Installation issues

Error response from daemon: dial unix /run/docker/plugins/a-very-long-hash-value/storageos.sock: connect: connection refused

This error indicates that the StorageOS did not successfully start during installation. This is usually due to incorrect option values being passed to the docker plugin installation command.

Common causes of this issue are

  • Missing or invalid JOIN information
  • Inability to contact the StorageOS discovery service (when using a cluster token)
  • Missing -v /run/docker/plugins:/run/docker/plugins

Refer to the container install instructions, or run docker logs storageos for more debug output.

Clearing cached JOIN

When trying out StorageOS in Dev/POC environments it is common to change the value of the JOIN variable between installs. However, if a cluster has previously been made, the old JOIN value is cached and used in preference. When re-installing StorageOS, you should clear the old cached data:

$ rm -rf /var/lib/storageos*

Note this will remove any data stored on this node.

Re-using an old cluster token

When re-installing StorageOS it is possible to make the mistake of using an old cluster token from a previous install. This will cause the new StorageOS install to attempt to join the old cluster, normally resulting in a failure.

$ docker plugin install --alias storageos storageos/plugin:0.9.2 JOIN=$PREVIOUS_CLUSTER_ID
0.9.2: Pulling from storageos/plugin
a4eba3fe5636: Download complete
Digest: sha256:4f4a87e1506b7357815f574ded1ef7fd53e94683ce9d802a134019dfd8e9580a
Status: Downloaded newer image for storageos/plugin:0.9.2
Installed plugin storageos/plugin:0.9.2

$ storageos cluster health
API not responding to list nodes: API error (Service Unavailable): KV Store Unavailable