Cannot get SeaDoc working on fresh install

ouifi · March 27, 2026, 9:08pm

Hello all!

I am looking for help regarding initial setup. On my fresh instance, Seadoc and related features are not working. Wiki features do not work, and I am unable to create/edit .sdoc files.

My configuration is relatively simple, and close to the stock setup. I am using the latest Seafile 13 community edition Docker setup instructions. I consolidated the compose files into a single compose file, and deployed using Portainer’s stack feature. Because I have multiple stacks/containers running on one server IP, I use Nginx Proxy Manager (NPM) to terminate SSL and handle all incoming connections on one port.

Because of this, I had to make adjustments to the compose file to account for a reverse proxy handling SSL. I have caddy container attached to a second network, so that NPM can reach it. In several places, I had to customize the XXX_PROTOCOL or other service URLs to be http or https , depending on whether the URL was internal or external. All of the caddy labels use http because they are internal traffic. And in the seahub_settings.py, I added CSRF_TRUSTED_ORIGINS = [“https://seafile.example.com”] . These steps fixed the issue of not being able to proceed after initial login (403).

I am able to access https://seafile.example.com/sdoc-server/ and I see a page with: Welcome to sdoc-server. The current version is 2.0.9.

The symptom I see is when opening an sdoc file, the request for the document (ex, https://seafile.example.com/sdoc-server/api/v1/docs/bd262c77-0e5c-4f30-895f-4c8ef22bbae3/ hangs and eventually terminates. A similar problem happens when I try to create a page in a Wiki.

The error in sdoc-server.log that appears once the request terminates is

[2026-03-27 16:30:25] [ERROR] index.js[177] - Service unknown error
[2026-03-27 16:30:25] [ERROR] index.js[178] - error_type:  UNKNOWN_ERROR
[2026-03-27 16:30:25] [ERROR] index.js[179] - error_message:  timeout of 60000ms exceeded
[2026-03-27 16:30:25] [ERROR] index.js[180] - error_stack:  AxiosError: timeout of 60000ms exceeded
    at RedirectableRequest.handleRequestTimeout (/opt/sdoc-server/sdoc-server-2.0.9/sdoc-server/node_modules/axios/dist/node/axios.cjs:3365:16)
    at RedirectableRequest.emit (node:events:524:28)
    at Timeout.<anonymous> (/opt/sdoc-server/sdoc-server-2.0.9/sdoc-server/node_modules/follow-redirects/index.js:210:12)
    at listOnTimeout (node:internal/timers:581:17)
    at process.processTimers (node:internal/timers:519:7)
    at Axios.request (/opt/sdoc-server/sdoc-server-2.0.9/sdoc-server/node_modules/axios/dist/node/axios.cjs:4483:41)
[2026-03-27 16:30:25] [ERROR] document-controller.js[56] - Get doc download link error
[2026-03-27 16:30:25] [ERROR] document-controller.js[64] - Get doc download link error. request url is: https://seafile.example.com/api/v2.1/seadoc/download-link/bd262c77-0e5c-4f30-895f-4c8ef22bbae3/

I cannot find any cross-correlating log in seafile-data/seafile/logs.

I tried:

Setting SEAFILE_SERVICE_URL from http://seafile to https://seafile.example.com from this post: "Load doc content error" - #18 by hayato

I have not tried:

Editing the NGINX inside seafile from this post: "Load doc content error" - #7 by TechnicallyReal
- I do not want to be tinkering with an auto generated NGINX file. That just sounds like a recipe for a fragile build that I have to redo every time the container is restarted.
Setting up SeaDoc as a standalone service
- I do not believe this would fix the issue because I have proven I can access the SeaDoc server just fine from the end client.

Any tips or suggestions are helpful. Hopefully i have not missed something simple, but I am willing to look like a fool on the internet if it solves this problem, and maybe helps someone else. I really like Seafile and would like to use it for my home file management. If I cannot get Seadoc working, it is not the end of the world. From reading many forum posts and github issues trying to dig for a solution, it seems many people run into issues trying to terminate SSL outside of the seafile stack. My suggestion would be to include some additional guide on how to do an “http-only” seafile deployment for these kinds of users. Thank you all for your hard work on this project and for answering questions on this forum.

daniel.pan · March 28, 2026, 12:32am

Here is the analysis by our AI:

Based on the logs and configuration details provided, here is an analysis of the issue where SeaDoc (sdoc-server) fails to open or create files in a fresh Seafile 13 installation with a reverse proxy setup.

Problem Analysis

The core issue is a network loopback or routing failure between the sdoc-server container and the Seafile server.

According to the sdoc-server.log provided:

[ERROR] document-controller.js[64] - Get doc download link error. request url is: https://seafile.example.com/api/v2.1/seadoc/download-link/...
...
[ERROR] index.js[180] - error_stack: AxiosError: timeout of 60000ms exceeded

When a user tries to open an .sdoc file, the sdoc-server attempts to contact the Seafile API to fetch a download link. In this setup, sdoc-server is trying to reach Seafile via the external URL https://seafile.example.com. Because the environment uses Nginx Proxy Manager (NPM) and multiple network layers, the sdoc-server container is likely unable to resolve or route to its own external domain name from within the Docker network, leading to the 60-second timeout.

Key Points to Check

Internal URL Communication:
The sdoc-server needs to talk to the Seafile API. In Docker environments, it is often more reliable to use the internal container name (e.g., http://seafile:80) for service-to-service communication rather than the external HTTPS URL. Check the sdoc_server_config.json (or the environment variables passed to the SeaDoc container) to see if SEAFILE_SERVER_ROOT is pointing to the external URL.
Docker Network Loopback:
If you must use https://seafile.example.com internally, ensure the sdoc-server container can actually reach that address. You can test this by running a shell inside the container:
```
docker exec -it sdoc-server curl -v https://seafile.example.com
```
If this fails or hangs, it confirms a routing issue.
Nginx Proxy Manager (NPM) Configuration:
Since you are using NPM → Caddy → Seafile, there are multiple layers of proxying. Ensure that the headers (especially Host, X-Forwarded-For, and X-Forwarded-Proto) are being passed correctly through both NPM and Caddy. SeaDoc is sensitive to the Origin and Host headers.
DNS Resolution (extra_hosts):
A common fix for this in Portainer/Docker Compose is to add an extra_hosts entry to the sdoc-server service in your compose file, mapping seafile.example.com to the internal IP of your proxy or the Seafile container:
```
extra_hosts:
  - "seafile.example.com:172.xx.x.x" # Internal IP of the gateway or seafile container
```

Recommended Next Step

Check the configuration file for SeaDoc (usually located at /shared/seafile/conf/sdoc_server_config.json or passed as SDOC_SERVER_HOSTNAME in Docker). Ensure the internal communication URL is set to use the internal Docker network (HTTP) instead of the external HTTPS address to avoid the hairpin NAT/routing timeout.

Personally, I suggest you to remove the Caddy to avoid another middle layer in your setup.