SERVER-46024 Collect /proc/vmstat swap statistics in FTDC

(cherry picked from commit 97e0731610f18576229c322c3a4186202c77c521)

4 files changed, 199 insertions, 0 deletions

diff --git a/src/mongo/db/ftdc/ftdc_system_stats_linux.cpp b/src/mongo/db/ftdc/ftdc_system_stats_linux.cpp
index 21c0138cfde..70ab7d0c240 100644
--- a/src/mongo/db/ftdc/ftdc_system_stats_linux.cpp
+++ b/src/mongo/db/ftdc/ftdc_system_stats_linux.cpp
@@ -72,6 +72,16 @@ static const std::vector<StringData> kNetstatKeys{
     "Tcp:"_sd, "Ip:"_sd, "TcpExt:"_sd, "IpExt:"_sd,
 };
 
+static const std::vector<StringData> kVMKeys{
+    "balloon_deflate"_sd,
+    "balloon_inflate"_sd,
+    "nr_mlock"_sd,
+    "pgfault"_sd,
+    "pgmajfault"_sd,
+    "pswpin"_sd,
+    "pswpout"_sd,
+};
+
 /**
  *  Collect metrics from the Linux /proc file system.
  */
@@ -125,6 +135,14 @@ public:
                                 &subObjBuilder);
             subObjBuilder.doneFast();
         }
+
+        {
+            BSONObjBuilder subObjBuilder(builder.subobjStart("vmstat"_sd));
+            processStatusErrors(
+                procparser::parseProcVMStatFile("/proc/vmstat"_sd, kVMKeys, &subObjBuilder),
+                &subObjBuilder);
+            subObjBuilder.doneFast();
+        }
     }
 
 private:
diff --git a/src/mongo/util/procparser.cpp b/src/mongo/util/procparser.cpp
index 625b8b7466a..e03006da07c 100644
--- a/src/mongo/util/procparser.cpp
+++ b/src/mongo/util/procparser.cpp
@@ -714,5 +714,94 @@ std::vector<std::string> findPhysicalDisks(StringData sysBlockPath) {
     return files;
 }
 
+// Here is an example of the type of string it supports:
+// Note: output has been trimmed
+//
+// For more information, see:
+// proc(5) man page
+//
+// > cat /proc/vmstat
+// nr_free_pages 2732282
+// nr_zone_inactive_anon 686253
+// nr_zone_active_anon 4975441
+// nr_zone_inactive_file 2332485
+// nr_zone_active_file 4791149
+// nr_zone_unevictable 0
+// nr_zone_write_pending 0
+// nr_mlock 0
+//
+Status parseProcVMStat(const std::vector<StringData>& keys,
+                       StringData data,
+                       BSONObjBuilder* builder) {
+    bool foundKeys = false;
+
+    using string_split_iterator = boost::split_iterator<StringData::const_iterator>;
+
+    // Split the file by lines.
+    // token_compress_on means the iterator skips over consecutive '\n'. This should not be a
+    // problem in normal /proc/vmstat output.
+    for (string_split_iterator lineIt = string_split_iterator(
+             data.begin(),
+             data.end(),
+             boost::token_finder([](char c) { return c == '\n'; }, boost::token_compress_on));
+         lineIt != string_split_iterator();
+         ++lineIt) {
+
+        StringData line(lineIt->begin(), lineIt->end());
+
+        // Split the line by spaces since this the delimiters for vmstat files.
+        // token_compress_on means the iterator skips over consecutive ' '. This is needed for
+        // every line.
+        string_split_iterator partIt = string_split_iterator(
+            line.begin(),
+            line.end(),
+            boost::token_finder([](char c) { return c == ' '; }, boost::token_compress_on));
+
+        // Skip processing this line if we do not have a key.
+        if (partIt == string_split_iterator()) {
+            continue;
+        }
+
+        StringData key(partIt->begin(), partIt->end());
+
+        ++partIt;
+
+        // Skip processing this line if we only have a key, and no number.
+        if (partIt == string_split_iterator()) {
+            continue;
+        }
+
+        // Check if the key is in the list. /proc/vmstat will have extra keys, and may not have the
+        // keys we want.
+        if (keys.empty() || std::find(keys.begin(), keys.end(), key) != keys.end()) {
+            foundKeys = true;
+
+            StringData stringValue(partIt->begin(), partIt->end());
+
+            uint64_t value;
+
+            if (!parseNumberFromString(stringValue, &value).isOK()) {
+                value = 0;
+            }
+
+            builder->appendNumber(key, static_cast<long long>(value));
+        }
+    }
+
+    return foundKeys ? Status::OK()
+                     : Status(ErrorCodes::NoSuchKey, "Failed to find any keys in vmstat string");
+}
+
+Status parseProcVMStatFile(StringData filename,
+                           const std::vector<StringData>& keys,
+                           BSONObjBuilder* builder) {
+    auto swString = readFileAsString(filename);
+    if (!swString.isOK()) {
+        return swString.getStatus();
+    }
+
+    return parseProcVMStat(keys, swString.getValue(), builder);
+}
+
 }  // namespace procparser
 }  // namespace mongo
diff --git a/src/mongo/util/procparser.h b/src/mongo/util/procparser.h
index 0db4fce3617..578fd68a3ab 100644
--- a/src/mongo/util/procparser.h
+++ b/src/mongo/util/procparser.h
@@ -136,5 +136,24 @@ Status parseProcDiskStatsFile(StringData filename,
  */
 std::vector<std::string> findPhysicalDisks(StringData directory);
 
