You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
cgroups: fix io.stat parsing so IO stats are recorded
cgroups: emit throttled/burst/force_idle-time, percpu, thp_swpout, and zswpwb stats
cgroups: fix container IP mapping key and cache-save error message
cgroups: populate path with container rootfs, falling back to the roots
seen by the procs in the cgroup, matching jail path semantics
cgroups: treat nr_periods, nr_throttled, and nr_bursts as counters so
they are rates like throttled-time and burst-time, matching what the
LibreNMS graphs expect both: when a counter is seen for the first
time, return 0 instead of the accumulated total over 300 if what it is for
is older than the polling interval, preventing spikes when something old
is first seen... procs younger than the polling interval are still processed
as fresh counters so short lived procs don't go uncounted
cgroups: don't zero the cached raw counter value when discarding a
garbage value, preventing a inflated delta on the following run
cgroups: remove unused raw usec names from the counters list
FreeBSD: fix MM:SS CPU time parsing
FreeBSD: guard against counter rollback
FreeBSD: populate jail path
both: self-heal a corrupt/truncated cache instead of erroring every run
both: write the cache atomically to avoid truncated cache files
fix leaked $@ that turned a handled cache error into "Failed to run backend"