This is a different problem from a host being Down — the connection and authentication are working, but the data itself is missing, sparse, or doesn't cover the period you expect.
Common causes
- Too soon — a newly added check simply hasn't run enough times yet to populate a meaningful graph. Give it a few check intervals before assuming something is wrong.
- Wrong graph timespan — a graph set to a very long timespan aggregates data (see Performance Metrics), and a recently added check will show as a flat or empty line for the portion of that timespan before it existed, which can look like missing data at a glance.
- Data retention — historical data beyond your environment's retention window has been rolled up or aged out, which is expected behavior rather than a fault — see Performance Metrics for how retention works.
- Service on the target stopped responding — for checks that rely on an application's own status endpoint (Apache's mod_status, PHP-FPM's status page, Tomcat's JMX, RabbitMQ's management API — see Application Monitoring), if that specific service or plugin stops responding, the underlying host/SSH/SNMP connection can still be fine while this one check goes dark.
- Permissions changed — an account that lost read access to a status endpoint or management API partway through its life will show the same symptom as it never having worked, which can be confusing if it was working previously.
How to isolate it
- Check whether the gap is on every metric for that host, or just one specific check — a single affected check points at that check's specific data source (an application status endpoint, for example), not the host connection as a whole.
- Confirm the check's own status (not just the host's) — a host can show Up while an individual service check on it shows Unknown or has simply stopped updating.
- Try a shorter graph timespan to rule out an aggregation/display issue before assuming data collection itself has actually failed.
If the underlying connection method itself seems to be the problem rather than one specific check, see SNMP Problems or SSH Problems depending on how that host is monitored.