prometheus-client-mmap 0.19.1 → 0.20.0

data/ext/fast_mmaped_file_rs/src/map.rs

@@ -0,0 +1,458 @@
+use ahash::AHashMap;
+use magnus::{exception::*, Error, RArray};
+use std::mem::size_of;
+
+use crate::error::MmapError;
+use crate::file_entry::{EntryData, EntryMetadata, FileEntry};
+use crate::file_info::FileInfo;
+use crate::raw_entry::RawEntry;
+use crate::util::read_u32;
+use crate::Result;
+use crate::{err, HEADER_SIZE};
+
+/// A HashMap of JSON strings and their associated metadata.
+/// Used to print metrics in text format.
+///
+/// The map key is the entry's JSON string and an optional pid string. The latter
+/// allows us to have multiple entries on the map for multiple pids using the
+/// same string.
+#[derive(Default, Debug)]
+pub struct EntryMap(AHashMap<EntryData, EntryMetadata>);
+
+impl EntryMap {
+    /// Construct a new EntryMap.
+    pub fn new() -> Self {
+        Self(AHashMap::new())
+    }
+
+    /// Given a list of files, read each one into memory and parse the metrics it contains.
+    pub fn aggregate_files(&mut self, list_of_files: RArray) -> magnus::error::Result<()> {
+        // Pre-allocate the `HashMap` and validate we don't OOM. The C implementation
+        // ignores allocation failures here. We perform this check to avoid potential
+        // panics. We assume ~1,000 entries per file, so 72 KiB allocated per file.
+        self.0
+            .try_reserve(list_of_files.len() * 1024)
+            .map_err(|_| {
+                err!(
+                    no_mem_error(),
+                    "Couldn't allocate for {} memory",
+                    size_of::<FileEntry>() * list_of_files.len() * 1024
+                )
+            })?;
+
+        // We expect file sizes between 4KiB and 4MiB. Pre-allocate 16KiB to reduce reallocations
+        // a bit.
+        let mut buf = Vec::new();
+        buf.try_reserve(16_384)
+            .map_err(|_| err!(no_mem_error(), "Couldn't allocate for {} memory", 16_384))?;
+
+        for item in list_of_files.each() {
+            let params = RArray::from_value(item?).expect("file list was not a Ruby Array");
+            if params.len() != 4 {
+                return Err(err!(
+                    arg_error(),
+                    "wrong number of arguments {} instead of 4",
+                    params.len()
+                ));
+            }
+
+            let params = params.to_value_array::<4>()?;
+
+            let mut file_info = FileInfo::open_from_params(&params)?;
+            file_info.read_from_file(&mut buf)?;
+            self.process_buffer(file_info, &buf)?;
+        }
+        Ok(())
+    }
+
+    /// Consume the `EntryMap` and convert the key/value into`FileEntry`
+    /// objects, sorting them by their JSON strings.
+    pub fn into_sorted(self) -> Result<Vec<FileEntry>> {
+        let mut sorted = Vec::new();
+
+        // To match the behavior of the C version, pre-allocate the entries
+        // and check for allocation failure. Generally idiomatic Rust would
+        // `collect` the iterator into a new `Vec` in place, but this panics
+        // if it can't allocate and we want to continue execution in that
+        // scenario.
+        if sorted.try_reserve_exact(self.0.len()).is_err() {
+            return Err(MmapError::OutOfMemory(
+                self.0.len() * size_of::<FileEntry>(),
+            ));
+        }
+
+        sorted.extend(
+            self.0
+                .into_iter()
+                .map(|(data, meta)| FileEntry { data, meta }),
+        );
+
+        sorted.sort_unstable_by(|x, y| x.data.cmp(&y.data));
+
+        Ok(sorted)
+    }
+
+    /// Check if the `EntryMap` already contains the JSON string.
+    /// If yes, update the associated value, if not insert the
+    /// entry into the map.
+    pub fn merge_or_store(&mut self, data: EntryData, meta: EntryMetadata) {
+        if let Some(existing) = self.0.get_mut(&data) {
+            existing.merge(&meta);
+        } else {
+            self.0.insert(data, meta);
+        }
+    }
+
+    /// Parse metrics data from a `.db` file and store in the `EntryMap`.
+    fn process_buffer(&mut self, file_info: FileInfo, source: &[u8]) -> Result<()> {
+        if source.len() < HEADER_SIZE {
+            // Nothing to read, OK.
+            return Ok(());
+        }
+
+        // CAST: no-op on 32-bit, widening on 64-bit.
