If you're running the SeasonsEcowitt skin (or any skin derived from it) and
noticing high CPU usage and slow report generation, check line 233 of
hilo.inc:
#elif $x[3] == '1' and (($getattr($year, $x[0]).has_data) or ('wx_binding'
in $txt and $getattr($year(data_binding=$txt), $x[0]).has_data) or
($year.wind.has_data and ($x[0] == 'windavg' or $x[0] == 'windrms' or $x[0]
== 'windrms' or $x[0] == 'windvecavg')))
This check uses $year to determine whether to display observations in the
statistics table. On a database with 6+ years of data (~640,000 archive
records), this triggers MySQL to scan the entire year's worth of raw
archive records for *every observation* in the hilo table — even though the
year timespan is not displayed in the statistics table at all.
*Impact on my 6-year MySQL database:*
- Report generation time: *4 minutes 10 seconds*
- MySQL archive table scans per report: *39 full-year scans* (Jan 1 →
present, ~640k records each)
*Fix:* Replace $year with $week on that line:
#elif $x[3] == '1' and (($getattr($week, $x[0]).has_data) or ('wx_binding'
in $txt and $getattr($week(data_binding=$txt), $x[0]).has_data) or
($week.wind.has_data and ($x[0] == 'windavg' or $x[0] == 'windrms' or $x[0]
== 'windrms' or $x[0] == 'windvecavg')))
Since $week data is already loaded for the statistics table, this change
adds zero additional queries.
*Result:*
- Report generation time: *31 seconds* (from 4:10 → 0:31, *87% reduction*
)
- No functional change — the statistics table displays identically
*Environment:* weewx 5.4.0, Python 3.14.4, Fedora 43, MySQL backend,
~640,000 archive records (6 years), SeasonsEcowitt skin 5.1.0
--
You received this message because you are subscribed to the Google Groups
"weewx-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion visit
https://groups.google.com/d/msgid/weewx-user/94f6fc97-2e30-4ebc-bf57-264909c7ab67n%40googlegroups.com.
