feat(cardinality): Add clickhouse host and port parameters (#5237)

In our prod systems, there seems to be an issue when cron jobs try to
access other systems using envoy. In order to bypass that, we can use
the dns name of the cluster to reach the clickhouse host. That is
what we do today in the optimize and cleanup jobs as well. So added the
requirement to pass in clickhouse host information as a CLI argument.

snuba/cli/spans_cardinality_analyzer.py

@@ -6,6 +6,7 @@
 
 from snuba import settings
 from snuba.admin.notifications.slack.client import SlackClient
+from snuba.clickhouse.native import ClickhousePool
 from snuba.clickhouse.span_cardinality_analyzer import (
     SpanGroupingCardinalityResult,
     span_grouping_cardinality_query,
@@ -30,9 +31,22 @@ def write_cardnaltiy_to_csv(
 
 
 @click.command()
+@click.option(
+    "--clickhouse-host",
+    help="Clickhouse server to write to.",
+    required=True,
+)
+@click.option(
+    "--clickhouse-port",
+    type=int,
+    help="Clickhouse native port to write to.",
+    required=True,
+)
 @click.option("--log-level", help="Logging level to use.")
 def spans_cardinality_analyzer(
     *,
+    clickhouse_host: str,
+    clickhouse_port: int,
     log_level: Optional[str] = None,
 ) -> None:
     """
@@ -47,9 +61,15 @@ def spans_cardinality_analyzer(
 
     storage_key = StorageKey("generic_metrics_distributions")
     storage = get_storage(storage_key)
+    (clickhouse_user, clickhouse_password) = storage.get_cluster().get_credentials()
 
-    connection = storage.get_cluster().get_query_connection(
-        ClickhouseClientSettings.CARDINALITY_ANALYZER
+    connection = ClickhousePool(
+        host=clickhouse_host,
+        port=clickhouse_port,
+        user=clickhouse_user,
+        password=clickhouse_password,
+        database=storage.get_cluster().get_database(),
+        client_settings=ClickhouseClientSettings.CARDINALITY_ANALYZER.value.settings,
     )
 
     # Get the distinct span modules we are ingesting.