Speakr is a personal, self-hosted web application designed for transcribing audio recordings
Python
docker-publish: separate Actions cache scopes for the lite builds, no cache for the full image (cache at its 10 GB limit evicted a layer mid-build)
Murtaza Nasir8ff5b0dbea8e
diff --git a/.github/workflows/docker-publish.yml b/.github/workflows/docker-publish.ymlindex ba23d2aa..60799d68 100644--- a/.github/workflows/docker-publish.yml+++ b/.github/workflows/docker-publish.yml@@ -89,8 +89,8 @@ jobs:load: truepush: falsetags: speakr:smoke- cache-from: type=gha- cache-to: type=gha,mode=max+ cache-from: type=gha,scope=lite-amd64+ cache-to: type=gha,scope=lite-amd64,mode=maxbuild-args: |PRODUCTION=1LIGHTWEIGHT=1@@ -181,8 +181,10 @@ jobs:push: ${{ github.event_name != 'pull_request' }}tags: ${{ steps.meta.outputs.tags }}labels: ${{ steps.meta.outputs.labels }}- cache-from: type=gha- cache-to: type=gha,mode=max+ # No Actions cache for the full image. Its PyTorch layers for two+ # architectures alone fill the 10 GB cache, and a layer evicted+ # mid-build failed the v0.10.10 release build ("cache entry no longer+ # exists"). It is built only for releases.build-args: |PRODUCTION=1@@ -249,8 +251,10 @@ jobs:push: ${{ github.event_name != 'pull_request' }}tags: ${{ steps.meta-lite.outputs.tags }}labels: ${{ steps.meta-lite.outputs.labels }}- cache-from: type=gha- cache-to: type=gha,mode=max+ cache-from: |+ type=gha,scope=lite+ type=gha,scope=lite-amd64+ cache-to: type=gha,scope=lite,mode=maxbuild-args: |PRODUCTION=1LIGHTWEIGHT=1