Rundeck setup: ssh keyfile does not exist

5,279

By default a project will include a local path to an ssh key, even if the key does not exist on the file system. You need to clear this even if you've already uploaded and selected a key for your project.

  1. Click 'configure' to the right of the project name.
  2. Click 'Simple Configuration'. Under 'Default Node Executor' clear 'SSH Key File path'.
  3. Make sure 'SSH Key Storage Path' lists the expected ssh key you uploaded.
  4. Scroll to the bottom of the config page and click 'save'.

sshkeys

Share:
5,279

Related videos on Youtube

David
Author by

David

Updated on September 18, 2022

Comments

  • David
    David over 1 year

    I'm setting up Rundeck for the first time. I'm testing by configuring a node with username="david". I've set ssh-password-storage-path="keys/foo/david" which is where I've uploaded the private keyfile in rundeck. I've tested that I can manually log in as david to the remote machine using passwordless login, with my public key in the remote machine's .ssh directory.

    It seems as though this should work on rundeck, that it would attempt to connect to the remote machine as 'david' since that's the username configuration in the node definition. Do I understand that correctly?

    But I'm getting the following error in the rundeck.log file:

    Node failures: {remote-machine-name=[ConfigurationFailure: SSH keyfile does not exist: /var/lib/rundeck/.ssh/id_rsa]}...
    

    Is this error referring to a path on the remote machine or the rundeck machine? And why would it even be looking for an id_rsa file when I've already told rundeck that it should be looking for the private key (id_rsa) in the key storageJabra Mic?

    Thanks!

  • David
    David about 8 years
    Thanks very much! Works great. Is the best practice to create a 'rundeck' user with sudo on all the remote machines?
  • jscott
    jscott about 8 years
    I can't really speak to a best practice, sorry, I'm a RunDeck novice in that regard. I was only able to answer because we had the exact issue you described during our initial deployment.