everything works

getsentry · Dec 16, 2024 · 0a9d96b · 0a9d96b
1 parent ebab5ff
commit 0a9d96b
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 25 deletions.
diff --git a/snuba/manual_jobs/scrub_ips_from_eap_spans.py b/snuba/manual_jobs/scrub_ips_from_eap_spans.py
@@ -42,27 +42,8 @@ def execute(self, logger: JobLogger) -> None:
         else:
             cluster_name = None
         query = self._get_query(cluster_name)
-        print("queryyyy", query)
         logger.info("Executing query: {query}")
         result = connection.execute(query=query, settings={"mutations_sync": 0})
 
-        # this shows all the user.ip are scrubbed, so the mapping works
-        print(
-            "mapApplyresult",
-            connection.execute(
-                query="SELECT mapApply((k, v) -> (k, if(k = 'user.ip', 'scrubbed', v)), `attr_str_1`) from eap_spans_2_local",
-                settings={"mutations_sync": 0},
-            ),
-        )
-
-        # this shows the table is not updated even though the mapping is correct
-        print(
-            "selectresult",
-            connection.execute(
-                query="SELECT attr_str_1 from eap_spans_2_local LIMIT 1",
-                settings={"mutations_sync": 0},
-            ),
-        )
-
         logger.info("complete")
         logger.info(repr(result))
diff --git a/snuba/web/db_query.py b/snuba/web/db_query.py
@@ -302,7 +302,6 @@ def execute_query_with_readthrough_caching(
         clickhouse_query_settings["query_id"] = query_id
         if span:
             span.set_data("query_id", query_id)
-
         return execute_query(
             clickhouse_query,
             query_settings,

diff --git a/tests/manual_jobs/test_scrub_ips_from_eap_spans.py b/tests/manual_jobs/test_scrub_ips_from_eap_spans.py
@@ -198,10 +198,13 @@ def _gen_message(
     }
 
 
-def _generate_request(ts: Any, hour_ago: int) -> TraceItemTableRequest:
+def _generate_request(
+    ts: Any, hour_ago: int, project_ids: list[int]
+) -> TraceItemTableRequest:
+    # project_ids is added as an argument to avoid this query getting cached
     return TraceItemTableRequest(
         meta=RequestMeta(
-            project_ids=[1, 2, 3],
+            project_ids=project_ids,
             organization_id=1,
             cogs_category="something",
             referrer="something",
@@ -239,8 +242,9 @@ def test_span_is_scrubbed() -> None:
 
     ts = Timestamp(seconds=int(BASE_TIME.timestamp()))
     hour_ago = int((BASE_TIME - timedelta(hours=1)).timestamp())
-    message = _generate_request(ts, hour_ago)
-    response = EndpointTraceItemTable().execute(message)
+    response = EndpointTraceItemTable().execute(
+        _generate_request(ts, hour_ago, [1, 2, 3])
+    )
     expected_response = TraceItemTableResponse(
         column_values=[
             TraceItemColumnValues(
@@ -271,7 +275,9 @@ def test_span_is_scrubbed() -> None:
         )
     )
 
-    print(response)
+    response = EndpointTraceItemTable().execute(
+        _generate_request(ts, hour_ago, [3, 2, 1])
+    )
     assert response == TraceItemTableResponse(
         column_values=[
             TraceItemColumnValues(