nagadit
libtinfo.so.6: no version information available
I think my install is mostly correct. The deepfacelab conda env shows my nvidia GPU in the TF test script, but when I run 1_clear_workspace.sh i get this error/warning? This is my first time trying deepfacelab and I have no idea what the expected output is from 1_clear_workspace.sh.
~/DeepFaceLab_Linux/scripts$ bash 1_clear_workspace.sh
bash: /home/mymachine/anaconda3/envs/deepfacelab/lib/libtinfo.so.6: no version information available (required by bash)
I have tried installing reinstalling libtinfo several times, restarting bash, etc. Nothing gets rid of this error. Within my native (non conda) environment, there is no issue with bash.
Just so you're not left hanging, this is also now happening to me, whereby everything WAS correctly running before. I think an update in the OS must have caused it, in my case on OpenSuse. I can't even run the Nano editor now. It might also be an anaconda update, which I am also running. The error only happens while in anaconda environment.
By chance, did you use this thread to install? https://github.com/nagadit/DeepFaceLab_Linux/issues/75
Because that's what I'm using, and it seems to start happening after running this line: echo 'export LD_LIBRARY_PATH=$CONDA_PREFIX/lib/:$CUDNN_PATH/lib:$LD_LIBRARY_PATH' >> $CONDA_PREFIX/etc/conda/activate.d/env_vars.sh
After that, even running nano I get the below: (and nano actually won't start)
nano: /home/username/anaconda3/envs/deepfacelab/lib/libtinfo.so.6: no version information available (required by nano)
nano: /home/username/anaconda3/envs/deepfacelab/lib/libtinfo.so.6: no version information available (required by nano)
nano: /home/username/anaconda3/envs/deepfacelab/lib/libtinfo.so.6: no version information available (required by nano)
nano: /home/username/anaconda3/envs/deepfacelab/lib/libncursesw.so.6: no version information available (required by nano)
nano: symbol lookup error: nano: undefined symbol: _nc_curscr, version NCURSES6_TINFO_5.7.20081102
(deepfacelab) username@OBIWAN:~/projects/DFL1/scripts> ```
It was gteachy that gave us this line, will see if I can ping them in the other thread to see if they have any clue.
Based on this page: https://stackoverflow.com/questions/72103046/libtinfo-so-6-no-version-information-available-message-using-conda-environment a workaround seems to be to rename the named files to e.g. *.bk. For me that gets it working. e.g cd into /home/username/anaconda3/envs/deepfacelab/lib and mv libncursesw.so.6 libncurses.so.6.bk etc
Unfortunately there is another problem with the update causing ffmpeg to fail, so I am now working through compiling that manually with the right flags as well, but that's another story.
Update: Fixed that, now another problem:
2023-08-26 11:08:19.243837: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT 2023-08-26 11:08:19.859979: E tensorflow/compiler/xla/stream_executor/cuda/cuda_driver.cc:266] failed call to cuInit: CUDA_ERROR_UNKNOWN: unknown error 2023-08-26 11:08:19.860049: I tensorflow/compiler/xla/stream_executor/cuda/cuda_diagnostics.cc:168] retrieving CUDA diagnostic information for host: OBIWAN 2023-08-26 11:08:19.860063: I tensorflow/compiler/xla/stream_executor/cuda/cuda_diagnostics.cc:175] hostname: OBIWAN 2023-08-26 11:08:19.860160: I tensorflow/compiler/xla/stream_executor/cuda/cuda_diagnostics.cc:199] libcuda reported version is: 535.86.10 2023-08-26 11:08:19.860193: I tensorflow/compiler/xla/stream_executor/cuda/cuda_diagnostics.cc:203] kernel reported version is: 535.86.10 2023-08-26 11:08:19.860204: I tensorflow/compiler/xla/stream_executor/cuda/cuda_diagnostics.cc:309] kernel version seems to match DSO: 535.86.10
I don't know what changed, but it's a nightmare.
Edit2 Running the below fixed it so that basic ffmpeg operations run, no idea why, but now everything running on CPU. Gotta fix that next I guess. This is nuts. conda install cuda -c nvidia
...huh. Yeah I got these errors as well. When I looked them up it was a simply annoying message it prompted so I ignored it.
I'm now realizing I don't see them anymore. I just recently refreshed my PC and reinstalled the requirements. My guess is something I had installed previously was sending this message...unfortunately I don't know what changed.
Thanks, I also took the drastic step of completely re-installing my OS - opensuse - exactly the same error afterward so it seems to be, maybe I'll have to wait and get the same improvement as you eventually.
Just reporting back, that an update came through on opensuse, including a new Nvidia driver and it all started again. So same experience as others on this front.