This issue should be fixed very soon, but I prefer signaling it here: on Centreon platforms having tens of thousands services and using the BAM module, the central broker daemon needs time to completely load its cache due to SQL queries that take too long to return the cache entries.
The symptoms are that after a restart of the `cbd` service, the “last update” time will remain frozen for each poller for a while, and the “top counter” icon in the black banner will warn you that some pollers are not up-to-date.
Once the cache is loaded, the broker daemon will catch-up handling the monitoring data and will work properly.
The team is actively working on this issue and a fix will soon be released (before the end of January, hopefully).
This issue is also referenced here (French version here).
Fix release status: