Hi there,

We're recently being seeing issues with both classic all-purpose compute clusters (multi node specifically) and job clusters (also multi node) failing due to STORAGE_DOWNLOAD_FAILURE_SLOW → BOOTSTRAP_TIMEOUT.

Our understanding is that this is a network/infrastructure. After some discussion with Azure support, we were recommended to update some missing network configurations. e.g. 

After the update to the network configurations, we are seeing an improvement to the all-purpose compute (multi node) but are still seeing the same issues pop up for the job compute (multi node).

I'm wondering if anyone here has seen this or something like this? It's been a real headache and would appreciate some input on this!