Forum Replies Created
-
Posts
-
Sure you can do that, but its a ssh config issue/wrong key issue, then it may not resolve the issue. GO ahead and no the new key.
Looking at the authentication logs from the bastion, your machine is present the wrong ssh key to authenticate at the bastion. From our logs the fingerprint for the key you are using is
SHA256:Oz6TG1SqPn3WXDeMmkkO+aj4sLqqgXCjScfLKc0LB2k
Where as the key registered for your account uses
SHA256:O2vzi6TM/5gF2pQaZw6IQbAD/iZt9zkkfJW8fvDlPcA (which is the SHA256 equivalent of the fingerprint you mentioned.)
So the ssh key you have is right, but what is being used by ssh is not this correct one. Please check you ssh using a -vvv to grab a verbose log of the failed attempt. That may tell us whats going on.
Do you have your latest bastion key copied on your local machine. If so, could you please provide the fingerprint the bastion key you are using ?
Can you also post your ssh_config from your machine ? From your output above looks like your ssh ke to the bastion is not right. If this key worked from jupyterhub then most likely your local ssh config may be incorrect.
@prateek. Arash is right on this. It all depends on what kinds of aplication and your experiment is. PTP is helpful when clock synchornization is needed down to a micro/milli second level between multiple nodes that may or may not be geographically distributed.
Yes there was a power outage around that time. All nodes and VM have been restored. Please try again now.
1 user thanked author for this post.
Have you tried uploading your files directly to your FABRIC VMs ?
Do you mean that firs t instrumentized the slice, then logged out and came back and tried to re-instrumentize the same slice ?
Hello,
I looked at your nodes, but as I mentioned earlier, the relevant log files are on your jupyterhub session in /tmp/mflib/*. We would require those to look into the cause. I tried the same slice using my account and have not encountered the error you mentioned.
All slices have been recovered and the site is functional again—apologies for the inconvenience caused.
Try now. My colleague Mert Cevik has restored all VMs. But FYI any data-plane interface config that was not persistent may need to be re-created
