Title
#tech-support
Gholamhossein Tavasoli

Gholamhossein Tavasoli

11/04/2022, 3:07 PM
Hi everyone. I upgraded Orchest to v2022.10.9, but there is an issue in our instance. Before upgrading to this version, everything was okay. The error is
HTTP 599: Unknown (Error attempting to connect to Gateway server url '<http://jupyter-eg-00ee2e2d-5982-4d9a86d313c8-94bd-4c1e:8888/jupyter-server-00ee2e2d-5982-4d9a86d313c8-94bd-4c1e>'.  Ensure gateway url is valid and the Gateway instance is running.)
Then I upgraded it to the latest version v2022.11.0, but this issue is still there. Could you please help me?
Yannick

Yannick

11/04/2022, 3:17 PM
Did you customize JupyterLab in any way where you build a new image for it (through the settings page)? I guess the issue occurs when you open Jupyterlab? Or does it happen the moment you open a notebook?
Jacopo

Jacopo

11/04/2022, 3:30 PM
What kind of deployment is this? Minikube?
3:31 PM
If the issue happens when opening a notebook, did you do anything particular in the environment setup script that could affect it when starting as a container? Kernels run on environment images as well
Gholamhossein Tavasoli

Gholamhossein Tavasoli

11/04/2022, 4:53 PM
microk8s
4:53 PM
We did not change anything actually.
4:55 PM
We build a custom kernel base on
orchest/base-kernel-py:v2022.08.3
and
#!/bin/bash
# Install any dependencies you have in this shell script,
# see <https://docs.orchest.io/en/latest/fundamentals/environments.html#install-packages>
# E.g. mamba install -y tensorflow
#sudo apt-get update
#sudo apt-get install libpq-dev
pip install -U ./packages/*
pip install -U scikit-learn pandas
pip install -U requests pyarrow xgboost pandas-datareader
pip install -U ujson
pip install plotly==5.8.0
pip install -U matplotlib
pip install -U seaborn
pip install dateparser
pip install openpyxl
pip install xlwt
pip install -U imbalanced-learn
pip install -U psycopg2-binary
pip install tensorflow
pip install tensorflow_datasets
pip install pydot
pip install pydotplus
pip install graphviz
pip install statsmodels
pip install scipy
4:56 PM
The kernel has been built successfully.
5:06 PM
Ah, I found the issue. The issue is the difference between version of the base kernel
orchest/base-kernel-py:v2022.08.3
and Orchest version
v2022.10.9
I built another image based on the base python kernel again. It worked. Is it meaningful?
Jacopo

Jacopo

11/04/2022, 5:47 PM
The base image version shouldn't matter in this case šŸ¤”, will take a better look on Monday
šŸ‘ 1
8:41 AM
Hi @Gholamhossein Tavasoli, some more questions: ā€¢ is this a single node deployment? ā€¢ what do you get when you run
kubectl exec -n orchest -it deploy/docker-registry -- du -sh /var/lib/registry
?