Qiskit Serverless Connection Error when switching wifi networks

[morford-oberst]

Steps to reproduce the problem

1. Instantiate a Serverless Job object using wifi network 1.
2. Connect to another wifi network (network 2).
3. The following exception is raised when calling any method (e.g. serverless_job.status()) on the Serverless Job object:

QiskitServerlessException:
| Message: Connection error. Make sure configuration (host and auth details) is correct.
| Code: AUTH1001
| Details: (MaxRetryError("HTTPSConnectionPool(host='qiskit-serverless.quantum.ibm.com', port=443): Max retries exceeded with url: /api/v1/jobs/190e3b34-62eb-489c-b03e-fee464ffa00e/?with_result=false (Caused by NewConnectionError('<urllib3.connection.HTTPSConnection object at 0x7f645359e3d0>: Failed to establish a new connection: [Errno -3] Temporary failure in name resolution'))"),)

Use the same jupyter notebook for all steps and do not restart the kernel between steps.

What is the current behavior?

If I instantiate and work with a Serverless Job object using wifi network 1, then later switch physical locations with my laptop and therefore connect to the internet with a new wifi network (network 2), I am no longer able to use the serverless job object I originally instantiated. I can no longer access the job information. I also cannot reinstantiate the serverless job object using the serverless job id, even if I reinstantiate my QiskitServerless object to instantiate a new Serverless job object. This renders a major functionality of Serverless useless since Serverless should allow users to disconnect from the Internet and come back to their work on a different network in the future. If users need to stay connected to the same network while working with Serverless, then there really is no reason to use Serverless.

What is the expected behavior?

It is expected that users can connect, disconnect, then reconnect to the internet and interact with Serverless, regardless of the network being used to connect to the internet.

[ElePT]

Thanks for your report morford-oberst! I am not 100% sure I am following the exact issue here, so I'll start describing what I consider "expected behavior":

1. If you switch networks, lose connection, close your laptop, restart your kernel, etc, the handle to the service will be lost. This is expected, and implies that your "serverless job" object won't work anymore until the handle is restored.
2. After the events in 1, you need to instantiate a new serverless client object (I don't know which one you are using, but one of IBMServerlessClient, QiskitServerless, QiskitFunctionsCatalog) with your credentials.
3. Once you have restored the handle to your service through this reauthentication/reinstantiation, you should be able to fetch your job object doing serverless.job("MY_JOB_ID").

In other words, every time your connection is lost, you must run at least something equivalent to:

from qiskit_ibm_catalog import QiskitServerless
serverless = QiskitServerless(token, instance, channel)
job = serverless.job("MY_JOB_ID")

# back to your code

This is behavior has always been supported by qiskit serverless and tested extensively. Is this what you tried? If you did, did you experience this issue more than once? You may have experienced temporary network issues that caused the errors.