Nginx Configuration
Dokku uses nginx as its server for routing requests to specific applications. By default, access and error logs are written for each app to /var/log/nginx/${APP}-access.log and /var/log/nginx/${APP}-error.log respectively
nginx:access-logs <app> [-t] # Show the nginx access logs for an application (-t follows)
nginx:error-logs <app> [-t] # Show the nginx error logs for an application (-t follows)
nginx:report [<app>] [<flag>] # Displays a nginx report for one or more apps
nginx:set <app> <property> (<value>) # Set or clear an nginx property for an app
nginx:show-config <app> # Display app nginx config
nginx:validate-config [<app>] [--clean] # Validates and optionally cleans up invalid nginx configurations
Usage
Request Proxying
By default, the web process is the only process proxied by the nginx proxy implementation. Proxying to other process types may be handled by a custom nginx.conf.sigil file, as generally described below
Nginx will proxy the requests in a round-robin balancing fashion to the different deployed (scaled) containers running the web proctype. This way, the host's resources can be fully leveraged for single-threaded applications (e.g. dokku ps:scale node-js-app web=4 on a 4-core machine)
Binding to specific addresses
New
Introduced in 0.19.2
By default, nginx will listen to all interfaces ([::] for IPv6, 0.0.0.0 for IPv4) when proxying requests to applications. This may be changed using the bind-address-ipv4 and bind-address-ipv6 properties. This is useful in cases where the proxying should be internal to a network or if there are multiple network interfaces that should respond with different content.
dokku nginx:set node-js-app bind-address-ipv4 127.0.0.1
dokku nginx:set node-js-app bind-address-ipv6 ::1
This may be reverted by setting an empty bind address.
Warning
Validation is not performed on either value.
Users with apps that contain a custom nginx.conf.sigil file will need to modify the files to respect the new NGINX_BIND_ADDRESS_IPV4 and NGINX_BIND_ADDRESS_IPV6 variables.
HSTS Header
New
Introduced in 0.20.0
If SSL certificates are present, HSTS will be automatically enabled. It can be toggled via nginx:set:
The following options are also available via the nginx:set command:
hsts(type: boolean, default:true): Enables or disables HSTS for your application.hsts-include-subdomains(type: boolean, default:true): Tells the browser that the HSTS policy also applies to all subdomains of the current domain.hsts-max-age(type: integer, default:15724800): Time in seconds to cache HSTS configuration.hsts-preload(type: boolean, default:false): Tells most major web browsers to include the domain in their HSTS preload lists.
Beware that if you enable the header and a subsequent deploy of your application results in an HTTP deploy (for whatever reason), the way the header works means that a browser will not attempt to request the HTTP version of your site if the HTTPS version fails until the max-age is reached.
Globally disabling the HSTS Header
HSTS Header can be disabled for all apps by setting the hsts property to false after passing the --global flag to nginx:set. Changing this value globally or on a per-app basis will require rebuilding the nginx config via the proxy:build-config command.
Once the HSTS setting is disabled globally, it can be re-enabled on a per-app basis by setting the hsts property as normal.
Checking access logs
You may check nginx access logs via the nginx:access-logs command. This assumes that app access logs are being stored in /var/log/nginx/$APP-access.log, as is the default in the generated nginx.conf.
You may also follow the logs by specifying the -t flag.
Checking error logs
You may check nginx error logs via the nginx:error-logs command. This assumes that app error logs are being stored in /var/log/nginx/$APP-error.log, as is the default in the generated nginx.conf.
You may also follow the logs by specifying the -t flag.
Changing log path
New
Introduced in 0.20.1
The path to where log files are stored can be changed by calling the nginx:set command with the following options:
access-log-path(type: string, default:${NGINX_LOG_ROOT}/${APP}-access.log): Log path for nginx access logserror-log-path(type: string, default:${NGINX_LOG_ROOT}/${APP}-error.log): Log path for nginx error logs
The defaults should not be changed without verifying that the paths will be writeable by nginx. However, this setting is useful for enabling or disabling logging by setting the values to off.
The default value may be set by passing an empty value for the option:
In all cases, the nginx config must be regenerated after setting the above values.
Changing log format
New
Introduced in 0.22.0
The format of the access log can be changed by calling the nginx:set command as follows:
Prior to changing the log-format, log formats should be specified at a file such as /etc/nginx/conf.d/00-log-formats.conf. This will ensure they are available within your app's nginx context. For instance, the following may be added to the above file. It only needs to be specified once to be used for all apps.
