Werk #19832: Systemd timesyncd: jitter and stratum levels reworked
| Component | Checks & agents | ||
| Title | Systemd timesyncd: jitter and stratum levels reworked | ||
| Date | Sep 7, 2026 | ||
| Level | Trivial Change | ||
| Class | New Feature | ||
| Compatibility | Incompatible - Manual interaction might be required | ||
| Checkmk versions & editions |
|
You are affected if you monitor "Systemd Timesyncd Time" services: those that were WARN or CRIT because of the jitter or the stratum may change their state after the update.
Previous situation
The jitter shared its levels with the offset: the rule field "Thresholds for quality of time" (default 200 ms / 500 ms) applied to jitter and offset.
The stratum was configured by "Critical at stratum" (default 10). Because of a bug the check turned CRIT one stratum below that value and never went WARN, so with the default a host reported CRIT from stratum 9 on.
New situation
Jitter and offset now have separate level fields in the rule. The jitter levels are not set by default, so the jitter is informational only.
The stratum has levels of its own, defaulting to WARN at 10 and CRIT at 16. Stratum 16 is the NTP value for an unsynchronised clock; 10 flags an unusually deep chain to the reference clock. Existing rules are migrated.
The time since the last NTP message and the time server moved from the service summary to the details; the time server now also shows its host name, if it resolves.