I got an error about saving a notebook. It's just 1.1MB. I can upload this notebook but cannot save any change. I got last update recently, a few minute ago. How can I solve this ?

hi @nuhyurduseven! I think this is an instance of https://github.com/orchest/orchest/issues/1236

Ok. I can try these steps.Nb has many matplotlib plots and It's possible that is the source of the problem.

Hi @nuhyurduseven, thanks for reporting the issue, if what @juanlu suggested doesn't work, could you try the following? •

kubectl get -n orchest ingress --no-headers -o custom-columns=":metadata.name" | grep jupyter-server

-> you will get one or more ingresses depending on how many interactive sessions you are currently running, either stop the ones you aren't using or the next step needs to be applied to every ingress • apply this annotation to the ingress

kubectl patch -n orchest ingress <INGRESS NAME> -p '{"metadata":{"annotations":{"<http://nginx.ingress.kubernetes.io/proxy-body-size|nginx.ingress.kubernetes.io/proxy-body-size>":"0"}} }'

Example:

kubectl patch -n orchest ingress jupyter-server-da8e8078-791a-4e730915b350-b929-4cbd -p '{"metadata":{"annotations":{"<http://nginx.ingress.kubernetes.io/proxy-body-size|nginx.ingress.kubernetes.io/proxy-body-size>":"0"}} }'

Thanks, I can try if @juanlu's solution doesn't work.

so, if I understand correctly, • clearing the outputs makes the "entity too large" error go away, and you can save the notebook ✔️ • but

%matplotlib ipympl

does not work, even though

ipympl

is installed in the environment? ❌

yes, I have installed ipympl and ipywidgets via mamba on jupyterlab terminal. my python env is shown at left of the promt.

@nuhyurduseven my advice would be to try out the kubectl commands I suggested, at the end of the day we wouldn't want to limit the size of the notebook outputs being saved

@Jacobo de Haro I will try. I belive your configs work because it's related with nb otputs size.

That would be good to hear, we have just published a new release that should fix this exact issue

Actually, I cannot understand. why am I getting a "no named module" error? They are installed, kernel can see other modules but cannot see these.

About the particular issue with the matplotlib modules I have no idea tbh, I'll leave this to the expert @juanlu

I just installed seaborn for testing accessibility. I can access the module , it's imported. no problem for others

I think you can't just

!mamba install ipympl

in the notebook because there are some extra mechanisms involved in setting up the widgets and so on. I'll provide some more instructions in a moment

No, I didn't use notebook to install modules. I used jupyterlab terminal.

it's already released, feel free to update 🙂

Sorry, fixed now:face_with_peeking_eye:

ipympl

works if you install it in the environment setup script and JupyterLab config.

I already used setup.sh. Actually I can see the plots, it's not a problem but I cannot save the notebook which has many plots

Ah, did you update to v2022.10.7?

@Jacopo’s ingress config is works for me. I can save nb now

Ah great

@Jacopo’s ingress config is works for me. I can save nb now

Note that the fix will only work for that interactive session, if you want this change to be permanent you should update to v2022.10.7

update released 2 hours ago@Rick Lamers. You're so quick. I have updated 8 hours ago. 😄

Hehe. All @Jacopo here 👏

Btw, I used to get an error while importing ipympl. Just updated v2022.10.7 and now there is no error.

Interesting, this could have been related to max network payload.