Sagemaker Studio Lab Temporary Storage Prefilled

[markoliver196]

As of approximately one week ago, my Sagemaker Studio Lab GPU runtimes have started with 26 GB of 37 GB in the /tmp partition already occupied (possibly due to preinstalled Conda envs and runtime packages). Is this an oversight or intended behavior going forward?

[markoliver196]

!df -h output from a new session without doing anything:

Filesystem      Size  Used Avail Use% Mounted on
overlay          37G   26G   12G  71% /
tmpfs            64M     0   64M   0% /dev
shm             4.0G     0  4.0G   0% /dev/shm
/dev/nvme0n1p1   80G   69G   12G  87% /mnt/sagemaker-nvme
/dev/nvme2n1     25G   16G  9.4G  63% /home/studio-lab-user
devtmpfs        4.0M     0  4.0M   0% /dev/tty
tmpfs           7.8G   12K  7.8G   1% /proc/driver/nvidia
tmpfs           3.1G  8.9M  3.1G   1% /run/nvidia-persistenced/socket
tmpfs           7.8G     0  7.8G   0% /proc/acpi
tmpfs           7.8G     0  7.8G   0% /sys/firmware

Also pinging @icoxfog417 since he was the last contributor to respond to an issue.