Hello,
I have noticed a persistent issue with my deployed Streamlit app. The app suddenly crashes with the following (uninformative) message (and screenshot):
[manager] Error checking Streamlit healthz: Get "http://localhost:8501/healthz": dial tcp 127.0.0.1:8501: connect: connection refused
[manager] Streamlit server consistently failed status checks
[manager] Please fix the errors, push an update to the git repo, or reboot the app.
There is no issue with the code in the repository as I have made no changes plus the app runs perfect on locally on my device. Now, once I reboot the app, everything works perfectly again, then it crashes after a while. I cannot make sense of the error message so I can’t pinpoint the source.
Kindly advise. The log can be found here. Thank you.
Hi @Outsiders17711.
I’m suspecting a memory issue here, I’m also checking with our cloud engineers now to confirm this.
Out of curiosity, which libraries are you using?
Thanks,
Charly
Hello @Charly_Wargnier,
Thanks for your response. I am using the following libraries:
- mediapipe==0.8.6
- streamlit==0.86.0
- numpy==1.20.2
- opencv_contrib_python==4.4.0.46
Plus, I had to add the following in a packages.txt
to get the opencv
to work:
- freeglut3-dev
- libgtk2.0-dev
- libgl1-mesa-glx
- tesseract-ocr
- libtesseract-dev
- libtesseract4
- tesseract-ocr-all
Thank you.
1 Like
Thanks.
Let’s see what the cloud engineers are saying.
We’ll be back to you shortly.
Best,
Charly
1 Like
Btw @Outsiders17711, can you please provide the code + Sharing URL?
Thanks,
Charly
Hello @Charly_Wargnier,
Thanks for following up.
The code can be found at this repository. The main file is streamlitMediapipe.py
.
The sharing URL is: https://share.streamlit.io/outsiders17711/streamlit-mediapipe-webapp/main/streamlitMediapipe.py. The app is currently down. The latest logs are here.
Does that answer your question?
Thank you.
1 Like
Thanks @Outsiders17711!
Devs took a quick look - This is the log from Streamlit server, most likely a memory error:
2021-08-11 04:18:14,240 INFO exited: streamlit (terminated by SIGKILL; not expected)
We see in our metrics that the app has very quick spikes to ~3gb before crashing, so we would recommend you to try and work on the memory management of the app.
Good luck!
Best,
Charly
Helllo @Charly_Wargnier,
Thanks for your response and effort so far.
I’ll check out the memory management. Thanks a lot.
1 Like
Hello @Charly_Wargnier ,
Thanks for the pointers. I took a look at my memory management and I found out I didn’t cache the Mediapipe classes. So, the Mediapipe models were being reread into memory every time the app ran. I fixed that and did some basic garbage collection and the app has been stable every since.
Thanks’ for the help.
1 Like
That’s great to hear @Outsiders17711!
Glad I’ve been able to help you!
Best,
Charly