# /etc/nginx/conf.d/00-log-formats.conf
# escape=json was added in nginx 1.11.8
log_format json_combined escape=json
'{'
'"time_local":"$time_local",'
'"remote_addr":"$remote_addr",'
'"remote_user":"$remote_user",'
'"request":"$request",'
'"status":"$status",'
'"body_bytes_sent":"$body_bytes_sent",'
'"request_time":"$request_time",'
'"http_referrer":"$http_referer",'
'"http_user_agent":"$http_user_agent"'
'}';
Next, the format should be set for the given app.
Finally, a proxy rebuild will change the format as desired.
Specifying a read timeout
New
Introduced in 0.21.0
When proxying requests to your applications, it may be useful to specify a proxy read timeout. This can be done via the nginx:set command as follows:
The default value is 60s, and all numeric values must have a trailing time value specified (s for seconds, m for minutes).
The default value may be set by passing an empty value for the option:
In all cases, the nginx config must be regenerated after setting the above value.
Specifying a custom client_max_body_size
New
Introduced in 0.23.0
Users can override the default client_max_body_size value - which limits file uploads - via nginx:set. Changing this value will only apply to every server stanza of the default nginx.conf.sigil; users of custom nginx.conf.sigil files must update their templates to support the new value.
The default value is empty string, which will result in nginx falling back to any configured, higher-level defaults (or 1m if unconfigued; all numerical values must have a size unit specified (k for kilobytes, m for megabytes).
The default value may be set by passing an empty value for the option:
In all cases, the nginx config must be regenerated after setting the above value.
Changing this value when using the PHP buildpack (or any other buildpack that uses an intermediary server) will require changing the value in the server config shipped with that buildpack. Consult your buildpack documentation for further details.
Showing the nginx config
For debugging purposes, it may be useful to show the nginx config. This can be achieved via the nginx:show-config command.
Validating nginx configs
It may be desired to validate an nginx config outside of the deployment process. To do so, run the nginx:validate-config command. With no arguments, this will validate all app nginx configs, one at a time. A minimal wrapper nginx config is generated for each app's nginx config, upon which nginx -t will be run.
As app nginx configs are actually executed within a shared context, it is possible for an individual config to be invalid when being validated standalone but also be valid within the global server context. As such, the exit code for the nginx:validate-config command is the exit code of nginx -t against the server's real nginx config.
The nginx:validate-config command also takes an optional --clean flag. If specified, invalid nginx configs will be removed.
Warning
Invalid app nginx config's will be removed even if the config is valid in the global server context.
The --clean flag may also be specified for a given app:
Customizing the nginx configuration
New
Introduced in 0.5.0
Dokku uses a templating library by the name of sigil to generate nginx configuration for each app. You may also provide a custom template for your application as follows:
- Copy the following example template to a file named
nginx.conf.sigiland either: - If using a buildpack application, you must check it into the root of your app repo.
- If using a dockerfile or docker image for deploys, either:
- If
WORKDIRis specified, add the file to theWORKDIRspecified in the last Dockerfile stage (example:WORKDIR /appandADD nginx.conf.sigil /app). - If no
WORKDIRis specified, add the file to the root (/) of the docker image (example:ADD nginx.conf.sigil /).
- If
Info
When using a custom nginx.conf.sigil file, depending upon your application configuration, you may be exposing the file externally. As this file is extracted before the container is run, you can, safely delete it in a custom entrypoint.sh configured in a Dockerfile ENTRYPOINT.
Info
The default template is available here, and can be used as a guide for your own, custom nginx.conf.sigil file. Please refer to the appropriate template file version for your Dokku version.
While enabled by default, using a custom nginx config can be disabled via nginx:set. This may be useful in cases where you do not want to allow users to override any higher-level customization of app nginx config.
# enable fetching custom config (default)
dokku nginx:set node-js-app disable-custom-config false
# disable fetching custom config
dokku nginx:set node-js-app disable-custom-config true
Unsetting this value is the same as enabling custom nginx config usage.
Available template variables
{{ .APP }} Application name
{{ .APP_SSL_PATH }} Path to SSL certificate and key
{{ .DOKKU_ROOT }} Global Dokku root directory (ex: app dir would be `{{ .DOKKU_ROOT }}/{{ .APP }}`)
{{ .PROXY_PORT }} Non-SSL nginx listener port (same as `DOKKU_PROXY_PORT` config var)
{{ .PROXY_SSL_PORT }} SSL nginx listener port (same as `DOKKU_PROXY_SSL_PORT` config var)
{{ .NOSSL_SERVER_NAME }} List of non-SSL VHOSTS
{{ .PROXY_PORT_MAP }} List of port mappings (same as `DOKKU_PROXY_PORT_MAP` config var)
{{ .PROXY_UPSTREAM_PORTS }} List of configured upstream ports (derived from `DOKKU_PROXY_PORT_MAP` config var)
{{ .RAW_TCP_PORTS }} List of exposed tcp ports as defined by Dockerfile `EXPOSE` directive (**Dockerfile apps only**)
{{ .SSL_INUSE }} Boolean set when an app is SSL-enabled
{{ .SSL_SERVER_NAME }} List of SSL VHOSTS
Finally, each process type has it's network listeners - a list of IP:PORT pairs for the respective app containers - exposed via an .DOKKU_APP_${PROCESS_TYPE}_LISTENERS variable - the PROCESS_TYPE will be upper-cased with hyphens transformed into underscores. Users can use the new variables to expose non-web processes via the nginx proxy.
