cancel
Showing results for 
Search instead for 
Did you mean: 

pg_receivewal and timeline issue

SOLVED
Level 3 Adventurer

pg_receivewal and timeline issue

newly migrated database, but older BART instance with a week-old backup on it.  Trying to start WAL streaming, but receiving the following issue:

 

Tue Jul 31 09:02:01 EDT 2018 pg_receivewal is not running
pg_receivewal: could not send replication command "START_REPLICATION": ERROR:  requested timeline 27 is not in this server's history
pg_receivewal: disconnected

My timeline is much further along, and I would like to establish this properly.  Is there a way to reset the timeline, force it forward, and/or reset the BART database?

 

FYI - here is a screenshot from my EFM instance to show timeline status:

 

Every 5.0s: /usr/edb/efm-3.1/bin/efm cluster-status efm                                                                                                                            Tue Jul 31 09:07:33 2018

Cluster Status: efm

        Agent Type  Address              Agent  DB       VIP
        -----------------------------------------------------------------------
        Witness     10.200.0.90          UP     N/A      10.200.0.98
        Master      10.200.0.91          UP     UP       10.200.0.98*
        Standby     10.200.0.92          UP     UP       10.200.0.98

Allowed node host list:
        10.200.0.92 10.200.0.91 10.200.0.90

Membership coordinator: 10.200.0.90

Standby priority host list:
        10.200.0.92

Promote Status:

        DB Type     Address              XLog Loc         Info
        --------------------------------------------------------------
        Master      10.200.0.91          62/71000108
        Standby     10.200.0.92          62/71000108
2 ACCEPTED SOLUTIONS

Accepted Solutions
Level 3 Adventurer

Re: pg_receivewal and timeline issue

Determined last night that it was looking at older wal files in the archived_wal folder. Once those were destroyed, it worked fine.

Highlighted
EDB Team Member

Re: pg_receivewal and timeline issue

Hi  lgwapnitsky,

 
 
 
 
 

That sound good. 

 

 

When pg_receivewal  (known as pg_receivexlog from 9.3 - 9.6) streams a wal file the name is with extension ".partial". The name is switched back to the wal file name is when the wal file is streamed successfully.  If you try starting the process again and if there is partial file present in the archive folder the pg_receivewal process won't start. You will have to manually rsync the wal file and remove the .partial file to successfully start the process again.

 

Thank you. 

4 REPLIES
EDB Team Member

Re: pg_receivewal and timeline issue


@lgwapnitsky wrote:

newly migrated database, but older BART instance with a week-old backup on it.  Trying to start WAL streaming, but receiving the following issue:

 

Tue Jul 31 09:02:01 EDT 2018 pg_receivewal is not running
pg_receivewal: could not send replication command "START_REPLICATION": ERROR:  requested timeline 27 is not in this server's history
pg_receivewal: disconnected

My timeline is much further along, and I would like to establish this properly.  Is there a way to reset the timeline, force it forward, and/or reset the BART database?

 

FYI - here is a screenshot from my EFM instance to show timeline status:

 

Every 5.0s: /usr/edb/efm-3.1/bin/efm cluster-status efm                                                                                                                            Tue Jul 31 09:07:33 2018

Cluster Status: efm

        Agent Type  Address              Agent  DB       VIP
        -----------------------------------------------------------------------
        Witness     10.200.0.90          UP     N/A      10.200.0.98
        Master      10.200.0.91          UP     UP       10.200.0.98*
        Standby     10.200.0.92          UP     UP       10.200.0.98

Allowed node host list:
        10.200.0.92 10.200.0.91 10.200.0.90

Membership coordinator: 10.200.0.90

Standby priority host list:
        10.200.0.92

Promote Status:

        DB Type     Address              XLog Loc         Info
        --------------------------------------------------------------
        Master      10.200.0.91          62/71000108
        Standby     10.200.0.92          62/71000108


Hi lgwapnitsky,

 

Can you please share following details so that it will be easy for the investigation.

1. BART configuration

2. Show all of the master database cluster.

3. BART backup logs.

4. Complete startup pg_log for the slave.

 

 

Level 3 Adventurer

Re: pg_receivewal and timeline issue

Determined last night that it was looking at older wal files in the archived_wal folder. Once those were destroyed, it worked fine.

EDB Team Member

Re: pg_receivewal and timeline issue

Hi Larry,

 

It is nice to hear that issue has been resolved now.

 

Kindly let us know if you need any other help.

Highlighted
EDB Team Member

Re: pg_receivewal and timeline issue

Hi  lgwapnitsky,

 
 
 
 
 

That sound good. 

 

 

When pg_receivewal  (known as pg_receivexlog from 9.3 - 9.6) streams a wal file the name is with extension ".partial". The name is switched back to the wal file name is when the wal file is streamed successfully.  If you try starting the process again and if there is partial file present in the archive folder the pg_receivewal process won't start. You will have to manually rsync the wal file and remove the .partial file to successfully start the process again.

 

Thank you.