From dda4d0e34e975e248848dcbd08956e675d06825d Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Mon, 5 Jul 2021 04:18:12 +0300
Subject: [PATCH] Add EDAC

---
 src/Interpreters/AsynchronousMetrics.cpp | 47 ++++++++++++++++++++++++
 src/Interpreters/AsynchronousMetrics.h   |  8 +++-
 2 files changed, 54 insertions(+), 1 deletion(-)

diff --git a/src/Interpreters/AsynchronousMetrics.cpp b/src/Interpreters/AsynchronousMetrics.cpp
index 014ac4639c0..c2184f78d6f 100644
--- a/src/Interpreters/AsynchronousMetrics.cpp
+++ b/src/Interpreters/AsynchronousMetrics.cpp
@@ -109,6 +109,7 @@ AsynchronousMetrics::AsynchronousMetrics(
         String hwmon_name;
         ReadBufferFromFile hwmon_name_in(hwmon_name_file, small_buffer_size);
         readText(hwmon_name, hwmon_name_in);
+        std::replace(hwmon_name.begin(), hwmon_name.end(), ' ', '_');
 
         for (size_t sensor_index = 0;; ++sensor_index)
         {
@@ -136,11 +137,36 @@ AsynchronousMetrics::AsynchronousMetrics(
             {
                 ReadBufferFromFile sensor_name_in(sensor_name_file, small_buffer_size);
                 readText(sensor_name, sensor_name_in);
+                std::replace(sensor_name.begin(), sensor_name.end(), ' ', '_');
             }
 
             hwmon_devices[hwmon_name][sensor_name] = std::move(file);
         }
     }
+
+    for (size_t edac_index = 0;; ++edac_index)
+    {
+        String edac_correctable_file = fmt::format("/sys/devices/system/edac/mc/mc{}/ce_count", edac_index);
+        String edac_uncorrectable_file = fmt::format("/sys/devices/system/edac/mc/mc{}/ue_count", edac_index);
+
+        bool edac_correctable_file_exists = std::filesystem::exists(edac_correctable_file);
+        bool edac_uncorrectable_file_exists = std::filesystem::exists(edac_uncorrectable_file);
+
+        if (!edac_correctable_file_exists && !edac_uncorrectable_file_exists)
+        {
+            if (edac_index == 0)
+                continue;
+            else
+                break;
+        }
+
+        edac.emplace_back();
+
+        if (edac_correctable_file_exists)
+            edac.back().first = openFileIfExists(edac_correctable_file);
+        if (edac_uncorrectable_file_exists)
+            edac.back().second = openFileIfExists(edac_uncorrectable_file);
+    }
 #endif
 }
 
@@ -741,6 +767,27 @@ void AsynchronousMetrics::update(std::chrono::system_clock::time_point update_ti
                 new_values[fmt::format("Temperature_{}_{}", hwmon_name, sensor_name)] = temperature * 0.001;
         }
     }
+
+    for (size_t i = 0, size = edac.size(); i < size; ++i)
+    {
+        if (edac[i].first)
+        {
+            ReadBufferFromFile & in = *edac[i].first;
+            in.rewind();
+            uint64_t errors = 0;
+            readText(errors, in);
+            new_values[fmt::format("EDAC{}_Correctable", i)] = errors;
+        }
+
+        if (edac[i].second)
+        {
+            ReadBufferFromFile & in = *edac[i].second;
+            in.rewind();
+            uint64_t errors = 0;
+            readText(errors, in);
+            new_values[fmt::format("EDAC{}_Uncorrectable", i)] = errors;
+        }
+    }
 #endif
 
     /// Free space in filesystems at data path and logs path.
diff --git a/src/Interpreters/AsynchronousMetrics.h b/src/Interpreters/AsynchronousMetrics.h
index 32226d5d25b..8c0a00df2a3 100644
--- a/src/Interpreters/AsynchronousMetrics.h
+++ b/src/Interpreters/AsynchronousMetrics.h
@@ -6,6 +6,7 @@
 #include <IO/ReadBufferFromFile.h>
 
 #include <condition_variable>
+#include <map>
 #include <mutex>
 #include <string>
 #include <thread>
@@ -84,13 +85,18 @@ private:
     std::optional<ReadBufferFromFile> cpuinfo;
     std::optional<ReadBufferFromFile> file_nr;
     std::optional<ReadBufferFromFile> uptime;
+
     std::vector<std::unique_ptr<ReadBufferFromFile>> thermal;
 
     std::unordered_map<String /* device name */,
         std::unordered_map<String /* label name */,
             std::unique_ptr<ReadBufferFromFile>>> hwmon_devices;
 
-    /// TODO: IO load, Network rx/tx, sockets, EDAC.
+    std::vector<std::pair<
+        std::unique_ptr<ReadBufferFromFile> /* correctable errors */,
+        std::unique_ptr<ReadBufferFromFile> /* uncorrectable errors */>> edac;
+
+    /// TODO: IO load, Network rx/tx, sockets.
 
     struct ProcStatValuesCPU
     {