Re: PostgresSQL 9.5 and systemd autorestart but without replication.

RHEL 7.2, which made the postgresql-9.5.service restart on both virtual servers because of some network problem I am almost sure. 

The problem now is to make the replication server understand that he is not the master now until next failover. If I can not do this by reestablishing

recovery.conf and then restarting the postgresql.-9.5.service then I am running out of ideas. I have to make sure to be able to handle the failover's and the reestablishment of  the "replica" or standby server again when going in production on the physical servers.

Thanks

Poul 

2016-10-14 14:53 GMT+02:00 Scott Mead <scottm@openscg.com>:

On Fri, Oct 14, 2016 at 7:42 AM, Poul Kristensen <bcc5226@gmail.com> wrote:

Hi !

Hopefully this is a situation in virtuel environments and not on physical

servers with storage on SAN.as replication have to follow up in a autorestart without a manual new basebackup.

What OS are you running?  recovery.conf gets renamed to 'recovery.done' only when a promotion (failover) occurs.  This can happen a number of ways (pg_ctl promote, trigger_file, etc..).  Typically, these commands are not executed automatically for you, you would have had to either install a tool that does auto failover, or the startup scripts in your OS did something (highly doubtful).

--Scott

 

Does anyone have experience with replication on physical servers(slaves)?

 

Thanks

 

Poul

2016-10-13 17:45 GMT+02:00 Poul Kristensen <bcc5226@gmail.com>:

Thank you for fast respons.

No. The recovery.conf has now replace a recovery.done file on the slave. 

No errors in the pg_log. I have not earlier experienced this situation. 

The feature of autorestart is really nice. 

BTW: I do not quite understand why the recovery.conf are replaced with a recovery.done file.

Thanks

/Poul

2016-10-13 16:30 GMT+02:00 Hunley, Douglas <douglas.hunley@openscg.com>:

On Thu, Oct 13, 2016 at 6:00 AM, Poul Kristensen <bcc5226@gmail.com> wrote:

This morning I noticed that my 2 PG virtual servers had made an automatic

restart at exactly the same time 00:00. The 2 servers is running and accepting connections but no WALL sender and no WALL receicver is running.

Why don't replication follow up! 

Do I have to make a new pg_basebackup from primarily to replica?

OS: RHEL 7.2

Is the recovery.conf file still present and valid in $PGDATA on the slave? Any errors in the slave's pg_log directory concerning the replication?

--

{
  "name" : "douglas j hunley",
  "email" : "douglas.hunley@openscg.com",
  "title" : "database engineer",

  "phone" : "+1 732 339 3419 x163"

}

--

Med venlig hilsen / Best regards
Poul Kristensen
Linux-OS/Virtualizationexpert and Oracle DBA

--

Med venlig hilsen / Best regards
Poul Kristensen
Linux-OS/Virtualizationexpert and Oracle DBA

--

--
Scott Mead
Sr. Architect
OpenSCG

http://openscg.com

--

Med venlig hilsen / Best regards
Poul Kristensen
Linux-OS/Virtualizationexpert and Oracle DBA