Add notification for user on max map count (#1408)

src/memgraph.cpp

@@ -9,6 +9,7 @@
 // by the Apache License, Version 2.0, included in the file
 // licenses/APL.txt.
 
+#include <cstdint>
 #include "audit/log.hpp"
 #include "communication/metrics.hpp"
 #include "communication/websocket/auth.hpp"
@@ -48,6 +49,7 @@ namespace {
 constexpr const char *kMgUser = "MEMGRAPH_USER";
 constexpr const char *kMgPassword = "MEMGRAPH_PASSWORD";
 constexpr const char *kMgPassfile = "MEMGRAPH_PASSFILE";
+constexpr uint64_t kMgVmMaxMapCount = 262144;
 
 // TODO: move elsewhere so that we can remove need of interpreter.hpp
 void InitFromCypherlFile(memgraph::query::InterpreterContext &ctx, memgraph::dbms::DatabaseAccess &db_acc,
@@ -207,6 +209,17 @@ int main(int argc, char **argv) {
   std::cout << "You are running Memgraph v" << gflags::VersionString() << std::endl;
   std::cout << "To get started with Memgraph, visit https://memgr.ph/start" << std::endl;
 
+  const auto vm_max_map_count = memgraph::utils::GetVmMaxMapCount();
+  if (vm_max_map_count.has_value()) {
+    if (vm_max_map_count.value() < kMgVmMaxMapCount) {
+      std::cout << "Max virtual memory areas vm.max_map_count " << vm_max_map_count.value()
+                << " is too low, increase to at least " << kMgVmMaxMapCount << std::endl;
+    }
+  } else {
+    std::cout << "Can't get info on vm.max_map_count, check whether it is too low, vm.max_map_count is at least "
+              << kMgVmMaxMapCount << std::endl;
+  }
+
   auto data_directory = std::filesystem::path(FLAGS_data_directory);
 
   memgraph::utils::EnsureDirOrDie(data_directory);

src/query/interpreter.cpp

@@ -95,6 +95,7 @@
 #include "utils/on_scope_exit.hpp"
 #include "utils/readable_size.hpp"
 #include "utils/settings.hpp"
+#include "utils/stat.hpp"
 #include "utils/string.hpp"
 #include "utils/tsc.hpp"
 #include "utils/typeinfo.hpp"
@@ -3092,11 +3093,15 @@ PreparedQuery PrepareSystemInfoQuery(ParsedQuery parsed_query, bool in_explicit_
       handler = [storage = current_db.db_acc_->get()->storage(), interpreter_isolation_level,
                  next_transaction_isolation_level] {
         auto info = storage->GetBaseInfo();
+        const auto vm_max_map_count = utils::GetVmMaxMapCount();
+        const int64_t vm_max_map_count_storage_info =
+            vm_max_map_count.has_value() ? vm_max_map_count.value() : memgraph::utils::VM_MAX_MAP_COUNT_DEFAULT;
         std::vector<std::vector<TypedValue>> results{
             {TypedValue("name"), TypedValue(storage->id())},
             {TypedValue("vertex_count"), TypedValue(static_cast<int64_t>(info.vertex_count))},
             {TypedValue("edge_count"), TypedValue(static_cast<int64_t>(info.edge_count))},
             {TypedValue("average_degree"), TypedValue(info.average_degree)},
+            {TypedValue("vm_max_map_count"), TypedValue(vm_max_map_count_storage_info)},
             {TypedValue("memory_res"), TypedValue(utils::GetReadableSize(static_cast<double>(info.memory_res)))},
             {TypedValue("disk_usage"), TypedValue(utils::GetReadableSize(static_cast<double>(info.disk_usage)))},
             {TypedValue("memory_tracked"),

src/storage/v2/inmemory/storage.cpp

@@ -20,6 +20,7 @@
 #include "storage/v2/inmemory/replication/replication_client.hpp"
 #include "storage/v2/inmemory/unique_constraints.hpp"
 #include "utils/resource_lock.hpp"
+#include "utils/stat.hpp"
 
 namespace memgraph::storage {
 

src/telemetry/collectors.cpp

@@ -78,8 +78,12 @@ const nlohmann::json GetResourceUsage(std::filesystem::path root_directory) {
   }
   auto cpu_total = GetCpuUsage(pid);
   cpu["usage"] = cpu_total.second;
-
-  return {{"cpu", cpu}, {"memory", utils::GetMemoryRES()}, {"disk", utils::GetDirDiskUsage(root_directory)}};
+  const auto vm_max_map_count = utils::GetVmMaxMapCount();
+  return {{"cpu", cpu},
+          {"memory", utils::GetMemoryRES()},
+          {"disk", utils::GetDirDiskUsage(root_directory)},
+          {"vm_max_map_count",
+           vm_max_map_count.has_value() ? *vm_max_map_count : memgraph::utils::VM_MAX_MAP_COUNT_DEFAULT}};
 }
 
 }  // namespace memgraph::telemetry

src/utils/stat.hpp

@@ -11,7 +11,9 @@
 
 #pragma once
 
+#include <cstddef>
 #include <filesystem>
+#include <optional>
 
 #include <unistd.h>
 
@@ -20,6 +22,8 @@
 
 namespace memgraph::utils {
 
+static constexpr int64_t VM_MAX_MAP_COUNT_DEFAULT{-1};
+
 /// Returns the number of bytes a directory is using on disk. If the given path
 /// isn't a directory, zero will be returned.
 template <bool IgnoreSymlink = true>
@@ -54,4 +58,20 @@ inline uint64_t GetMemoryRES() {
   return memory;
 }
 
+/// Returns the size of vm.max_map_count
+inline std::optional<int64_t> GetVmMaxMapCount() {
+  auto vm_max_map_count_data = utils::ReadLines("/proc/sys/vm/max_map_count");
+  if (vm_max_map_count_data.empty()) {
+    return std::nullopt;
+  }
+  if (vm_max_map_count_data.size() != 1) {
+    return std::nullopt;
+  }
+  const auto parts{utils::Split(vm_max_map_count_data[0])};
+  if (parts.size() != 1) {
+    return std::nullopt;
+  }
+  return std::stoi(parts[0]);
+}
+
 }  // namespace memgraph::utils

tests/e2e/configuration/storage_info.py

@@ -53,10 +53,15 @@ def test_does_default_config_match():
     config = cursor.fetchall()
 
     # The default value of these is dependent on the given machine.
-    machine_dependent_configurations = ["memory_res", "disk_usage", "memory_tracked", "allocation_limit"]
-
+    machine_dependent_configurations = [
+        "memory_res",
+        "disk_usage",
+        "memory_tracked",
+        "allocation_limit",
+        "vm_max_map_count",
+    ]
     # Number of different data-points returned by SHOW STORAGE INFO
-    assert len(config) == 12
+    assert len(config) == 13
 
     for conf in config:
         conf_name = conf[0]

tests/integration/telemetry/server.py

@@ -148,6 +148,7 @@ def verify_storage(storage, args):
             assert "cpu" in item["data"]["resources"]
             assert "memory" in item["data"]["resources"]
             assert "disk" in item["data"]["resources"]
+            assert "vm_max_map_count" in item["data"]["resources"]
             assert "uptime" in item["data"]
 
             uptime = item["data"]["uptime"]