Problems with Jupyter Notebook

Part 1 (2017)
1

Hi guys,

Every once in awhile my jupyter notebook has this popup, and I won’t be able to run anything at all:

A connection to the notebook server could not be established. The notebook will continue trying to reconnect. Check your network connection or notebook server configuration.

I’ll then have to reboot my entire aws instance and re-run my models when I start jupyter notebook again. It’s become quite frustrating esp since I have to re-train all models from the start again, has anyone encountered this problem before and found a solution to it? Really appreciate the help, thanks in advance!

2

@henrywjr, I haven’t encountered this with AWS but did a few times with FloydHub, and that was due to some spotty wifi issues.

Are you accessing AWS from your home wifi or from open network like Starbucks?

Also you can save your models at various points so you do not have to re-run from the beginning again, just from your last save. Hope that helps.

3

It may be due out of memory of the server, if your instance is some of small. Try to start an instance with more memory, or find if you can limit the memory consumption during the training.

4

@amritv I noticed that the problem happens (disconnected after approx 15min on the notebook, and during that period I can run code and models etc., ubuntu message says Connection reset by port 22) only when I’m on home networks. When I’m in public spaces on open networks, everything seems to be fine, and I’m able to operate on the notebook for hours on end with 0 disruption. Do you happen to have solved a similar problem like this before?

5

Thanks for helping Kirill! However I don’t think it’s a memory issue, have checked my AWS instance logs and everything seems to be fine. Seems to be a connection problem with private networks… :confused: