Linux Cgroup from First Principles

a year ago

Linux cgroups (control groups) are used to limit resource allocations such as memory, CPU, and network bandwidth for processes.
Cgroup v2, introduced in Linux kernel 4.5, offers a simpler design compared to v1.
Linux follows the Unix philosophy 'everything is a file,' allowing interaction with kernel subsystems via file operations.
Cgroups are organized under `/sys/fs/cgroup`, with each cgroup containing virtual files for setting limits and monitoring usage.
Controllers like `memory.max` and `cpu.max` can be enabled in cgroups to enforce resource constraints.
Processes can be moved to a cgroup by writing their PID to `cgroup.procs` or using tools like `cgexec`.
Memory limits can be set (e.g., 5MiB) to prevent processes from exceeding allocated resources, triggering OOM (Out of Memory) kills.
CPU throttling can be applied (e.g., 1% CPU allocation) to simulate over-subscribed machines.
Cgroups provide isolation and are useful for performance benchmarking under constrained conditions.
Error messages for invalid cgroup operations (e.g., 'no internal process' constraint) could be more descriptive.

Hasty Briefsbeta