Closed RehanSaeed closed 7 years ago
If you see this sh: yes: unknown operand
it means that something didn't get loaded properly on boot.
You can check with a docker ps
on the managers and there's a chance they didn't have any container deployed beyond the shell.
I believe there's a fix for it in our current build of 17.06.1-CE being tested which should be pushed out to stable soon.
In the meantime, you can try the test release of 17.07 and see if you can replicate the issue: https://editions-us-east-1.s3.amazonaws.com/aws/test/Docker.tmpl
Thanks for the update, I'll wait for the 17.06.1 bug fix template.
The new update seems to have fixed this issue but raised others. I've raised https://github.com/docker/for-aws/issues/100
Expected behavior
Start or upgrade a new swarm, all nodes should join the cluster. Also,
docker-diagnose
should return a session ID.Actual behavior
Randomly 1 or 2 of my 3 nodes fails to initialize swarm mode. Also
docker-diagnose
returns nothing due to a timeout. I think both issues may be related.Information
Copied from https://github.com/docker/for-aws/issues/85:
I confirmed that the meta-aws container is running on all three of my nodes. When I
curl {IP Address}
, I do get a response ourputting/token
. When I curl{IP Address}:9024/token/manager/
I get an error:Unfortunately, for the swarm initialization failure issue, 10.2.0.209 got blown away due to issue https://github.com/docker/for-aws/issues/52. I have three new nodes which joined the swarm correctly but still return
curl: (52) Empty reply from server
when I curl that endpoint. Here are the system logs for one of my nodes, I have also attached the syslog syslog.txt: