Use the same volume of old pgmaster when brings it back to the cluster
fhbenatti opened this issue · 1 comments
Guys, sorry for rising up this question again, but i need some instructions to how recreate a service for pgmaster to join in the cluster like slave.
After read some posted issues i noticed witch the correct manner to do this is removed the old service pgmaster and creates a new one:
If I understand all of this correct, you should change master to standby config before returning it in cluster or using PARTNERS_NODE
Here described how to return master to cluster:
https://stackoverflow.com/questions/37710868/how-to-promote-master-after-failover-on-postgresql-with-docker
And here we had had discussion about its process:
#178
If I'm wrong, please attach your configuration for more clear understanding.
Originally posted by @Kealman in #184 (comment)
But, is it possible to reuse a previous volume used for old master?
I always need to create a new volume to add a new service pgmaster, otherwise, i get the same error mentioned in issue #184. I want to reuse a previous volume to avoid a full backup (doing a rewind, instead) for the new instance master.
Just for testing, i'd created a new docker-compose file to create a new service for pgmaster:
AddNewNode.docker-compose.yml
pgmaster:
image: postdock/postgres:latest-postgres10-repmgr40
environment:
NODE_ID: 3
NODE_NAME: node1
CLUSTER_NODE_NETWORK_NAME: pgmaster
SSH_ENABLE: 0
PARTNER_NODES: 'pgmaster,pgslave1'
REPLICATION_PRIMARY_HOST: pgslave1
CLEAN_OVER_REWIND: 1
CONFIGS_DELIMITER_SYMBOL: ;
CONFIGS: 'max_replication_slots:10'
ports:
- 5422:5432
volumes:
- pgmaster:/var/lib/postgresql/data
- ./ssh:/home/postgres/.ssh/keys
networks:
cluster:
aliases:
- pgmaster
And then, run:
docker-compose -f "AddNewNode.docker-compose.yml" up
As i said, I got the same error mentioned in issue #184.
Is there any wrong parameter in my AddNewNode.docker-compose.yml file?
Any update about this issue please ?