I have Seafile 11.0.13 on Ubuntu 24.04 server, running in a virtual Python environment with the exact Python dependencies mentioned in the documentation with MariaDB 10.11.8 backend. The frontend is Apache 2.4.58.
There’s nothing special about this setup and by the time I upgraded from Seafile 10, it ran Ubuntu 22.04. This issue started with upgrade to Seafile 11. In about 1 in 50 cases Seahub’s GUI gets stuck showing the loading animation and after some minutes it will just say “Error” in red text.
Absolutely nothing gets left behind Seafile’s logs.
Apache’s proxy logs following:
[Thu Nov 28 20:35:38.083129 2024] [proxy:error] [pid 557065] [client ip.ad.dr.ess:33404] AH00898: Error reading from remote server returned by /api/v2.1/notifications/, referer: https://my-domain-here.com/shared-libs/
[Thu Nov 28 20:35:40.259103 2024] [proxy_http:error] [pid 557990] (70007)The timeout specified has expired: [client ip.ad.dr.ess:34690] AH01102: error reading status line from remote server 127.0.0.1:8000, referer: https://my-domain-here.com/my-libs/
[Thu Nov 28 20:35:40.259153 2024] [proxy:error] [pid 557990] [client ip.ad.dr.ess:34690] AH00898: Error reading from remote server returned by /api/v2.1/repos/, referer: https://my-domain-here.com/my-libs/
Seafile client activity never causes this error, only the Seahub usage. I tried “daemon = false” in gunicorn.conf.py and starting Seahub manually, but while it started and behave the same way, it did not give out any extra clues on what’s happening.
The server is not under any meaningful load nor there are any other issues in it. It just seems like Seahub sleeps on some requests. If reload is smashed with 2 separate browser instances, while the other gets stuck, the other keeps on responding so this isn’t like the system got slow or stuck. When it works, it works very fast.
Any pointers on debugging this further?