I got this running on a gentoo headless server. #15241
jargoman
started this conversation in
Show and tell
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
from a barebones system. (I started with openrc-multilib stage3 tarball)
openrc init system but that probably makes no difference
I eventually added use-flag "X" globally (I considered installing fluxbox to get server running)
python3 was already installed. 3.11.8
install default gentoo kernel (or advanced users could configure their own)
emerge sys-kernel/linux-firmware
emerge sys-kernel/gentoo-kernel
nvidia drivers:
To get the driver libraries to compile I needed to add to the following line to /etc/profile:
export LANG=en_US.utf8
use flags= dist-kernel persistenced modules (I didn't use the "tools" flag, it's not needed on a headless, more on that later)
emerge x11-drivers/nvidia-drivers
I figure the script needs git to clone the repo. wget already installed on barebones gentoo
emerge dev-vcs/git
added my user to the video group!!! that one had me stumped for a while
from there I followed the instructions for linux
downloaded the webui.sh
wget -q https://raw.githubusercontent.com/AUTOMATIC1111/stable-diffusion-webui/master/webui.sh
I originally got the warning about libtcmalloc, which I couldn't figure out how to install but I found the perftools which eliminated the warning
emerge dev-util/google-perftools
I didn't use the "optimisememory" flag and instead opted for performance by leaving it out. nor did I change the paging size with use flags
changed the permissions obviously
chmod u+x webui.sh
I mentioned earlier I didn't include the "tools" flag for nvidia so that's possibly the reason I needed to skip cuda test
./webui.sh ----skip-torch-cuda-test
I then added --listen as a command line argument because it's a headless server. I used a monitor to get the install started
Beta Was this translation helpful? Give feedback.
All reactions