RUNINSTALL_ENABLE_NO_GIT
: if set totrue
, It allows running tool update without any git repository. it will bypassRUNINSTALL_INCLUDES
andRUNINSTALL_EXCLUDES
RUNINSTALL_INCLUDES
: a comma seperated list of strings for repository URLs to include (fallback toRUNINSTALL_MATCH
)RUNINSTALL_EXCLUDES
: a comma seperated list of strings for repository URLs to exclude
If RUNINSTALL_INCLUDES
is undefined, or does not match, or RUNINSTALL_EXCLUDES
is matched, then all logic will be skipped.
The command will be passed through to /usr/local/bin/${cmd}
.
Example:
export RUNINSTALL_INCLUDES=org_a,org_b
mvn --version
RUNINSTALL_MATCH
is deprecated and has been replaced by RUNINSTALL_INCLUDES
.
Runinstall only works for mvn
, pipenv
and poetry
.
If any other command is used, the command will exit with an error.
In development, the command can be overridden/hardcoded with RUNINSTALL_CMD
.
Example:
export RUNINSTALL_CMD=mvn
node src/index.js --version
Runinstall will skip all remaining logic if the same command has been run in the same directory right immediately before.
It achieves this by writing the previous command to /tmp/runinstall_history
.
This can be overridden by setting RUNINSTALL_FORCE=1
, which means Runinstall will ignore the history.
Next, Runinstall will attempt to extract constraints from the cwd
.
If no constaints were found in the previous step, Runinstall will skip install command generation and not install anything.
To override this and instruct Runinstall to install every time, set RUNINSTALL_ALWAYS_INSTALL=1
.
Runinstall then maps constraints to install commands. This may require access to a github.com token in order to fetch release lists.
The token is found from either:
.gitAccessToken
in the file system in the current directory or a parent directory, or- From
RUNINSTALL_GITHUB_TOKEN
(used as a fallback)
Runinstall will then shell out to execute each install command.
Finally, runinstall will execute the desired command - passing all stdio through - before then logging the result and exiting.
Runinstall does not log to standard output or eerror, because doing so might interfere with the Unified Agent's parsing of package manager output. Instead, Runinstall sends its own logs to a remote CloudWatch Log Group if configured.
Configuration:
RUNINSTALL_DEBUG
: If defined, logs will be output to console instead of, or in addition to, CloudwatchRUNINSTALL_KEY_ID
/RUNINSTALL_ACCESS_KEY
: AWS credentials necessary for the remote Cloudwatch service
If all the above are undefined then Runinstall won't log a thing.
At the end of an invocation, Runinstall creates a log like the following:
{
"args": [
"install"
],
"cmd": "poetry",
"cwd": "/tmp/ws-scm/poetry-project",
"installCommands": [
"install-tool python 3.10.11",
"install-tool poetry 1.4.2"
],
"installSuccess": true,
"level": "info",
"message": "runinstall result",
"remote": "https://github.com/dawsonbooth/poetry-project",
"runSuccess": true,
"toolConstraints": [
{
"constraint": ">=3.8,<3.11",
"toolName": "python"
},
{
"constraint": ">=0.12",
"toolName": "poetry"
}
]
}