Dependencies are not read when I open a new pod, I use 1TB storage
It broke again, I'm wasting my time and money on this, please fix it now.
Something's wrong with RunPod. I have the dependencies in the ComfyUI venv. It crashed, and all the dependencies weren't reading. I reinstalled everything, and it worked perfectly.
I closed the pod, reopened it in a new pod running Comfyui using the same venv as before, and it has the same problem: it doesn't read the dependencies.
i work with storage, 1 TB
My commandline:
cd /workspace/ComfyUI
source venv/bin/activate
python main.py --listen 0.0.0.0 --port 9999
21 Replies
you did not included any error
root@c997c51df8a9:/# cd /workspace/ComfyUI
source venv/bin/activate
kill -9 $(ss -tulpn | grep :9999 | grep -oP 'pid=\K[0-9]+') 2>/dev/null; \
python main.py --listen 0.0.0.0 --port 7777
Traceback (most recent call last):
File "/workspace/ComfyUI/main.py", line 11, in <module>
import utils.extra_config
File "/workspace/ComfyUI/utils/extra_config.py", line 2, in <module>
import yaml
ModuleNotFoundError: No module named 'yaml'
(venv) root@c997c51df8a9:/workspace/ComfyUI# deactivate
practically the venv breaks
-
I've been working with the same storage for a month, everything was working fine, but since 2 days ago when runpod broke, now I get this error every time I run comfyui in different pods
-
(venv) root@c997c51df8a9:/workspace/ComfyUI# pip show
Traceback (most recent call last):
File "/workspace/ComfyUI/venv/bin/pip", line 5, in <module>
from pip._internal.cli.main import main
ModuleNotFoundError: No module named 'pip'
(venv) root@c997c51df8a9:/workspace/ComfyUI#
-
not even the pip works
-
with this, rtx 5090
my account: aipower954@gmail.com
Unrelated, but why are you doing this kill?
It's not the same port as your service.
Regardless, the Pytorch 2.8.0 template changed Python versions recently (this morning, and some of yesterday while we ironed out a bug)
ohhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh
Can you recreate your venv and reinstall? It should "just work" from there.
I know it's been driving you crazy, I've been seeing your messages but doing support is a bit of a commitment and we were really heads down on resolving an issue related to the template.
You would've seen this start "randomly" at about 2 hours ago yesterday, then stop like 6 hours later when we rolled back
And it would've started again this morning at like 11:27 AM Central Time.
I am the person doing this.
So this is why you're telling me, when they're in that process, is there a way to know? Because I had no idea what was going on.
i go to recreate the venv
We need to announce these changes still and indicate that it may affect workloads that are very sensitive to Python versions.
We were trying to do this yesterday, but we didn't have our bug resolved and we withdrew the changes instead of announcing.
Yeah, I think it's important that you let me know somewhere, because you rack your brains and have clients who expect me to fulfill my service, so they avoid me spamming them LOL
i go to make this
I fully understand, trust me - I wanted to make sure you got a good personal answer other than "oh just remake the venv sorry"
thanks my king
@Dj I'm wondering will it become regular event to upgrade the python version? which means every time you upgrade it, we need to recreate our python environment?
Hey guys any news on this? I know it was late last night I hope everyone got some rest
n/a, i'm still getting the same issue
Same 😭
I don't want to press too hard because I too have a whole discord community + Fiverr clients asking me when this will be fixed :(
looks like the technical team is getting buried under dozens of bugs
severe slowdowns, server outages, and now this?
if you have some pressing clients, you could move to a different template in the meantime, that's what i'm doing atm
just a heads up that we're still getting dependency issues on the official runpod pytorch templates, the same issue persists even after the push from last night. any word on what's happening behind the scenes?
I was advised to push a new version tag to docker, that build finished a little bit ago and I'm running a script to help preload it onto some machines
understood, much thanks
I'm so stupid
I just figured it out
cd ComfyUI
rm -r venv
python3.10 venv venv
source venv/bin/activate
pip install -r requirements.txt
python3.10 main.py --listen
paste that in your terminal in workspace and it fixes it
I can't believe I never re-installed requirements
but we back baby
