Inject Resource data for use in pre/post deployment tasks

[troponaut]

It would be very helpful if resource-specific metadata, such as the stack name, were available as interpolatable variables during pre and post deployment shell commands.

Example Use Case

Say I want to create a stack-specific directory in a shared mount. This is useful when we try to persist data for multiple stacks that share the exact same repo.

For example:

CONF_FILE=/mnt/shared/<stack_name>/file.conf
[ -s "$CONF_FILE" ] || install -D "$CONF_FILE"

For instance, if I am deploying three traefik stacks on different servers, and want to persist say the acme.json files on a central location, being able to reference the stack name helps avoid conflicts by letting each stack write to its own folder.

Current Workaround

One workaround is to use the current working directory as a stand-in for the stack name:

STACK_NAME=$(basename "$PWD")

This can work, but it falls short when the same linked repo is used to deploy multiple identical stacks on the same server. In that case, $PWD is the same for all of them (repo cloned dir), and we can’t tell the stacks apart.

Suggestion

It would help to have the stack name (and possibly other metadata) available as environment variables during hook execution. This would make it much easier to write reliable, stack-aware logic in shell scripts.

⸻

Let me know if you want to include other metadata ideas besides the stack name.