Werk #20009: Availability: fix false "row limit reached" warnings and detect unreachable sites
| Component | User interface | ||
| Title | Availability: fix false "row limit reached" warnings and detect unreachable sites | ||
| Date | Jun 1, 2026 | ||
| Level | Trivial Change | ||
| Class | Bug Fix | ||
| Compatibility | Compatible - no manual interaction needed | ||
| Checkmk versions & editions |
|
This werk fixes two long-standing problems in the availability computation, affecting both BI availability and host/service availability.
False "row limit reached" warnings in distributed setups
The computation option Process at most ... status entries limits the number of state history rows fetched per query. This limit is applied to each site individually, but the check whether the limit was reached compared the summed number of rows from all sites against the per-site limit.
In a distributed (multisite) setup this raised the warning
Your query matched more than N log entries. [...] the result might be
incomplete.
even when every site returned a complete dataset below its own limit - for example three sites returning 200000 rows each would trip a 400000 limit although none of them was actually truncated. For host/service availability the opposite could also happen: with a single site the check could never trigger at all.
The limit is now evaluated per site, so the warning is only shown when a single site genuinely had to truncate its data.
No warning when a site is unreachable
If one of several sites was down, the availability was silently computed on the reduced data of the remaining sites without any indication that data was missing.
The availability views and the availability report element now show a warning listing the sites that could not be reached, so an incomplete and potentially incorrect availability is no longer presented as complete.