By default, Dokku will constrain the amount of output displayed for any given command run. The verbosity of output can be increased by enabling trace mode. Trace mode will turn on the `set -x` flag for bash plugins, while other plugins are free to respect the environment variable `DOKKU_TRACE` and log differently as approprate. Trace mode can be useful to see _where_ plugins are running commands that would otherwise be unexpected.
Using the `EXPOSE` directive in a Dockerfile may be another reason why you might see the default site. When the `EXPOSE` directive is in use and a proxy plugin is enabled (the default), the proxy plugin will listen to requests on the ports specified in the `EXPOSE` stanza.
For example, if you have an `EXPOSE` directive like so:
```Dockerfile
EXPOSE 8000
```
The proxy port mapping will be `http:8000:8000`.
To avoid this issue, either of the following can be done:
- Remove `EXPOSE` directive: This will require respecting the `$PORT` environment variable (automatically set by Dokku). Once that change is deployed, the port mapping should be cleared via the `dokku proxy:ports-clear $APP` command (where `$APP` is your app name).
- Update the port mapping: Updating the port mapping to redirect port `80` to your app's exposed port via `dokku proxy:ports-set $APP http:80:$EXPOSED_PORT` can also fix the issue. This will also allow certificate management and the letsencrypt plugin to work correctly.
See the [port management documentation](/docs/networking/port-management.md) for more information on how Dokku exposes ports for applications and how you can configure these for your app.
The `remote rejected` error does not give enough information. Anything could have failed. Enable trace mode and begin debugging. If this does not help you, create a [gist](https://gist.github.com) containing the full log, and create an issue.
One the reasons why you may get this error is because the command that is run in the container exited (without errors). For example, (in Procfile) when you define a new worker container to run Delayed Job and use the bin/delayed_job start command. This command deamonizes the process and exists. The container thinks it's done so it closes itself. The error you get is the one above. To fix the above problem for Delayed Job, you must define the worker to user rake jobs:work, which doesn't deamonize the process.
Start a new container with the failed image and poke around (i.e. ensure you can access the internet from within the container or attempt the failed command, if known).
Sometimes (especially on DigitalOcean) deploying again seems to get past these seemingly transient issues. Additionally we've seen issues if changing networks that have different DNS resolvers. In this case, you can run the following to update your `resolv.conf`.
Certain systems may have access to the `dokku` user via SSH disabled. Please check that the `dokku` user is allowed access to the system in the file `/etc/security/access.conf`. As Dokku does not manage this file, please consult your Operating System's documentation for more information.
You get asked for a password because your SSH secret key can't be found. This may happen if the private key corresponding to the public key you added with `sshcommand acl-add` is not located in the default location `~/.ssh/id_rsa`.
it might that the cURL command that is supposed to fetch the buildpack (anything in the low megabyte file size range) takes too long to finish, due to slowish connection. To overwrite the default values (connection timeout: 90 seconds, total maximum time for operation: 600 seconds), set the following environment variables:
Another reason for this error (although it may respond immediately ruling out a timeout issue) may be because you've set the config setting `SSL_CERT_FILE`. Using a config setting with this key interferes with the buildpack's ability to download its dependencies, so you must rename the config setting to something else, e.g. `MY_APP_SSL_CERT_FILE`.
This generally occurs when the server runs out of memory. You can either add more RAM to your server or setup swap space. The follow script will create 2 GB of swap space.
This can occur if Dokku is running on a system with a firewall like UFW enabled (some OS versions like Ubuntu have this enabled by default). You can check if this is your case by running the following script:
If the previous script returned `Status: active` and a list of ports, UFW is enabled and is probably the cause of the symptom described above. To disable it, run:
This isn't usually an issue with Dokku, but rather an app config problem. This can happen when your application is configured to enforce secure connections/HSTS, but you don't have SSL set up for the app.
In Rails at least, if your `application.rb` or `environmnents/production.rb` include the line `configure.force_ssl = true` which includes HSTS, try commenting that out and redeploying.
This could be a result of a bad proxy configuration (`http:5000:5000` may be incorrect). Run `dokku proxy:report myapp` to check if your app has the correct proxy configuration. It should show something like the following.
```
=====> myapp proxy information
Proxy enabled: true
Proxy port map: http:80:5000 https:443:5000
Proxy type: nginx
```
Set `dokku proxy:ports-set front http:80:5000` to get proxy correctly configured for http endpoint.
### I deployed a new app but now subdomains are miss-routed
Sometimes nginx does something funky and cant actually reload for whatever reason.
```bash
# Validate the configuration
$ sudo nginx -t
nginx: the configuration file /etc/nginx/nginx.conf syntax is ok
nginx: configuration file /etc/nginx/nginx.conf test is successful
$ sudo service nginx restart
# Re-enable Letsencrypt
```
Example: existing AppA (https) and AppB(https), deployed NEW (non-https) then noticed right away NEW's subdomain showed AppB's content and cert but under its own subdomain (cert miss-match of course). AppB still works. AppA also changed to show AppB content and cert.. Other apps were unaffected.
Consider caddy and traefik support via docker containers if this continues to be an issue.