Note
Application config variables are available for use in custom templates. To do so, use the form of {{ var "FOO" }} to access a variable named FOO.
Customizing via configuration files included by the default templates
The default nginx.conf template will include everything from your apps nginx.conf.d/ subdirectory in the main server {} block (see above):
That means you can put additional configuration in separate files. To increase the client request header timeout, the following can be performed:
mkdir /home/dokku/node-js-app/nginx.conf.d/
echo 'client_header_timeout 50s;' > /home/dokku/node-js-app/nginx.conf.d/timeout.conf
chown dokku:dokku /home/dokku/node-js-app/nginx.conf.d/upload.conf
service nginx reload
The example above uses additional configuration files directly on the Dokku host. Unlike the nginx.conf.sigil file, these additional files will not be copied over from your application repo, and thus need to be placed in the /home/dokku/node-js-app/nginx.conf.d/ directory manually.
For PHP Buildpack users, you will also need to provide a Procfile and an accompanying nginx.conf file to customize the nginx config within the container. The following are example contents for your Procfile
web: vendor/bin/heroku-php-nginx -C nginx.conf -i php.ini php/
Your nginx.conf file - not to be confused with Dokku's nginx.conf.sigil - would also need to be configured as shown in this example:
client_header_timeout 50s;
location / {
index index.php;
try_files $uri $uri/ /index.php$is_args$args;
}
Please adjust the Procfile and nginx.conf file as appropriate.
Custom Error Pages
By default, Dokku provides custom error pages for the following three categories of errors:
- 4xx: For all non-404 errors with a 4xx response code.
- 404: For "404 Not Found" errors.
- 5xx: For all 5xx error responses
These are provided as an alternative to the generic Nginx error page, are shared for all applications, and their contents are located on disk at /var/lib/dokku/data/nginx-vhosts/dokku-errors. To customize them for a specific app, create a custom nginx.conf.sigil as described above and change the paths to point elsewhere.
Default site
By default, Dokku will route any received request with an unknown HOST header value to the lexicographically first site in the nginx config stack. If this is not the desired behavior, you may want to add the following configuration to the global nginx configuration.
Create the file at /etc/nginx/conf.d/00-default-vhost.conf:
server {
listen 80 default_server;
listen [::]:80 default_server;
server_name _;
access_log off;
return 410;
}
# To handle HTTPS requests, you can uncomment the following section.
#
# Please note that in order to let this work as expected, you need a valid
# SSL certificate for any domains being served. Browsers will show SSL
# errors in all other cases.
#
# Note that the key and certificate files in the below example need to
# be copied into /etc/nginx/ssl/ folder.
#
# server {
# listen 443 ssl;
# listen [::]:443 ssl;
# server_name _;
# ssl_certificate /etc/nginx/ssl/cert.crt;
# ssl_certificate_key /etc/nginx/ssl/cert.key;
# access_log off;
# return 410;
# }
Make sure to reload nginx after creating this file by running service nginx reload.
This will catch all unknown HOST header values and return a 410 Gone response. You can replace the return 410; with return 444; which will cause nginx to not respond to requests that do not match known domains (connection refused).
The configuration file must be loaded before /etc/nginx/conf.d/dokku.conf, so it can not be arranged as a vhost in /etc/nginx/sites-enabled that is only processed afterwards.
Alternatively, you may push an app to your Dokku host with a name like "00-default". As long as it lists first in ls /home/dokku/*/nginx.conf | head, it will be used as the default nginx vhost.
Other
Domains plugin
See the domain configuration documentation for more information on how to configure domains for your app.
Customizing hostnames
See the customizing hostnames documentation for more information on how to configure domains for your app.
Disabling VHOSTS
See the disabling vhosts documentation for more information on how to disable domain usage for your app.
Running behind a load balancer
See the load balancer documentation for more information on how to configure your nginx config for running behind a network load balancer.
SSL Configuration
See the ssl documentation for more information on how to configure SSL certificates for your application.
Disabling Nginx
See the proxy documentation for more information on how to disable nginx as the proxy implementation for your app.
Managing Proxy Port mappings
See the proxy documentation for more information on how to manage ports proxied for your app.
Regenerating nginx config
See the proxy documentation for more information on how to rebuild the nginx proxy configuration for your app.