Thread: Error "new timeline 2 forked off current database system timeline" in cascaded replication, when master changes

Hi,

Postgres Version - PostgreSQL 9.6.5 on x86_64-pc-linux-gnu, compiled by gcc (Debian 4.9.2-10) 4.9.2, 64-bit
Host - 4.15.0-43-generic #46~16.04.1-Ubuntu SMP Fri Dec 7 13:31:08 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux

I am experimenting with a cascaded replication between two datacenters and below is my setup (refer to attached image
ifthe alignment is not proper).
 

  ________________________________________
|  DC-1                           |
|            Server-1(master)       |
|            /\              \       |
|                    /     \Sync           \Async  |
|                  /           \             \       |
|       Server2(slave)      Server3(slave)     \        |
|____________________________________\__  |
                               \ VIP
             ___________________\_____________________
            |DC-2            /               |
            |      Server-4 (slave)                |
            |        /\                        |
            |                /     \Sync                   |
            |             /           \                           |
            |Server5(slave)      Server6(slave)                           |
            |________________________________________|


I have setup a virtual IP in DC-1 and the Server-4 in datacenter-2 is configured with upstream server as the virtual ip
address.After the initial set-up everything is working fine. 
 

Then I have stopped the server-1(master) and promoted Server-2 as master. Also configured the VIP in server 2. After
this,in the logs of Server-4, I am seeing the below error:
 

2019-02-10 21:00:25.708 PST [27602] LOG:  restarted WAL streaming at F/EC000000 on timeline 1
2019-02-10 21:00:25.840 PST [27602] LOG:  replication terminated by primary server
2019-02-10 21:00:25.840 PST [27602] DETAIL:  End of WAL reached on timeline 1 at F/EC7DD488.
2019-02-10 21:00:25.841 PST [14973] LOG:  new timeline 2 forked off current database system timeline 1 before current
recoverypoint F/EC7DD4C0
 

Server-4 did not recover from this error.

After restarting the server-4, the below error is displayed:

2019-02-10 21:01:10.131 PST [23920] FATAL:  requested timeline 2 does not contain minimum recovery point F/EC7DD4C0 on
timeline1
 
2019-02-10 21:01:10.133 PST [23915] LOG:  startup process (PID 23920) exited with exit code 1
2019-02-10 21:01:10.133 PST [23915] LOG:  aborting startup due to startup process failure
2019-02-10 21:01:10.136 PST [23915] LOG:  database system is shut down

Any pointers appreciated.
regards
Abhilash Mannathanil

Hi

do you have recovery_target_timeline = 'latest' in recovery.conf?

regards, Sergei

Hi Sergey,

Yes, I have recovery_target_timeline = 'latest' in recovery.conf

Regards
Abhilash

On 2/11/19, 1:55 AM, "Sergei Kornilov" <sk@zsrv.org> wrote:

    Hi
    
    do you have recovery_target_timeline = 'latest' in recovery.conf?
    
    regards, Sergei