It'd be nice to have more easily readable (and plottable, and maybe alertable?) stats about hourly webrequest sequence stats. We could augment christian's report, but I think an easy intermediate step would be to also create a new aggregate table that sums stats and percent_different from the webrequest_sequence_stats table. Something like:
create table otto.webrequest_sequence_stats_hourly as select webrequest_source, year, month, day, hour, sum(count_actual) as count_actual, sum(count_expected) as count_expected, sum(count_different) as count_different, ((sum(count_actual) / sum(count_expected)) - 1) * 100.0 as percent_different from wmf_raw.webrequest_sequence_stats where webrequest_source='text' and year=2015 and month=8 and day = 11 and percent_different <= 0.0 and sequence_min <> 0 group by webrequest_source, year, month, day, hour ;
Not sure about the percent_different <= 0.0 and sequence_min <> 0 bit, we should discuss that.