Add tnt_cartridge_config_checksum metric

Author: p0rtale

CHANGELOG.md

@@ -8,6 +8,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+- `tnt_cartridge_config_checksum` metric.
+
 ### Changed
 
 ### Fixed

doc/monitoring/api_reference.rst

@@ -570,6 +570,7 @@ Metrics functions
     *   ``luajit``
     *   ``cartridge_issues``
     *   ``cartridge_failover``
+    *   ``cartridge_config``
     *   ``clock``
     *   ``event_loop``
     *   ``config``

doc/monitoring/metrics_reference.rst

@@ -414,6 +414,10 @@ Cartridge
         *   -   ``tnt_cartridge_failover_trigger_total``
             -   Count of failover triggers in cluster.
 
+        *   -   ``tnt_cartridge_config_checksum``
+            -   Cartridge configuration checksum on the instance.
+                Can be used to detect configuration divergence across cluster nodes.
+
 ..  _metrics-reference-luajit:
 
 LuaJIT metrics

metrics/cartridge/config.lua

@@ -0,0 +1,28 @@
+local utils = require('metrics.utils')
+local collectors_list = {}
+
+local function update()
+    local is_cartridge = pcall(require, 'cartridge')
+    if not is_cartridge then
+        return
+    end
+
+    local confapplier = require('cartridge.confapplier')
+    local clusterwide_config = confapplier.get_active_config()
+    if clusterwide_config ~= nil then
+        collectors_list.config_checksum =
+            utils.set_gauge(
+                'cartridge_config_checksum',
+                'Cartridge configuration checksum on the instance',
+                clusterwide_config:get_checksum(),
+                nil,
+                nil,
+                {default = true}
+            )
+    end
+end
+
+return {
+    update = update,
+    list = collectors_list,
+}

metrics/tarantool.lua

@@ -22,6 +22,7 @@ local default_metrics = {
     luajit              = require('metrics.tarantool.luajit'),
     cartridge_issues    = require('metrics.cartridge.issues'),
     cartridge_failover  = require('metrics.cartridge.failover'),
+    cartridge_config    = require('metrics.cartridge.config'),
     clock               = require('metrics.tarantool.clock'),
     event_loop          = require('metrics.tarantool.event_loop'),
     config              = require('metrics.tarantool.config'),

test/integration/cartridge_metrics_test.lua

@@ -144,3 +144,62 @@ g.test_failover = function()
     g.cluster.main_server:start()
     g.cluster:wait_until_healthy()
 end
+
+local function get_config_checksum(server)
+    local resp = server:http_request('get', '/metrics')
+    local checksum_metric = utils.find_metric('tnt_cartridge_config_checksum', resp.json)
+
+    t.assert(checksum_metric, 'tnt_cartridge_config_checksum metric should be present')
+    t.assert_equals(#checksum_metric, 1)
+
+    local checksum = checksum_metric[1].value
+    t.assert_type(checksum, 'number', 'Checksum should be a number')
+
+    return checksum
+end
+
+g.test_config_checksum_match = function()
+    local main_server = g.cluster:server('main')
+    local replica_server = g.cluster:server('replica')
+
+    local initial_checksum = get_config_checksum(main_server)
+
+    main_server.net_box:eval([[
+        local patch = require('cartridge').config_patch_clusterwide
+        local ok, err = patch({
+            ['text'] = 'text',
+        })
+    ]])
+
+    local updated_checksum = get_config_checksum(main_server)
+    t.assert_not_equals(initial_checksum, updated_checksum, 'Config checksum should change after config update')
+
+    local replica_checksum = get_config_checksum(replica_server)
+    t.assert_equals(replica_checksum, updated_checksum, 'Nodes should have the same config checksum')
+end
+
+g.test_config_checksum_mismatch = function()
+    local main_server = g.cluster:server('main')
+    local replica_server = g.cluster:server('replica')
+
+    replica_server.net_box:eval([[
+        local vars = require('cartridge.vars').new('cartridge.confapplier')
+        __old_config = vars.clusterwide_config
+    ]])
+
+    main_server.net_box:eval([[
+        local patch = require('cartridge').config_patch_clusterwide
+        local ok, err = patch({
+            ['text'] = 'text',
+        })
+    ]])
+
+    replica_server.net_box:eval([[
+        local vars = require('cartridge.vars').new('cartridge.confapplier')
+        vars.clusterwide_config = __old_config
+    ]])
+
+    local main_checksum = get_config_checksum(main_server)
+    local replica_checksum = get_config_checksum(replica_server)
+    t.assert_not_equals(main_checksum, replica_checksum, 'Nodes should have different config checksum')
+end