Description
We have a huge build that uses 8 independent/parallel stages and the artifacts are copied into the final stage, however each build stage uses a lot of CPU/disk that the builder machine can not serve and eventually some of the stages fail. Rerunning the build will succeed on the second run because the build cache is already there so those stages that are already built won't be built again so the remaining stages and the image can built.
My proposal is that there should be a --multistage-parallelmax flag so we could control how many stages are built concurrently at any given time during the docker build process, default setting would be unlimited.
Description
We have a huge build that uses 8 independent/parallel stages and the artifacts are copied into the final stage, however each build stage uses a lot of CPU/disk that the builder machine can not serve and eventually some of the stages fail. Rerunning the build will succeed on the second run because the build cache is already there so those stages that are already built won't be built again so the remaining stages and the image can built.
My proposal is that there should be a
--multistage-parallelmaxflag so we could control how many stages are built concurrently at any given time during the docker build process, default setting would be unlimited.