Reproduced with official Zabbix 7.4.15 Ubuntu images and bundled fping 5.1, using a proxy with default pinger settings and resolver timeouts.
When DNS lookups take too long, a shared fping execution continues beyond 600 seconds. Healthy targets in the same batch stop receiving new values and become unsupported.

Steps to reproduce:

1. Create 124 proxy monitored hosts with DNS interfaces and icmpping items. Confirm successful checks and that targets are passed to one fping process.
2. Delay DNS replies for 121 names by 5.4 seconds, while resolving the remaining names immediately.
3. Test delayed successful replies and delayed NXDOMAIN separately, restoring normal DNS between runs.

Result:
Proxy logs:

fping execution time limit (600s) exceeded, stopping output processing

All 124 items become unsupported, including healthy targets. They receive the same error, which can refer to another host:

fping failed: slow-112.test: Name or service not known

or even contain a successful ping reply:

fping failed: healthy-a.test : [0], 64 bytes, 0.145 ms (0.145 avg, 0% loss)

Healthy items had 684 s gaps between history values. Collection recovered after restoring DNS, without restarting Zabbix or changing host/item configuration. Quick NXDOMAIN and ICMP loss alone did not interrupt healthy targets.