Determining conflicts: 0%| | 0/6 [00:00<?, ?it/s]e[91m
Examining conflict for python __glibc: 0%| | 0/6 [00:00<?, ?it/s]e[91m
Examining conflict for python xformers: 17%|█▋ | 1/6 [00:00<00:00, 33.85it/s]e[91m
Found conflicts! Looking for incompatible packages.
This can take several minutes. Press CTRL-C to abort.
UnsatisfiableError: The following specifications were found to be incompatible with each other:
Output in format: Requested package -> Available versionsThe following specifications were found to be incompatible with your system:
- python=3.10 -> libgcc-ng[version='>=11.2.0'] -> __glibc[version='>=2.17']
Your installed version is: 2.27
e[91mERROR conda.cli.main_run:execute(49): `conda run /bin/bash -c conda install xformers -c xformers/label/dev` failed. (See above for error)
The command '/opt/conda/bin/conda run --no-capture-output -n xformers /bin/bash -c conda install xformers -c xformers/label/dev' returned a non-zero code: 1
INTERNAL ERROR: Build server encountered an unexpected error
What is going on here? Why did suddenly did stop working after working just fine for several weeks maybe months?
Hey, haven’t changed anything my side, there actually haven’t been any new releases in a while. Is this on banana, with a Dockerfile based on docker-diffusers-api-build-download? Tracking latest/main/dev?
I just rebuilt from scratch locally without any issues, but of course a lot of variations are possible. Could you perhaps attach your full Dockerfile (with any secret info removed), and likewise for the full log above in case there’s any more info above the error, which could help me understand what’s going on.
Been trying to think how this is even possible… is this perhaps the old v0 version of docker-diffusers-api? In which case, it looks like maybe the latest dev release of xformers is breaking the build due to some incompatibility with glibc and python 3.10. (In v1, with the new split image architecture, these kinds of things can’t really happen).
If that is the case (you’re using v0), the “best” option is probably to try upgrade to v1 if you can, which won’t have this issue, and is better supported. However, for a “quick fix”, you could try changing the following line in the Dockerfile, to use a stable release of xformers rather than their latest dev release (which seems to have this issue):
# RUN conda install xformers -c xformers/label/dev # comment out this
RUN conda install xformers -c xformers # replace with this
which hopefully will fix the issue. (Back when I wrote the original code, the non-dev releases weren’t usable for diffusers yet).
In any event, let me know exactly which variants and versions of docker-diffusers-api you’re using so I can be sure, but hopefully if you’re still on v0, the above should be all you need. Hopefully! Keep me posted