WIP libstore: control parallel builds with gnu make jobserver protocol #14556
+140
−0
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Motivation
Another stab at the known issue involving parallel building on powerful machines with many cores.
This is a WIP draft implementation of the GNU make jobserver protocol I saw originally mentioned as a possible solution in #11143, although the load-limit setting from that pr may be a more generalized solution for this.
Another idea for this problem could involve some mechanism for resource allocation to the builder sandboxes themselves from the perspective of the system, starting minimal and ballooning resources as needed, and sleeping when they are not available. Ideally we can have some way to set a global hard cap on resource consumption.
Hacking on this at Thaigersprint/thaigersprint-2025#2
Context
Creates nix-daemon options
use-jobserverandjobserver-tokens.Places an impure FIFO populated with
jobserver-tokenstokens when the nix-daemon is started if theuse-jobserveroption is enabled. Cleans up at termination.The jobserver FIFO is added to sandbox paths for use by derivation builders with make.
Add 👍 to pull requests you find important.
The Nix maintainer team uses a GitHub project board to schedule and track reviews.