[09:14] Customer: Hi, after upgrading my self-hosted photo library last night, the mobile app stopped backing up new photos. Existing photos are still visible.

[09:15] Support: Hi! Sorry about the trouble. Which server and mobile app versions are you running, and what error appears when you try a manual upload?

[09:17] Customer: Server is 1.108.0 in Docker, Android app is 1.108.0. Manual upload says “Unable to reach server,” but I can browse the timeline normally.

[09:18] Support: Thanks. That usually means browsing requests work, but the upload endpoint is being blocked or routed incorrectly. Are you accessing the server through a reverse proxy?

[09:19] Customer: Yes, Nginx Proxy Manager. The server itself is on port 2283.

[09:21] Support: Please check the proxy host’s advanced configuration and confirm there isn’t a small request-body limit. Photo and video uploads can fail if Nginx is using its default limit. You can add:

client_max_body_size 0;

Then save the proxy configuration and restart the mobile app.

[09:24] Customer: I added that. Photos upload now, but a 2.4 GB video still fails at about 1%.

[09:25] Support: Understood. Is Cloudflare or another proxy service in front of Nginx? Those services may impose their own upload-size or timeout limits.

[09:27] Customer: Cloudflare is enabled for the domain.

[09:28] Support: That explains the larger video failure. As a quick test, temporarily set the DNS record to “DNS only” and retry the upload. If it works, you can keep uploads on a DNS-only subdomain or use a tunnel/configuration that supports files of that size.

[09:34] Customer: Switching it to DNS only fixed the video upload. Is there anything else I should check after the upgrade?

[09:35] Support: Yes—confirm all containers use the same release version, then check that the server, database, Redis, and machine-learning containers are healthy. Avoid using mixed tags such as “latest” for one container and a fixed version for another.

[09:37] Customer: They’re all pinned to 1.108.0 and showing healthy. One more thing: face recognition hasn’t processed anything since the upgrade.

[09:39] Support: Open Administration → Jobs and check whether the facial-recognition queue is paused or has failed jobs. Also verify that the machine-learning container can reach the server container over the Docker network.

[09:42] Customer: The queue was paused. I resumed it, and jobs are processing now.

[09:43] Support: Great. The backup issue was caused by Nginx’s request-body limit, the large-video failure was caused by the Cloudflare proxy limit, and facial recognition was paused. Your library should now resume normal processing.

[09:44] Customer: Everything looks good now. Thanks!

[09:44] Support: You’re welcome! Before future upgrades, I recommend backing up the database and upload directory, reviewing the release notes, and keeping all container versions aligned.