Hello,
I am setting up Prodigy as a labeling tool for my user. In the local prodigy.json file, I put in my IP address as host, and use 9000 as port.
Before this week, everything works fine. But since yesterday, whenever my user open the prodigy address from her browser, the prodigy service stops at my side.
Could you share some thoughts on what is the reason for this?
Also is there any way to trouble shoot this problem?
Thanks!
Hi! This is definitely strange. Could you set PRODIGY_LOGGING=basic and share the latest log entries before the process dies? Also, which version of Prodigy are you running and what is your recipe doing (roughly)? Is your recipe calling into spaCy?
One possible explanation: When the annotator opens the app in the web browser, it will make a request to the server to fetch examples from the stream generator. So this is the first time the stream generator is actually consumed. If something within the stream or the model causes something like a segmentation fault or memory error, the process might die like this.
Just to be sure, did you double-check that your machine isn’t running out of memory?
Hi Please find the screenshot for the logging.
Our dataset is very small and prodigy is the only working process.
Thanks for the help!
This is very strange! If you listen on 0.0.0.0, or on localhost, do you get the same error?
I wonder whether there’s some extra process running that ends up killing Python when the connection is made? I don’t know what it could be, but perhaps you have something running like that. Perhaps a monitoring or security daemon?
localhost works fine.
I also tried fresh restart, and the only working process is Prodigy, the same thing happens.
My colleague tested on his machine as well. The same problem occurs.
You are right about killing Python process. The moment I hit enter in the address bar, the Prodigy python process is killed.
This is the screenshot of prodigy related process. When I entered the address in the address bar, and press enter, all these three process are killed.
Could you try having some other web server listen on that address, to see if the problem is specific to Prodigy? Try python -m http.server --help
