Resize compute without rebuilding cluster from scratch (Expand PATCH endpoint) - Githubissues

DataBiosphere / leonardo

Notebook service

BSD 3-Clause "New" or "Revised" License

44 stars 21 forks source link

Resize compute without rebuilding cluster from scratch (Expand PATCH endpoint) #842

Closed sharmatime closed 5 years ago

sharmatime commented 5 years ago

Note: Rob's finished the backend component of this story. The front end component is being tracked in #885

Current state:

Geetha is running an analysis and she needs more CPUs/memory/Disk space on her VM. She clicks on the "Custom" drop down in the Terra compute runtime UI, and selects more CPUs/memory/Disk, then clicks update -Update triggers a complete cluster rebuild, which takes 5-10 minutes.

Desired user outcome:

Resizing compute does not trigger a complete cluster rebuild, but instead triggers a stop/start, which will allow Geetha to access her compute much faster

Additional context: Expand cluster PATCH endpoint to include changing VM settings and enable the Terra UI to call it

CPUs
Mem
Disk
Autopause
Number of cluster nodes if Spark
Change all these by calling the PATCH endpoint, which triggers a stop/start, which will take time but won’t require a complete rebuild
Little bit of UI work to call the patch, and Leo work to do support CPU, Mem, disk. Current PATCH just has autopause and number of nodes in Spark cluster

rtitle commented 5 years ago

This is in production as of 4/24.