Please log in or register to answer this question.

1 Answer

0 like 0 dislike
answered by (5.7k points)  

Switch Secondary/slave database to become read/write

On the secondary server create the trigger file (as specified in recovery.conf)

touch opt/PostgresPlus/9.2AS/data2/recovery.trigger

Now connect again and check if your current live sessions/new sessions to the secondary database are able to do write operations to the database. If yes, then great! You just completed a successful failover.

 

 Rebuilding the Master Database

Now, we need to re-build the master (assuming that the master database server is up).

 

First of all clean up the database cluster on old Primary server:

     rm -Rf /opt/PostgresPlus/9.2AS/data

Now, take a backup from new primary (dbserver2) to rebuild the master:

pg_basebackup -D /opt/PostgresPlus/9.2AS/data -h 192.168.160.150 -W -p 5222 -Fp --xlog-method=stream

Once the backup is complete, copy the postgresql.conf.dbserver1 as postgresql.conf and then remove recovery.done and copy the recover.conf.dbserver2_is_master as recovery.conf.

 

Start the Primary DB as new slave

Now start the db cluster on master node. Once the start-up is successful, make sure everything is alright by connecting to the database and firing the below command:

SELECT pg_is_in_recovery(); #expected output is “t”

Then fire the below command on dbserver1 and dbserver2 and both should be same:

SELECT txid_current_snapshot();

Carefully inspect the log files on secondary node (dbserver1) to confirm the recovery is in progress and there is no issues in the replication.

Related questions

0 like 0 dislike
0 answers
...