Skip to content

The easy-to-use open source Business Intelligence and Embedded Analytics tool that lets everyone work with data :bar_chart:

[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.clj
index 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.clj
index 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-disabled
add-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]