cgroups
cgroups copied to clipboard
cgroups package for Go
I noticed that creating a cgroup will silently ignore timeouts and continue on. Concretely, I've hit cases where a cgroup fails to get created, and the caller ends up looking...
The unified hierarchy provides the cpu.stat file for every cgroup, regardless if the CPU controller is enabled (in fact, setting the systemd property CPUAccounting=True does not enable this controller because...
We are missing hugetlb failcnt metric for cgroup2. According to [kernel doc][1], this can be sourced from `hugetlb..events`. > hugetlb..events > A read-only flat-keyed file which exists on non-root cgroups....
Addresses https://github.com/containerd/cgroups/issues/365 This change adds `CPUQuotaPeriodUSec` property for cgroupsv2. Heavily influenced by github.com/opencontainers/cgroups. Refactors a bit of the existing functions for testability. Added some unit tests for validating the property...
Bumps [golangci/golangci-lint-action](https://github.com/golangci/golangci-lint-action) from 6 to 8. Release notes Sourced from golangci/golangci-lint-action's releases. v8.0.0 Requires golangci-lint version >= v2.1.0 What's Changed Changes feat: use absolute paths by default when using working-directory...
Currently `cgroup2` systemd implementation does not set `CPUQuotaPeriodUSec` property. Similar support was added into runc's cgroup implementation: * https://github.com/opencontainers/runc/issues/2465 * https://github.com/opencontainers/runc/pull/2466 Shall we replicate the same logic in this package?
cgroup v2 use unix.InotifyInit() to monitor oom,wil cause too many open files, because every container will create an inotify instance see:https://github.com/containerd/containerd/issues/10468#issuecomment-2747079482
`Load` does not check for cgroup existence: this proposal fixes this
Check LinuxResources field, and skip Update() operation if not specified or invalid to avoid the potential panic issue of misconfigure