Title: node-cache-pruner reads every cache's size even when a running pod means it cannot prune · Issue #146 · ExaDev/github-runner · GitHub
Open Graph Title: node-cache-pruner reads every cache's size even when a running pod means it cannot prune · Issue #146 · ExaDev/github-runner
X Title: node-cache-pruner reads every cache's size even when a running pod means it cannot prune · Issue #146 · ExaDev/github-runner
Description: node-cache-pruner.sh runs du -sk over every profile's cache on every round, before it checks whether a runner pod mounts the cache. It can only empty a cache when no pod mounts it, so on a node that is always running jobs nearly every ro...
Open Graph Description: node-cache-pruner.sh runs du -sk over every profile's cache on every round, before it checks whether a runner pod mounts the cache. It can only empty a cache when no pod mounts it, so on a node tha...
X Description: node-cache-pruner.sh runs du -sk over every profile's cache on every round, before it checks whether a runner pod mounts the cache. It can only empty a cache when no pod mounts it, so on a node...
Opengraph URL: https://github.com/ExaDev/github-runner/issues/146
X: @github
Domain: github.com
{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"node-cache-pruner reads every cache's size even when a running pod means it cannot prune","articleBody":"`node-cache-pruner.sh` runs `du -sk` over every profile's cache on every round, before it checks whether a runner pod mounts the cache. It can only empty a cache when no pod mounts it, so on a node that is always running jobs nearly every round reads the size and then does nothing with it.\n\nOn a node with a consumer SATA SSD and a shared pnpm store of about 300k files, that read is expensive: the jobs' page cache churn evicts the inode tables between rounds, so each `du` re-reads them, and with ext4's inode readahead a round read 5 to 6 GiB at about 90 MiB/s for about a minute. While it ran, `/proc/pressure/io` `full` on the node was 14 to 32%, and over a six-minute sample the pruner's windows held 19% and 54% of the node's whole I/O stall on two such nodes. The pruner container's cgroup `io.stat` showed about 3.5 TiB read in a little over two days at the default `CHECK_SECONDS` of 300.\n\nListing the pods first and reading the size only when the cache could be emptied (no pod mounts it, or it is retired) would remove almost all of it on a busy node. The hard-limit report still needs a size while pods run; that could be read on a slower cadence of its own. A consumer can raise `github_runner_arc_node_cache_prune_seconds` meanwhile, which only spreads the same reads out.\n","author":{"url":"https://github.com/Mearman","@type":"Person","name":"Mearman"},"datePublished":"2026-10-27T15:49:18Z","url":"https://github.com/ExaDev/github-runner/issues/146"}
| react-profiling | 0 |
| react-import-map | react,react-dom,react-dom/client,react-dom/profiling,react-is,react/compiler-runtime,react/jsx-dev-runtime,react/jsx-runtime,scheduler@b8488ee2c322 |
| route-pattern | /_view_fragments/issues/show/:user_id/:repository/:id/issue_layout(.:format) |
| route-controller | voltron_issues_fragments |
| route-action | issue_layout |
| fetch-nonce | v2:cca04ab4-535a-d909-3e91-56e6b8be6491 |
| current-catalog-service-hash | 81bb79d38c15960b92d99bca9288a9108c7a47b18f2423d0f6438c5b7bcd2114 |
| request-id | A144:BB729:5058D9:6DEF6E:6B007312 |
| html-safe-nonce | 691cefa85493e578912ba3070d57d8eee10a464b4a8f336d870876dfa034a300 |
| visitor-payload | eyJyZWZlcnJlciI6IiIsInJlcXVlc3RfaWQiOiJBMTQ0OkJCNzI5OjUwNThEOTo2REVGNkU6NkIwMDczMTIiLCJ2aXNpdG9yX2lkIjoiODY5MTQxMzU2NDIxNDI0NDExNSIsInJlZ2lvbl9lZGdlIjoiaWFkIiwicmVnaW9uX3JlbmRlciI6ImlhZCJ9 |
| visitor-hmac | 35f89e7edbc61f9532a71b23715598f595d0b32d5233231acc0dda0e943930c2 |
| hovercard-subject-tag | issue:6062702193 |
| github-keyboard-shortcuts | repository,issues,copilot |
| google-site-verification | Apib7-x98H0j5cPqHWwSMm6dNU4GmODRoqxLiDzdx9I |
| octolytics-url | https://collector.github.com/github/collect |
| analytics-location | / |
| fb:app_id | 1401488693436528 |
| apple-itunes-app | app-id=1477376905, app-argument=https://github.com/_view_fragments/issues/show/ExaDev/github-runner/146/issue_layout |
| twitter:image | https://opengraph.githubassets.com/424b8e04f39519b87216fe3ddcae5908c1feb315aece9d0635657d6dc78b4b07/ExaDev/github-runner/issues/146 |
| twitter:card | summary_large_image |
| og:image | https://opengraph.githubassets.com/424b8e04f39519b87216fe3ddcae5908c1feb315aece9d0635657d6dc78b4b07/ExaDev/github-runner/issues/146 |
| og:image:alt | node-cache-pruner.sh runs du -sk over every profile's cache on every round, before it checks whether a runner pod mounts the cache. It can only empty a cache when no pod mounts it, so on a node tha... |
| og:image:width | 1200 |
| og:image:height | 600 |
| og:site_name | GitHub |
| og:type | object |
| og:author:username | Mearman |
| hostname | github.com |
| expected-hostname | github.com |
| None | 75c2b65dd205f26961d0c29afed1a8d0bb7cca9f05780571dfd1ef93eb539d16 |
| turbo-cache-control | no-preview |
| go-import | github.com/ExaDev/github-runner git https://github.com/ExaDev/github-runner.git |
| octolytics-dimension-user_id | 140118018 |
| octolytics-dimension-user_login | ExaDev |
| octolytics-dimension-repository_id | 1388126104 |
| octolytics-dimension-repository_nwo | ExaDev/github-runner |
| octolytics-dimension-repository_public | true |
| octolytics-dimension-repository_is_fork | false |
| octolytics-dimension-repository_network_root_id | 1388126104 |
| octolytics-dimension-repository_network_root_nwo | ExaDev/github-runner |
| turbo-body-classes | logged-out env-production page-responsive |
| disable-turbo | false |
| browser-stats-url | https://api.github.com/_private/browser/stats |
| browser-errors-url | https://api.github.com/_private/browser/errors |
| release | 84e563eb4b55f8fa102f325f117533cee9e3df4a |
| ui-target | full |
| theme-color | #1e2327 |
| color-scheme | light dark |
| react-app-name | issues |
Links:
Viewport: width=device-width