If the user adds some bad commands in startup.sh, the run script will exit prematurely, without starting the Datalab server, which means datalab connect will fail. We should handle these errors gracefully and maybe show an alert on the tree page saying some steps did not complete in the startup script.
I also experienced that just now. I noticed the datalab instance just kept on restarting. I tried deleting the startup.sh script but it just kept on coming back. In the end I had to delete the VM.
If the user adds some bad commands in startup.sh, the run script will exit prematurely, without starting the Datalab server, which means
datalab connect
will fail. We should handle these errors gracefully and maybe show an alert on the tree page saying some steps did not complete in the startup script.This is originally an SO question here.