perf: download actions as archives once per job (#1316)

Each step refreshed its own clone of an action, so a composite using one action repository ten times fetched it ten times per job, and a single network error failed the job.

Like actions/runner, a job now downloads each action repository and ref once, as an archive extracted into a directory per commit. Each action still gets its own copy in the job container. This restores the per-job dedupe of https://gitea.com/gitea/runner/pulls/1178, which only reached a since-deleted branch.

How a download works:

1. The action URL is taken after `insteadOf` rewriting, as git would use it.
2. The archive comes from `<url>/archive/<ref>.tar.gz`, or from codeload.github.com for github.com, as for actions/runner.
3. A cached tag or branch is revalidated with its ETag, which GitHub and Gitea 28+ answer with a 304. A cached commit pin needs no request.
4. Transient failures (5xx, 408, 429 honouring `Retry-After`, dropped transfers) are retried three times with a 10 to 30 second backoff.
5. Git takes over, cloning and exporting the commit into the same directory layout, when the URL is not HTTP(S), the ref is a short commit, the host serves no usable archive, the host cannot be connected to, or the retries run out.

Median of 3 runs, git (shallow clone, then fetch) against archives, on a cold and a warm cache:

| Action | git cold | git warm | archive cold | archive warm |
|---|---|---|---|---|
| github.com actions/checkout@v4 | 937ms | 726ms | 286ms | 171ms |
| github.com actions/cache@v4 | 1006ms | 801ms | 516ms | 161ms |
| github.com actions/setup-node@v4 | 1046ms | 786ms | 463ms | 156ms |
| gitea.com actions/checkout@v4 | 2007ms | 1165ms | 667ms | 222ms |

The action cache layout changes, so existing caches download once more.

*Written by Claude.*

Fixes: https://gitea.com/gitea/runner/issues/1220
Reviewed-on: https://gitea.com/gitea/runner/pulls/1316
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
This commit is contained in:
silverwind authored and bircni committed 2026-10-15 12:49:37 +00:00
1 parent 9c4c66e70f
commit b769a9765f
19 files changed
+664 -267

No files matched your search

-40
View File
@@ -21,7 +21,6 @@ import (
"testing/synctest"
"time"
"gitea.com/gitea/runner/act/common"
"gitea.com/gitea/runner/act/common/logging"
"gitea.com/gitea/runner/act/common/logging/logtest"
"gitea.com/gitea/runner/internal/pkg/lock"
@@ -426,45 +425,6 @@ func TestGitCloneExecutorOfflineMode(t *testing.T) {
})
}
func TestGitCloneExecutorQuietDemotesCloneLine(t *testing.T) {
t.Parallel()
remoteDir := t.TempDir()
require.NoError(t, gitCmd("init", "--bare", "--initial-branch=main", remoteDir))
workDir := t.TempDir()
require.NoError(t, gitCmd("clone", remoteDir, workDir))
require.NoError(t, gitCmd("-C", workDir, "checkout", "-b", "main"))
require.NoError(t, gitCmd("-C", workDir, "commit", "--allow-empty", "-m", "initial"))
require.NoError(t, gitCmd("-C", workDir, "push", "-u", "origin", "main"))
// Quiet callers report the download themselves, so the clone line must not reach the job log.
for name, quiet := range map[string]bool{"quiet": true, "not quiet": false} {
t.Run(name, func(t *testing.T) {
hook := logtest.New()
logger := logging.New(logging.NewLevelHandler(hook, logging.LevelInfo))
ctx := common.WithLogger(t.Context(), logger.With("job", "j1"))
require.NoError(t, NewGitCloneExecutor(NewGitCloneExecutorInput{
URL: remoteDir,
Ref: "main",
Dir: t.TempDir(),
Quiet: quiet,
})(ctx))
var cloneLines int
for _, entry := range hook.AllRecords() {
if strings.HasPrefix(entry.Message, "git clone ") {
cloneLines++
}
}
if quiet {
assert.Zero(t, cloneLines)
} else {
assert.Equal(t, 1, cloneLines)
}
})
}
}
func TestGitCloneExecutorShallow(t *testing.T) {
t.Parallel()
// Build a local "remote" with several commits on main plus a tag, so a full clone would pull noticeably more history than a shallow one.