Base image to be used as source (Dockerfile:FROM) for other images. It is based on the well tested Phusion Baseimage. Applause for the baseimage-docker contributors!
Though the file structure has evolved since the fork, many tweaks and optimizations are still used from the original images. Also the whole my_init/runit stuff is still the same.
Our philosophy at CYLEDGE does not see a SSH daemon useful in a container - it's completly removed. On the other hand the syslog daemon is one of the main reasons we are using a multi-process container concept. Docker is still lacking a rich logging/output handling like - for example - journald would offer. Therefore our syslog-ng configuration is improved as described below.
Please dive into the nature of this image by reading the README of phusion/base-image
cd $base_dir
docker build --build-arg=UBUNTU_VERSION="20.04" image/
Regional information is set in prepare.sh through ONBUILD triggered ARGS as follwos:
TZ(timezone, default 'Europe/Vienna')LANGUAGE(default 'en_US:en')LOCALE(a.k.a.LANG, default 'en_US.UTF-8')
The modified syslog-ng configuration logs either to stdout of the container or to a fluentd service. To control the logging destination set the env variable "LOG_TO" to either "stdout" (default, if not set) or to "fluent".
To send logs to a fluentd service the env variables FLUENT_HOST and FLUENT_SYSLOG_PORT are used. Logs are sent as RFC5424 messages via a TCP connection and contain a faked HOSTNAME header part: It uses the value of the env variable DOCKER_HOST_NAME - which might be set to the real (bare metal) host name in order to see more valuable results in the fluentd collector. (The container ID gets also sent in the message as structured data.)
Set the targeting variables FLUENT_HOST and FLUENT_SYSLOG_PORT to match your infrastructure.
To replace the whole syslog-ng configuration set the env variable SYSLOG_CONF to a config file mounted as a volume into the container.