Greatly appreciate the continued development behind Docker Swarm.
At this time, we are evaluating Docker Swarm failover with the following setup and experiencing recovery times on the order of 40 seconds.
OS: CentOS 7.2.1511
Docker Version: 1.12.6
1)Swarm setup using Tokens re: discovery 2)NFS v3 export 3)Scality S3 Server running as a container on two physical machines Manager Worker -> S3 container with NFS mounted docker volumes of 2).
We’re finding that failing over w/Swarm using “docker stop” on one the servers in item 3), typically results in service outage on the S3 Server platform for a duration of approximately 40 seconds consistently. We are wondering if this is expected behavior or if this may be optimized?
Greatly appreciate any help/insight the Docker/Swarm team may be able to provide.