from beginning onwards still it is in on but in target server and source
server.

On Thu, 20 May, 2021, 5:45 am Michael Paquier, <michael(at)paquier(dot)xyz> wrote:

> On Wed, May 19, 2021 at 09:22:45PM +0530, Mohan Nagandlla wrote:
> > Yes but the wal_log_hints will be in on from the begining of the server.
> > Which means it is in on when I am starting the server.
>
> Please describe more precisely the flow of what you are doing here.
> It looks like what you are missing is that wal_log_hints needs to be
> enabled on the target server, not the source. Please note that this
> needs to be reflected in the control file of the target cluster (fetch
> for "wal_log_hints setting" in the output generated by
> pg_controldata).
> --
> Michael
>