search

cortex-lab / phy

phy: interactive visualization and manual spike sorting of large-scale ephys data
BSD 3-Clause "New" or "Revised" License
314 stars 157 forks source link

Error due to 'Segmentation Fault' on Linux #1246

Closed paulweger closed 4 months ago

paulweger commented 8 months ago

Hey, I keep running into the issue of "Segmentation Fault". I am working on Linux and tried reinstalling the environment/phy. The following error messages appear:

(phy2) paul@spikesorter:/storage3/paul/data/1950_2/KS$ phy template-gui params.py --debug 10:54:03.701 [D] init:68 Start capturing exceptions. 10:54:03.895 [W] model:603 Unreferenced clusters found in templates (generally not a problem) 10:54:04.034 [D] model:619 Loading spike clusters. 10:54:04.181 [W] model:625 Unreferenced clusters found in spike_clusters (generally not a problem) 10:54:04.334 [D] model:569 No channel shank file found. 10:54:04.335 [D] model:692 Loading templates. 10:54:04.415 [D] model:724 Templates are dense. 10:54:04.420 [W] model:667 Skipping spike waveforms that do not exist, they will be extracted on the fly from the raw data as needed. 10:54:04.420 [D] model:730 Loading the whitening matrix. 10:54:04.420 [D] model:434 Whitening matrix file not found. 10:54:04.421 [D] model:737 Loading the inverse of the whitening matrix. 10:54:04.430 [D] model:766 Loading features. 10:54:04.430 [D] model:803 Loading template features. 10:54:04.432 [D] model:504 Load cluster_Amplitude.tsv. 10:54:04.479 [D] model:504 Load cluster_ContamPct.tsv. 10:54:04.483 [D] model:504 Load cluster_group.tsv. 10:54:04.487 [D] model:504 Load cluster_KSLabel.tsv. 10:54:05.032 [D] context:100 Initialize joblib cache dir at /storage3/paul/data/1950_2/KS/.phy. 10:54:05.032 [D] context:101 Reducing the size of the cache if needed. 10:54:05.044 [D] base:102 Add filter high_pass. 10:54:05.045 [D] config:31 Load config file /home/paul/.phy/phy_config.py. 10:54:05.047 [D] plugin:145 Loading 0 plugins. 10:54:05.049 [D] context:209 The file /storage3/paul/data/1950_2/KS/.phy/new_cluster_id.pkl doesn't exist. 10:54:05.447 [D] context:185 Save data to /storage3/paul/data/1950_2/KS/.phy/spikes_per_cluster.pkl. 10:54:05.815 [D] gui:463 Creating GUI. 10:54:05.833 [D] state:46 Load /home/paul/.phy/TemplateGUI/state.json for GUIState. Segmentation fault

Incase that helps somehow: when using another environment (without the respective libaries), Phy is opening, but showing blank (white) ClusterView and SimilarityView; other panels are black except of TraceView. Even the test-dataset provided doesn't work with both scenarios (environments).

Thank you for your help, Paul

paulweger commented 8 months ago

As mentioned in #1231, I installed phy via environment.yml and ran into the following problem with the imp libary. Did you find a solution yet?

(phy2) paul@spikesorter:~/Downloads/phy-data-master/template$ phy template-gui params.py --debug Traceback (most recent call last): File "/home/paul/miniconda3/envs/phy2/bin/phy", line 5, in from phy.apps import phycli File "/home/paul/miniconda3/envs/phy2/lib/python3.12/site-packages/phy/init.py", line 21, in from .utils.config import load_master_config File "/home/paul/miniconda3/envs/phy2/lib/python3.12/site-packages/phy/utils/init.py", line 8, in from .plugin import IPlugin, attach_plugins File "/home/paul/miniconda3/envs/phy2/lib/python3.12/site-packages/phy/utils/plugin.py", line 14, in import imp ModuleNotFoundError: No module named 'imp'

paulweger commented 8 months ago

Found a workaround using the base version environment and the command in #1143. Please inform me about any other solution if possible.

fbartsch commented 8 months ago

What exactly was your workaround? I've had the same segfault issue on Ubuntu 22.04. I just tried to run env QTWEBENGINE_CHROMIUM_FLAGS="--single-process" phy template-gui params.py as recommended in #1143 in the regular phy2 environment, but still got the same error. I likely missed something, any help is greatly appreciated.

paulweger commented 8 months ago

@fbartsch try the preexisting base environment, that somehow worked for me

fbartsch commented 8 months ago

@paulweger still not working for me, but thanks for confirming - I'll keep looking.

zm711 commented 8 months ago

You should install with the environment yaml instead. But when you download it you need to edit it and limit it to python=3.11. You should never install in base. It can break your conda install. @fbartsch , @paulweger.

I have a PR open to fix the imp issue, but I'm waiting for review and merge.

fbartsch commented 8 months ago

@zm711 Ha, that (and a quick reboot after reinstalling using the yaml) did it! Thanks so much!