How to fix this issue "no suitable node (scheduling constraints not satisfied on 1 node)" in docker swarm while deploying registry?

Question

I have a docker swarm in a virtual machine with 2 core 4GB ram Centos.

In the swarm when I deploy docker private registry (registry 2.6.4) it shows service status as pending forever. I used docker service ps <<registry_name>>

And when i inspect using docker inspect <<task_id>> in message I got this "no suitable node (scheduling constraints not satisfied on 1 node)".

I tried service restart and redeployment.

How to fix this?

Magnetometer

Did you find a solution for this? – Quacksalver

No, That setup didn't worked. I reinstalled docker setup in the machine. – Magnetometer

Answer 1

I often run into this problem when there is a mismatch between the node labels defined in the compose file and the ones defined in the actual node, either because I set a wrong label (e.g. a typo) or simply forgot to label nodes at all.

To label nodes:

1) For each target node do:

docker-machine ssh <manager_node_name> 'docker node update --label-add <label_name>=<label_value> <target_node_name>'

2) Make sure they match the ones defined in the compose file.

3) restart docker service in manager node

for example:

compose file:

 dummycontainer:
    image: group/dummyimage
    deploy:
      mode: replicated
      replicas: 1
      placement:
        constraints: [node.labels.dummy_label == dummy]
      restart_policy:
        condition: on-failure

assuming that I want to deploy this replica in a node called dummy_node:

docker-machine ssh manager_node 'docker node update --label-add dummy_label=dummy dummy_node'

and restart docker in the manager node.

Finally, if you deploy you should expect dummycontainer running in dummy_node, assuming that the label was correctly set in both steps. Otherwise it is expectable to see the error you are getting.

Best regards

Neglectful

Hi, Thanks for the response. But I got this issue in normal service create without stack. – Magnetometer

Thanks for the help. My issue was a simple typo in the compose file. mode.role == manager instead of node.role == manager – Tarn

This was my issue had to give a complete hostname. – Dobla

Answer 2

I had a similar problem while deploying service, check what is the availability of node, by docker node ls and check if nodes are not set to drain and update to active using docker node update --availability active <node-id>

which will allow swarm to run containers on the nodes for that service.

Craze

Answer 3

I encountered this error when attempting to deploy viz on a Raspberry Pi Swarm.

It turns out there is a specific version for ARM architectures:

docker service create \
  --name=viz \
  --publish=8080:8080/tcp \
  --constraint=node.role==manager \
  --mount=type=bind,src=/var/run/docker.sock,dst=/var/run/docker.sock \
  alexellis2/visualizer-arm:latest

Cashmere

This was our issue. We were building on an Apple Silicon machine and encountered this issue due to the architecture being different. – Christo

Answer 4

Another way would be to remove the reservations unless you have it set for all.

Menashem