We've recently seen at least two incidents where single-core VMs have shown periodical patterns of saturation roughly matching chef-repo runs cadence (https://gitlab.com/gitlab-com/gl-infra/production/-/issues/6238 and https://gitlab.com/gitlab-com/gl-infra/production/-/issues/6230). @jacobvosmaer-gitlab has pointed out that it seems that every run performs unnecessary CPU intensive actions, like compiling and installing packages (e.g. `osqueryd`). We should try to identify such tasks executed by our chef roles, and where possible optimize them.