diff --git a/sql/histograms/evalJs.sql b/sql/histograms/evalJs.sql index 00ac184..4d2b200 100644 --- a/sql/histograms/evalJs.sql +++ b/sql/histograms/evalJs.sql @@ -13,6 +13,9 @@ FROM ( CAST(FLOAT64(payload['_cpu.EvaluateScript']) / 20 AS INT64) * 20 AS bin FROM `httparchive.crawl.requests` + INNER JOIN + `httparchive.crawl.pages` + USING (date, client, is_root_page, rank, page) WHERE date = '${YYYY-MM-DD}' AND is_root_page diff --git a/sql/timeseries/h2.sql b/sql/timeseries/h2.sql index 3d0eed2..2ce3681 100644 --- a/sql/timeseries/h2.sql +++ b/sql/timeseries/h2.sql @@ -8,7 +8,7 @@ FROM `httparchive.crawl.requests` r INNER JOIN `httparchive.crawl.pages` -USING (date, client, is_root_page, rank) +USING (date, client, is_root_page, rank, page) WHERE is_root_page AND date >= '2016-07-15' diff --git a/sql/timeseries/h3.sql b/sql/timeseries/h3.sql index 4e50661..dd15274 100644 --- a/sql/timeseries/h3.sql +++ b/sql/timeseries/h3.sql @@ -31,7 +31,7 @@ LEFT OUTER JOIN ON (resp.name = 'alt-svc') INNER JOIN `httparchive.crawl.pages` -USING (date, client, is_root_page, rank) +USING (date, client, is_root_page, rank, page) WHERE date >= '2020-01-01' AND is_root_page diff --git a/sql/timeseries/pctHttps.sql b/sql/timeseries/pctHttps.sql index 623f180..5ac3b7b 100644 --- a/sql/timeseries/pctHttps.sql +++ b/sql/timeseries/pctHttps.sql @@ -7,8 +7,8 @@ SELECT FROM `httparchive.crawl.requests` INNER JOIN - `httparchive.crawl.pages` p -USING (date, client, is_root_page, rank) + `httparchive.crawl.pages` +USING (date, client, is_root_page, rank, page) WHERE is_root_page AND date >= '2016-01-01'