For dynamic languages like PHP, Python, or Ruby, the build-time and run-time environments are the same. In this case using the builder as a base image for a resulting application image is natural.
For compiled languages like C, C++, Go, or Java, the dependencies necessary for compilation might dramatically outweigh the size of the actual runtime artifacts, or provide attack surface areas that are undesirable in an application image. To keep runtime images slim, S2I enables a multiple-step build processes, where a binary artifact such as an executable or Java WAR file is created in the first builder image, extracted, and injected into a second image that simply places the executable in the correct location for execution. To give you even more abilities to customize a resulting image, S2I is also executing an assemble-runtime script inside of this (runtime) image. In this way you may do final adjustments by modifying files before an image will be committed.
The following diagram illustrates the build workflow:
To make this work S2I needs to know the following:
- builder image
- mapping of the artifacts
- runtime image
This information is specified during S2I invocation:
s2i build <repo> <builder-image> <my-app> --runtime-image <runtime-image> --runtime-artifact </path/to/artifact>
The only arguments here are the --runtime-image and --runtime-artifact options. The first option specifies the image that will be used as the base image for the final application image. The second option specifies a full path to a file (or directory) that will exist in the builder container after assembly and will be copied into the WORKDIR of the runtime container.
For our example S2I will do the following steps:
- run a builder container and invoke the assemble script (as usual)
- after the builder finishes but before stopping the builder container, download the requested artifacts from the builder and place them in a temporary directory on the host
- start a container using the specified runtime image
- upload the scripts and build artifacts from the temporary directory into
WORKDIRof the runtime container - run the
assemble-runtimescript in the runtime container - commit the runtime container and tag it as the new application image
--runtime-artifact option (or its short equivalent -a) have the format <source>[:<destination>]. Here are some example values:
| Value | Meaning |
|---|---|
/tmp/app.war |
/tmp/app.war file will be extracted from the builder container and uploaded into the WORKDIR of the runtime container |
/tmp/app.war:. |
the same as above |
/tmp/app.war:app/dist |
/tmp/app.war file will be copied from the builder container into the app/dist subdirectory of the WORKDIR of the runtime container |
/tmp/app-0.1.war:app.war |
/tmp/app-0.1.war file will be uploaded into the app.war subdirectory of the WORKDIR of the runtime container |
/opt/data |
/opt/data directory will be copied from the builder container into the WORKDIR of the runtime container |
/opt/data/ |
the same as above |
/opt/data/*.jar |
invalid mapping because wildcards are not supported. The build will fail |
You can specify this option multiple times (for example, -a /first/artifact -a /second/artifact).
The source must be an absolute path. The destination must be a relative path and it must not start with .. Because destination is always a path to a directory, it is impossible to rename artifacts during copying, you only able to choose where S2I will create this file.
When copying the artifacts, S2I will modify their permissions. All directories and files with executable bit will be uploaded with 0755 mode. Other files will have 0644 mode.
assemble-runtime can be any executable script or binary. S2I searches the following locations for this script in the following order:
- A script found at the
--scripts-urlURL - A script found in the application source
.s2i/bindirectory - A script found at the default image URL (
io.openshift.s2i.scripts-urllabel)
The assemble-runtime script is always executed as the runtime image USER.
In most cases you can use any image as a runtime image. However, if you are using the --allowed-uids option then the image must have a numeric USER specified and the value must be within the range of allowed user ids.
To simplify the build workflow and provide some reasonable defaults, the author of the runtime image can use the following techniques:
runandassemble-runtimescripts can be placed inside of the runtime image. Scripts from the image will be used as a fallback when user does not provide them in the.s2i/bindirectory of the source repository and an alternative location is not specified with the--scripts-urloption. The location of the scripts is defined by the value of theio.openshift.s2i.scripts-urllabel that should be presented on the image. For example, you can set it toimage:///usr/libexec/s2i- default mapping for the files can be specified by adding the
io.openshift.s2i.assemble-input-fileslabel to the runtime image. This mapping will be used as a fallback when the user does not specify the artifacts explicitly with the--runtime-artifactoption.
To specify mappings for multiple files, separate them with a semicolon. For example: /tmp/app.war:app;/opt/data
Builder and runtime containers have the same environment. In other words assemble and assemble-runtime scripts are able to use environment variables defined with --env and --environment-file options along with the values from .s2i/environment file in the source repository.
In the current implementation it is not possible to do an extended incremental build. This combination is invalid and the build will fail.