Docker swarm NFS volumes, -


i playing docker's 1.12 swarm orchestration! there 1 issue not able find answer to:

  1. in case if you're running service nginx or redis don't worry data persistence,

  2. but if you're running service database need data persistance if happens docker instance master shuttle docker instance 1 of available nodes, default docker doesn't move data volumes other nodes address problem. can use third party plugins flocker (https://github.com/clusterhq/flocker), rexray ("https://github.com/emccode/rexray") solve issue.

  3. but problem is: when 1 node fails lose data. flocker or rexray not deal this.

  4. we can solve if use nfs. mount same volume across nodes in case don't have move data between 2 nodes. if 1 of nodes fail need remember docker mount location, can this? if can achieve docker swarm built-in orchestration!

using rexray, data stored outside docker swarm nodes (in amazon s3, openstack cinder, ...). if loose node, won't loose persistent data. if scheduler mounts new container needs data on host, retrieve external volume using rexray plugin , you're ok go.

note: external provider needs allow perform forced detach of volume unavailable old nodes.


Comments

Post a Comment