There are various queries in that log, but only one showed stale data. The others showed that only tag, “sites/pramorel/niveau/hvalue”, should have not shown, because its quality really was bad.
The stale quality mostly comes from the scan class execution validation. There is a table maintained of scan class execution windows, so that we can detect when the system was not actually running, and mark data as “stale”. But perhaps gaps in this table are causing your values to show as stale, as in the one query, each block was covering a lot of time (1 day range, but only 23 points? Is that right?)
The scan class records are contained in the “sqlth_sce” table. If you want, you can back that table up, then delete the contents, wait for rows to be inserted for each scan class, and then run a query to update the “start_time” to 0, so that each scan class has a record that covers from 0 to now. You then should not get “stale” reported.
Let me know if this helps.