The easy-to-use open source Business Intelligence and Embedded Analytics tool that lets everyone work with data :bar_chart:
TypeScript analyticsbibusiness-intelligencebusinessintelligenceclojuredashboarddatadata-analysisdata-visualizationdatabasemetabasemysqlpostgrespostgresqlreportingslacksql-editorvisualization
[o11y] Add a new metric to count unique cards with blank queries (#66903)
Braden Shepherdsonbd40119508b5
diff --git a/src/metabase/analytics/prometheus.clj b/src/metabase/analytics/prometheus.cljindex c1b7306fa6a..f33be5de57b 100644--- a/src/metabase/analytics/prometheus.clj+++ b/src/metabase/analytics/prometheus.clj@@ -251,6 +251,8 @@{:description "Number of errors when processing metrics in the metrics adjust middleware."})(prometheus/gauge :metabase-query-processor/computed-weak-map-queries{:description "Number of queries cached in lib.computed/weak-map."})+ (prometheus/gauge :metabase-card/unique-cards-failed-conversion+ {:description "Number of distinct cards which have :dataset_query {}, meaning MBQL 4 to 5 conversion failed."})(prometheus/gauge :metabase-database/status{:description "Does a given database using driver pass a health check.":labels [:driver :healthy :reason]})diff --git a/src/metabase/queries/models/card.clj b/src/metabase/queries/models/card.cljindex dd83693e71b..24060643e83 100644--- a/src/metabase/queries/models/card.clj+++ b/src/metabase/queries/models/card.clj@@ -7,6 +7,7 @@[clojure.walk :as walk][honey.sql.helpers :as sql.helpers][medley.core :as m]+ [metabase.analytics.core :as analytics][metabase.api.common :as api][metabase.app-db.core :as app-db][metabase.audit-app.core :as audit]@@ -705,6 +706,26 @@card)queries.schema/normalize-card))+(defonce ^:private unique-cards-with-blank-dataset-query+ (atom #{}))++(defn- monitor-blank-dataset-query+ "Captures the IDs of all cards which get read and have `:dataset_query {}`. That happens when the MBQL 4->5+ conversion fails, and usually indicates a problem with the MBQL 4 normalization.++ This publishes a Prometheus metric with a count of the unique cards on this instance which have been read since+ startup. If this is nonzero then there are known to be bad cards present. This function logs a message in that+ case which will point out the bad cards and enable us to localize the problem.++ Always returns `card`."+ [card]+ (when (= (:dataset_query card) {})+ (log/infof "Card %d has a blank :dataset_query - this indicates a Metabase issue" (:id card))+ (let [uniques (swap! unique-cards-with-blank-dataset-query conj (:id card))]+ (analytics/set! :metabase-card/unique-cards-failed-conversion (count uniques))))+ ;; Always returns the original card.+ card)+(t2/define-after-select :model/Card[card];; +===============================================================================================+@@ -718,7 +739,8 @@public-sharing/remove-public-uuid-if-public-sharing-is-disabledadd-query-description-to-metric-card;; At this point, the card should be at schema version 20 or higher.- upgrade-card-schema-to-latest))+ upgrade-card-schema-to-latest+ monitor-blank-dataset-query))(t2/define-before-insert :model/Card[card]