Werk #19789: Fix response time levels losing millisecond values in certain rules
| Component | Setup | ||
| Title | Fix response time levels losing millisecond values in certain rules | ||
| Date | May 18, 2026 | ||
| Level | Trivial Change | ||
| Class | Bug Fix | ||
| Compatibility | Compatible - no manual interaction needed | ||
| Checkmk versions & editions |
|
When editing rules whose response time thresholds are displayed in milliseconds only (without a separate seconds field), warning and critical thresholds were silently reset to 1/1000th of their intended value after saving.
For example, a warning threshold of 1000 ms would be stored as 1.0 s on first save, but displayed as 1 ms on the next edit. Saving again would then permanently store 1 ms.
The root cause was a wrong cascade factor in the TimeSpan valuespec: when
the seconds unit is not shown but the milliseconds field is, the whole-seconds
value was passed to the milliseconds input with a factor of 1 instead of 1000.
So 1.0 s was rendered as 1 ms. A second bug prevented rules originally
created in Checkmk 2.3 from loading at all in 2.4 when a migration function
was configured on the levels form spec.
The following rules are affected because they display response time thresholds in milliseconds only (i.e. without a seconds field alongside):
- Check TCP port connection (expected response time)
- Disk IO (read/write service time)
- Diskstat (read/write service time, read/write wait time, average wait time)
- MS Exchange: Database (transaction log write time), Info Store (read/write latency), RPC Client Access (RPC latency)
This has been fixed.