I am troubleshooting problems where the command classify-sklearn does not consistently finish running on a remote JupyterHub server using the Artifact API version of the command (I believe this is a faulty VPN connection issue, but am still trying to determine the root cause).
In the Command Line Interface version of classify-sklearn, there is a --verbose flag. Is there anything similar to a verbose flag for the Artifact API version, or another way to confirm it is still running properly? It takes ~13 minutes to finish running even on a small 2-sample test dataset, and so having an indication that the command is still running properly other than waiting for it to finish (or not) would be extremely helpful for this troubleshooting.
Thank you in advance for any advice you can give me!
Hello @geohouse. You've run into one of the most troubling issues in the entire field of computation. The halting problem. It is impossible to know if any given computation will halt, until it does halt. There is nothing you can pass to classify-sklearn to tell you whether it is still running or not.
That being said there may be something you can do to see if the process is still running, but it would involve poking around on that remote server that you may be having issues connecting to. When you run any QIIME 2 command, you should see a qiime process show up on your computer.
As long as this process exists, QIIME 2 is working on something. I don't know anything about the configuration of your remote JupyterHub server, but it may be possible for you to check whether this process is running in some way.
EDIT: Note, since you're running QIIME 2 through the artifact API, the process will probably be called python not qiime. Unfortunately, this is a less ambiguous way of determining if your specific QIIME 2 related python process is still running, but the principle remains the same.
Thanks @Oddant1 for the quick reply and helpful background information - it’s good to at least know I’m not missing an easy test or functionality in QIIME2 that could make troubleshooting this easier. The system administrator and I may try tracking specific python processes if other troubleshooting ideas don’t seem to resolve the connection problem.
