Error while loading mosaic in notebook - TimeoutException: Futures timed out after [80 seconds]

[pranav_k1]

I am working on reading spatial data with mosaic and gdal Previously I used databricks mosaic = 0.3.9 version with databricks cluster = 12.2 LTS version With following command - %pip install databricks-mosaic==0.3.9 --quiet Now It's giving timeout error - TimeoutException: Futures timed out after [80 seconds] And whenever I try to run it again it gives python kernel error - jupyter client is not available because the python kernel is not defined. The kernel may be restarting or the repl may have been shut down Is there any way to avoid this error without restarting cluster every time Also I tried adding timeout as it was giving timeout error %pip install databricks-mosaic==0.3.9 --quiet --timeout 1800 This also fails with same above error I also tried using DBR 13.3 version with installing databricks using %pip install 'databricks-mosaic<0.4,>=0.3' command but this also not working How to resolve this issue, Thanks in advance

[Alberto_Umana]

Hello @pranav_k1,

I hope you are doing well!

- Have you tried to install the package via cluster libraries?

- Is this the only library not installing or failing with timeout?

- Can you please try with: %pip install databricks-mosaic==0.3.9 --verbose to obtain more insight on the failure?

- Has there been any recent change on the network, like a firewall that might be blocking outgoing requests?

- I also would like to double-check cluster configurations. 

I just tried to download the package using DBR 12.2 LTS and worked for me but took some time to download.

[pranav_k1 · Accepted solution]

Thanks for your reply

I was able to install mosaic using following command  - 

%pip install 'geopandas==0.14.3' 'databricks-mosaic<0.4,>=0.3'

Which is combination of two commands mentioned in question

Not sure of the reason but they have mentioned similar things in mosaic release documentation

https://github.com/databrickslabs/mosaic/releases 

[Alberto_Umana]

Hi @pranav_k1,

Thanks for confirming it worked for you now!

I see that the usual %pip install databricks-mosaic cannot install due to the fact that it has thus far allowed geopandas to essentially install the latest... As of geopandas==0.14.4, the version of minimum numpy is changed such that it affects limits of scikit-learn. So the workaround is %pip install geopandas==0.14.3 databricks-mosaic to fix geopandas version.