[6/8/2016] I am going to leave the original post up to be honest about my own learning curve. Meanwhile, the better solution to this annoyance is to make an ssh call before starting the server in order to get the ~/.ssh/known_hosts file updated. That said, on to my previous misconception…
For the longest time I could not get the ssh NodeManager to work on my VirtualBox VM I use for proving out approaches. It was not a big problem to use the command line start, so I did, until recently I needed to precisely replicate the start up process used in production.
After much thumping of forehead on desk (and steering wheel, and pillow, and…) I found this nugget at http://www.pythian.com/blog/why-was-sshd-refusing-my-key/:
…In our environment, we are not using SELinux so we disabled it by setting
SELINUX=disabled
in/etc/selinux/configand rebooted the system….
A sad alternative is to go through a bit of a console dance. After starting the admin server for the first time, keep an eye in the command/bash console for:
[user@host scripts]$ The authenticity of host 'host (127.0.0.1)' can't be established. RSA key fingerprint is 78:a1:89:03:53:1c:5b:d2:8d:50:15:3f:84:b9:a2:6e. Are you sure you want to continue connecting (yes/no)?
Once this appears, type ‘yes’ (sans quotes) in the console and hit enter. Then shut down the server from the WebLogic Server console. In the command/bash console, you will see and endless series of ‘y’s. Hit CTRL+C to stop the y parade, then start the admin server again. You may need to repeat this for the managed servers as well, but eventually it will start behaving.
I have seen some other solutions in various posts, but they are not complete enough to just jump in and follow. Once I have the time and motivation to try them out, I will update this post with my findings.
© Scott S. Nelson
