René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Overlay-base cache restore misses on cache backends that only prefix-match restoreKeys","articleBody":"### Summary\n\n`downloadOverlayBaseDatabaseFromCache` restores the overlay-base database by passing the key **prefix** as the primary key, with no `restoreKeys`:\n\nhttps://github.com/github/codeql-action/blob/d208ec71e39ae46397cbf6569bef8439b9aab974/src/overlay/caching.ts#L290-L302\n\n```ts\nactionsCache.restoreCache(\n  [dbLocation],\n  cacheRestoreKeyPrefix,\n  undefined,\n  { segmentTimeoutInMs: 3000 },\n)\n```\n\nThis depends on the primary key matching by prefix. GitHub's cache service does that, but some Actions-cache-compatible backends only prefix-match `restoreKeys`. One example is the transparent cache on Blacksmith runners. On those backends, PR runs can never restore the overlay-base database, even though default-branch runs upload it successfully. Every PR run then falls back to a full analysis.\n\n### What we see\n\nDefault-branch run, on the same runner type:\n\n```\nSetting overlay database mode to overlay-base with caching because we are analyzing the default branch.\nUploading overlay-base database to Actions cache with key codeql-overlay-base-database-1-\u003chash>-python-2.27.1-\u003csha>-\u003crun_id>-1\nSuccessfully uploaded overlay-base database from /home/runner/_work/_temp/codeql_databases\n```\n\nPull request run:\n\n```\nSetting overlay database mode to overlay with caching because we are analyzing a pull request.\nLooking in Actions cache for overlay-base database with restore key codeql-overlay-base-database-1-\u003chash>-python-2.27.1-\nNo overlay-base database found in Actions cache\nNo overlay-base database found in cache, reverting overlay database mode to none.\n```\n\n### Minimal reproduction of the backend difference\n\nA cache entry saved as `probe-\u003crun_id>-full`, then restored in a later job on the same branch with `actions/cache/restore@v4`:\n\n| Restore | GitHub-hosted runner | Blacksmith runner |\n|---|---|---|\n| `key: probe-\u003crun_id>-` (prefix as the primary key, like this action) | hit | **miss** |\n| `key: probe-\u003crun_id>-nomatch`, `restore-keys: probe-\u003crun_id>-` | hit | hit |\n\n### Suggested fix\n\nAlso pass the prefix as a restore key:\n\n```ts\nactionsCache.restoreCache(\n  [dbLocation],\n  cacheRestoreKeyPrefix,\n  [cacheRestoreKeyPrefix],\n  { segmentTimeoutInMs: 3000 },\n)\n```\n\nOn GitHub's cache this matches exactly the same entries as today, since the prefix is tried first either way. On backends that only prefix-match restore keys, overlay analysis starts working.\n\n### Environment\n\n- `github/codeql-action/init` and `analyze` `@v4`, CodeQL bundle 2.27.1\n- `build-mode: none`, languages `python` and `javascript-typescript`\n- Third-party hosted runners (Blacksmith) whose `actions/cache` calls go to their own cache backend\n","author":{"url":"https://github.com/Yuppyboy","@type":"Person","name":"Yuppyboy"},"datePublished":"2026-10-07T05:43:31Z","url":"https://github.com/github/codeql-action/issues/4202"}