Any suggestions on how best to manage space on the Xavier AGX. I am just setting up ROS2 Nav and even with just a base install, I am running out of space.
I have some tensor and cuda installed, but still not sure why with 32GB storage, I have 26GB only showing and that is mostly taking by the Ubuntu OS?
What am I missing?
Have you consider external storage like SSD?
My system default space as below.
nvidia@nvidia-desktop:~$ df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/mmcblk0p1 28768380 4970448 22313544 19% /
none 15958532 0 15958532 0% /dev
tmpfs 16342368 4 16342364 1% /dev/shm
tmpfs 16342368 32388 16309980 1% /run
tmpfs 5120 4 5116 1% /run/lock
tmpfs 16342368 0 16342368 0% /sys/fs/cgroup
tmpfs 3268472 160 3268312 1% /run/user/1000
You are running from external SSD or NvME?
this is my default system which is what sdkmanager loads (Jetpack 4.4)
Filesystem Size Used Avail Use% Mounted on
/dev/mmcblk0p1 28G 27G 0 100% /
none 16G 0 16G 0% /dev
tmpfs 16G 464M 16G 3% /dev/shm
tmpfs 16G 121M 16G 1% /run
tmpfs 5.0M 4.0K 5.0M 1% /run/lock
tmpfs 16G 0 16G 0% /sys/fs/cgroup
/dev/loop0 22M 22M 0 100% /snap/bashtop/126
/dev/loop1 49M 49M 0 100% /snap/core18/1888
/dev/loop2 22M 22M 0 100% /snap/bashtop/133
/dev/loop3 26M 26M 0 100% /snap/snapd/8791
tmpfs 3.2G 12K 3.2G 1% /run/user/120
tmpfs 3.2G 132K 3.2G 1% /run/user/1000
I didn’t install cuda and those software by sdkmanager.
Don’t you need it?
Could you just install ubuntu system only and install those software one by one to narrow maybe which one have large usage.
i think the big data issue is the Cuda library and the TensorRT References - a couple of GB each. But why so much other space?
When I install those and ROS2 containers, it is empty.
My only solution was to use NVMe SSD, but can’t be the final solution.
I plan to cross compile this in my production setup…