Open-source Linux developer Chris Mason who is known for being the original lead developer of the Btrfs file-system recently began hacking on a new tool that he announced today, rsched.
Rsched is a new tool for dealing with Linux kernel scheduler metrics. Like much software written today, it’s written in the Rust programming language. Rsched makes use of the Linux kernel BPF integration for watching scheduler tracepoints and select performance counters.
Chris Mason elaborated in his announcement today introducing rsched:
“I’ve been wanting to write up a tool to give a broad overview of process scheduler metrics, and I used our recent regression hunt as an excuse to work something up.
rsched is meant to sit next to the running workload, kind of like vmstat or top. The goal is just being able to quickly decide if I should blame the scheduler, and maybe give some ideas about which part to blame. It doesn’t provide the full details like systing or wprof, both of which are much better for in-depth analysis.
rsched is written in rust and uses BPF to watch a few scheduler tracepoints, as well as some performance counters. I’m obviously still learning rust, so don’t be shy with the pull requests if you see chances to improve the code.”
Rsched can be downloaded and explored via GitHub. More details on rsched in general via the mailing list announcement.