+/**
+ * Read a string matching /proc/vmstat format, and write the specified list of keys in builder.
+ *
+ * keys - list of keys to output in BSON. If keys is empty, all keys are outputed.
+ * data - string to parsee
+ * builder - BSON output
+ */
+Status parseProcVMStat(const std::vector<StringData>& keys,
+                       StringData data,
+                       BSONObjBuilder* builder);
+
+/**
+ * Read from file, and write the specified list of keys in builder.
+ */
+Status parseProcVMStatFile(StringData filename,
+                           const std::vector<StringData>& keys,
+                           BSONObjBuilder* builder);
+
+
 }  // namespace procparser
 }  // namespace mongo
diff --git a/src/mongo/util/procparser_test.cpp b/src/mongo/util/procparser_test.cpp
index 1f6324543cb..d047cc4ab2b 100644
--- a/src/mongo/util/procparser_test.cpp
+++ b/src/mongo/util/procparser_test.cpp
@@ -97,6 +97,11 @@ StringMap toNestedStringMap(BSONObj& obj) {
     ASSERT_OK(procparser::parseProcDiskStats(_disks, _x, &builder)); \
     auto obj = builder.obj();                                        \
     auto stringMap = toNestedStringMap(obj);
+#define ASSERT_PARSE_VMSTAT(_keys, _x)                           \
+    BSONObjBuilder builder;                                      \
+    ASSERT_OK(procparser::parseProcVMStat(_keys, _x, &builder)); \
+    auto obj = builder.obj();                                    \
+    auto stringMap = toStringMap(obj);
 
 TEST(FTDCProcStat, TestStat) {
 
@@ -640,5 +645,73 @@ TEST(FTDCProcDiskStats, TestLocalDiskStats) {
     }
 }
 
+
+TEST(FTDCProcVMStat, TestVMStat) {
+
+    std::vector<StringData> keys{"Key1", "Key2", "Key3"};
+
+    // Normal case
+    {
+        ASSERT_PARSE_VMSTAT(keys, "Key1 123\nKey2 456");
+        ASSERT_KEY_AND_VALUE("Key1", 123UL);
+        ASSERT_KEY_AND_VALUE("Key2", 456UL);
+    }
+
+    // No newline
+    {
+        ASSERT_PARSE_VMSTAT(keys, "Key1 123 Key2 456");
+        ASSERT_KEY_AND_VALUE("Key1", 123UL);
+        ASSERT_NO_KEY("Key2");
+    }
+
+    // Key without value
+    {
+        ASSERT_PARSE_VMSTAT(keys, "Key1 123\nKey2");
+        ASSERT_KEY_AND_VALUE("Key1", 123UL);
+        ASSERT_NO_KEY("Key2");
+    }
+
+    // Empty string
+    {
+        BSONObjBuilder builder;
+        ASSERT_NOT_OK(procparser::parseProcVMStat(keys, "", &builder));
+    }
+}
+
+// Test we can parse the /proc/vmstat on this machine. Also assert we have the expected fields
+// This tests is designed to exercise our parsing code on various Linuxes and fail
+// Normally when run in the FTDC loop we return a non-fatal error so we may not notice the failure
+// otherwise.
+TEST(FTDCProcVMStat, TestLocalVMStat) {
+    std::vector<StringData> keys{
+        "balloon_deflate"_sd,
+        "balloon_inflate"_sd,
+        "nr_mlock"_sd,
+        "pgfault"_sd,
+        "pgmajfault"_sd,
+        "pswpin"_sd,
+        "pswpout"_sd,
+    };
+
+    BSONObjBuilder builder;
+
+    ASSERT_OK(procparser::parseProcVMStatFile("/proc/vmstat", keys, &builder));
+
+    BSONObj obj = builder.obj();
+    auto stringMap = toStringMap(obj);
+    ASSERT_KEY("nr_mlock");
+    ASSERT_KEY("pgmajfault");
+    ASSERT_KEY("pswpin");
+    ASSERT_KEY("pswpout");
+}
+
+
+TEST(FTDCProcVMStat, TestLocalNonExistentVMStat) {
+    std::vector<StringData> keys{};
+    BSONObjBuilder builder;
+
+    ASSERT_NOT_OK(procparser::parseProcVMStatFile("/proc/does_not_exist", keys, &builder));
+}
+
 }  // namespace
 }  // namespace mongo