Werk #19044: Fix inaccurate alert counts and infinite loading spinners in ntop dashlets
| Component | Ntopng integration | ||||
| Title | Fix inaccurate alert counts and infinite loading spinners in ntop dashlets | ||||
| Date | Mar 19, 2026 | ||||
| Level | Trivial Change | ||||
| Class | Bug Fix | ||||
| Compatibility | Compatible - no manual interaction needed | ||||
| Checkmk versions & editions |
|
This fix addresses three related issues in the ntop integration dashlets:
Quickstats alert badge showed wrong count: The "X Alerts" badge in the ntop interface quickstats widget was reading a pre-aggregated count from ntop's interface data endpoint, which includes all alert categories (host, interface, and system). The badge now queries the host alert endpoint directly, matching the count shown in the Engaged Alerts dashlet.
Alert table showed inflated total count: The total alert count in the Engaged Alerts dashlet was computed by summing the 31-day alert timeseries, which included historical resolved alerts. It now uses the recordsFiltered value from the alert list API, reflecting only the currently matching alerts.
Infinite loading spinners: Ntop dashlets could get stuck showing a loading spinner forever when ntop returned no data. Timeseries figures now show "No data available" instead, and traffic table cells correctly dismiss their loading image after processing.