SLU ending up in a lock state

[nvinayvarma189]

Issue:
I added uwsgi.lock() and uwsgi.unlock() in my endpoints.py file like this. When the first exception occurs in an SLU service, the SLU will be left in a locked state. The SLU will not be attending to any following requests. This behavior is bad because, if one error occurs, then everything will fail until we redeploy.

Steps to reproduce locally:

1. If not already present, add uwsgi.lock() and uwsgi.unlock() in your local endpoints.py file like this.
2. Run task serve to run your SLU service locally
3. Intentionally, create a curl request such that it causes an exception in the PREDICT_API function of the endpoints.py file. Something like 1 divided by 0
4. Send this curl request to your local SLU
5. You should see the Exception showing up in the output of your terminal. Your SLU service should still be running though.
6. Send the same curl request again
7. You should see that you are getting an empty reply from the SLU service after 1 minute

[nvinayvarma189]

from @ltbringer

try:
    uwsgi.lock()
    this_may_raise_err()
except SomeException as err:
     return hande(err)
finally:
     uwsgi.unlock()

or use a context manager