Managing logs when deploying Rasa Chatbots

I am aware that one should use a tracker store to log conversations, but I am wondering what I should do with the logs that normally get routed to the rasa_core.log file. Trying to deploy to AWS, we noticed that saving logs to files is a really bad practice while using container orchestration since pods come and go, and so do the files that come with them. So ideally these files should only be read, not written (such as configuration files).

For tracker stores, this is not a problem since one can use Mongo, Redis and they can be subclassed to be customized. But what solutions have been tried out there when it comes to the logs going to rasa_core.log? I am open to anything, because right now I am just “not logging” to make it work (quite literally) and this will become painful really, really soon.

Thanks!

1 Like

What would be your desired behavior? You can also just log to stdout and manage these logs with your cluster orchestration tool (e.g. + elastic).

I am a bit new to this, so I was hoping that I could rely on someone else’s experience to see if an already tested idea could also work for me, the same way that I took the MongoTrackerStore directly from Rasa but arguably with a bit more work. I tried to play a bit with the logging python library and create a custom logger that redirects the logs to a mongoDB container, but I cannot get it to work quite correctly just yet, I would have to spend a bit more time on it to make it work correctly. The question was basically about asking about other people’s experiences, since I can’t believe I’m the only one interested in what happens when my bot is running!