From 23765ae0e3b5a1ce10b38ea04bdba8639213d418 Mon Sep 17 00:00:00 2001
From: Jonathan Davies <jpds@protonmail.com>
Date: Fri, 13 Jun 2025 12:05:23 +0000
Subject: [PATCH] nixos/vector: Add Caddy+ClickHouse test case

---
 nixos/tests/vector/caddy-clickhouse.nix | 206 ++++++++++++++++++++++++
 nixos/tests/vector/default.nix          |   1 +
 2 files changed, 207 insertions(+)
 create mode 100644 nixos/tests/vector/caddy-clickhouse.nix

diff --git a/nixos/tests/vector/caddy-clickhouse.nix b/nixos/tests/vector/caddy-clickhouse.nix
new file mode 100644
index 000000000000..c1b478f3648c
--- /dev/null
+++ b/nixos/tests/vector/caddy-clickhouse.nix
@@ -0,0 +1,206 @@
+{ lib, pkgs, ... }:
+
+{
+  name = "vector-caddy-clickhouse";
+  meta.maintainers = [ pkgs.lib.maintainers.happysalada ];
+
+  nodes = {
+    caddy =
+      { config, pkgs, ... }:
+      {
+        networking.firewall.allowedTCPPorts = [ 80 ];
+
+        services.caddy = {
+          enable = true;
+          virtualHosts = {
+            "http://caddy" = {
+              extraConfig = ''
+                encode gzip
+
+                file_server
+                root /srv
+              '';
+              logFormat = "
+                  output file ${config.services.caddy.logDir}/access-caddy.log {
+                    mode 0640
+                  }
+                ";
+            };
+          };
+        };
+
+        systemd.services.vector.serviceConfig = {
+          SupplementaryGroups = [ "caddy" ];
+        };
+
+        services.vector = {
+          enable = true;
+
+          settings = {
+            sources = {
+              caddy-log = {
+                type = "file";
+                include = [ "/var/log/caddy/*.log" ];
+              };
+            };
+
+            transforms = {
+              caddy_logs_timestamp = {
+                type = "remap";
+                inputs = [ "caddy-log" ];
+                source = ''
+                  .tmp_timestamp, err = parse_json!(.message).ts * 1000000
+
+                  if err != null {
+                    log("Unable to parse ts value: " + err, level: "error")
+                  } else {
+                    .timestamp = from_unix_timestamp!(to_int!(.tmp_timestamp), unit: "microseconds")
+                  }
+
+                  del(.tmp_timestamp)
+                '';
+              };
+            };
+
+            sinks = {
+              vector_sink = {
+                type = "vector";
+                inputs = [ "caddy_logs_timestamp" ];
+                address = "clickhouse:6000";
+              };
+            };
+          };
+        };
+      };
+
+    client =
+      { config, pkgs, ... }:
+      {
+        environment.systemPackages = [ pkgs.curl ];
+      };
+
+    clickhouse =
+      { config, pkgs, ... }:
+      {
+        virtualisation.memorySize = 4096;
+
+        networking.firewall.allowedTCPPorts = [ 6000 ];
+
+        services.vector = {
+          enable = true;
+
+          settings = {
+            sources = {
+              vector_source = {
+                type = "vector";
+                address = "[::]:6000";
+              };
+            };
+
+            sinks = {
+              clickhouse = {
+                type = "clickhouse";
+                inputs = [
+                  "vector_source"
+                ];
+                endpoint = "http://localhost:8123";
+                database = "caddy";
+                table = "access_logs";
+                date_time_best_effort = true;
+                skip_unknown_fields = true;
+              };
+            };
+          };
+
+        };
+
+        services.clickhouse = {
+          enable = true;
+        };
+      };
+  };
+
+  testScript =
+    let
+      # work around quote/substitution complexity by Nix, Perl, bash and SQL.
+      databaseDDL = pkgs.writeText "database.sql" "CREATE DATABASE IF NOT EXISTS caddy";
+
+      tableDDL = pkgs.writeText "table.sql" ''
+        CREATE TABLE IF NOT EXISTS caddy.access_logs (
+          timestamp DateTime64(6),
+          host LowCardinality(String),
+          message String,
+        )
+        ENGINE = MergeTree()
+        ORDER BY timestamp
+        PARTITION BY toYYYYMM(timestamp)
+      '';
+
+      tableViewBase = pkgs.writeText "table-view-base.sql" ''
+        CREATE TABLE IF NOT EXISTS caddy.access_logs_view_base (
+          timestamp DateTime64(6),
+          host LowCardinality(String),
+          request JSON,
+          status UInt16,
+        )
+        ENGINE = MergeTree()
+        ORDER BY timestamp
+        PARTITION BY toYYYYMM(timestamp)
+      '';
+
+      tableView = pkgs.writeText "table-view.sql" ''
+        CREATE MATERIALIZED VIEW IF NOT EXISTS caddy.access_logs_view TO caddy.access_logs_view_base
+        AS SELECT
+          timestamp,
+          host,
+          simpleJSONExtractRaw(message, 'request') AS request,
+          simpleJSONExtractRaw(message, 'status') AS status
+        FROM caddy.access_logs;
+      '';
+
+      selectQuery = pkgs.writeText "select.sql" ''
+        SELECT
+          timestamp,
+          request.host,
+          request.remote_ip,
+          request.proto,
+          request.method,
+          request.uri,
+          status
+        FROM caddy.access_logs_view_base
+        WHERE request.uri LIKE '%test-uri%'
+        FORMAT Pretty
+      '';
+    in
+    ''
+      clickhouse.wait_for_unit("clickhouse")
+      clickhouse.wait_for_unit("vector")
+      clickhouse.wait_for_open_port(6000)
+      clickhouse.wait_for_open_port(8123)
+
+      clickhouse.succeed(
+        "cat ${databaseDDL} | clickhouse-client",
+        "cat ${tableDDL} | clickhouse-client",
+        "cat ${tableViewBase} | clickhouse-client",
+        "cat ${tableView} | clickhouse-client",
+      )
+
+      caddy.wait_for_unit("caddy")
+      caddy.wait_for_open_port(80)
+      caddy.wait_for_unit("vector")
+      caddy.wait_until_succeeds(
+        "journalctl -o cat -u vector.service | grep 'Vector has started'"
+      )
+
+      client.systemctl("start network-online.target")
+      client.wait_until_succeeds("curl http://caddy/test-uri")
+
+      caddy.wait_until_succeeds(
+        "journalctl -o cat -u vector.service | grep 'Found new file to watch. file=/var/log/caddy/access-caddy.log'"
+      )
+
+      clickhouse.wait_until_succeeds(
+        "cat ${selectQuery} | clickhouse-client | grep test-uri"
+      )
+    '';
+}
diff --git a/nixos/tests/vector/default.nix b/nixos/tests/vector/default.nix
index 79af701903a0..360fbd9a716f 100644
--- a/nixos/tests/vector/default.nix
+++ b/nixos/tests/vector/default.nix
@@ -3,6 +3,7 @@
 {
   file-sink = runTest ./file-sink.nix;
   api = runTest ./api.nix;
+  caddy-clickhouse = runTest ./caddy-clickhouse.nix;
   dnstap = runTest ./dnstap.nix;
   journald-clickhouse = runTest ./journald-clickhouse.nix;
   nginx-clickhouse = runTest ./nginx-clickhouse.nix;