+        let used = read_u32(source, 0)? as usize;
+
+        if used > source.len() {
+            return Err(MmapError::PromParsing(format!(
+                "source file {} corrupted, used {used} > file size {}",
+                file_info.path.display(),
+                source.len()
+            )));
+        }
+
+        let mut pos = HEADER_SIZE;
+
+        while pos + size_of::<u32>() < used {
+            let raw_entry = RawEntry::from_slice(&source[pos..used])?;
+
+            if pos + raw_entry.total_len() > used {
+                return Err(MmapError::PromParsing(format!(
+                    "source file {} corrupted, used {used} < stored data length {}",
+                    file_info.path.display(),
+                    pos + raw_entry.total_len()
+                )));
+            }
+
+            let meta = EntryMetadata::new(&raw_entry, &file_info)?;
+            let data = EntryData::new(&raw_entry, &file_info, meta.is_pid_significant())?;
+
+            self.merge_or_store(data, meta);
+
+            pos += raw_entry.total_len();
+        }
+
+        Ok(())
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use magnus::{StaticSymbol, Symbol};
+    use std::mem;
+
+    use super::*;
+    use crate::file_entry::FileEntry;
+    use crate::testhelper::{self, TestFile};
+
+    #[test]
+    fn test_into_sorted() {
+        let _cleanup = unsafe { magnus::embed::init() };
+        let ruby = magnus::Ruby::get().unwrap();
+        crate::init(&ruby).unwrap();
+
+        let entries = vec![
+            FileEntry {
+                data: EntryData {
+                    json: "zzzzzz".to_string(),
+                    pid: Some("worker-0_0".to_string()),
+                },
+                meta: EntryMetadata {
+                    multiprocess_mode: Symbol::new("max"),
+                    type_: StaticSymbol::new("gauge"),
+                    value: 1.0,
+                },
+            },
+            FileEntry {
+                data: EntryData {
+                    json: "zzz".to_string(),
+                    pid: Some("worker-0_0".to_string()),
+                },
+                meta: EntryMetadata {
+                    multiprocess_mode: Symbol::new("max"),
+                    type_: StaticSymbol::new("gauge"),
+                    value: 1.0,
+                },
+            },
+            FileEntry {
+                data: EntryData {
+                    json: "zzzaaa".to_string(),
+                    pid: Some("worker-0_0".to_string()),
+                },
+                meta: EntryMetadata {
+                    multiprocess_mode: Symbol::new("max"),
+                    type_: StaticSymbol::new("gauge"),
+                    value: 1.0,
+                },
+            },
+            FileEntry {
+                data: EntryData {
+                    json: "aaa".to_string(),
+                    pid: Some("worker-0_0".to_string()),
+                },
+                meta: EntryMetadata {
+                    multiprocess_mode: Symbol::new("max"),
+                    type_: StaticSymbol::new("gauge"),
+                    value: 1.0,
+                },
+            },
+            FileEntry {
+                data: EntryData {
+                    json: "ooo".to_string(),
+                    pid: Some("worker-1_0".to_string()),
+                },
+                meta: EntryMetadata {
+                    multiprocess_mode: Symbol::new("all"),
+                    type_: StaticSymbol::new("gauge"),
+                    value: 1.0,
+                },
+            },
+            FileEntry {
+                data: EntryData {
+                    json: "ooo".to_string(),
+                    pid: Some("worker-0_0".to_string()),
+                },
+                meta: EntryMetadata {
+                    multiprocess_mode: Symbol::new("all"),
+                    type_: StaticSymbol::new("gauge"),
+                    value: 1.0,
+                },
+            },
+        ];
+
+        let mut map = EntryMap::new();
+
+        for entry in entries {
+            map.0.insert(entry.data, entry.meta);
+        }
+
+        let result = map.into_sorted();
+        assert!(result.is_ok());
+        let sorted = result.unwrap();
+        assert_eq!(sorted.len(), 6);
+        assert_eq!(sorted[0].data.json, "aaa");
+        assert_eq!(sorted[1].data.json, "ooo");
+        assert_eq!(sorted[1].data.pid.as_deref(), Some("worker-0_0"));
+        assert_eq!(sorted[2].data.json, "ooo");
+        assert_eq!(sorted[2].data.pid.as_deref(), Some("worker-1_0"));
+        assert_eq!(sorted[3].data.json, "zzz");
+        assert_eq!(sorted[4].data.json, "zzzaaa");
+        assert_eq!(sorted[5].data.json, "zzzzzz");
+    }
+
+    #[test]
+    fn test_merge_or_store() {
+        let _cleanup = unsafe { magnus::embed::init() };
+        let ruby = magnus::Ruby::get().unwrap();
+        crate::init(&ruby).unwrap();
+
+        let key = "foobar";
+
+        let starting_entry = FileEntry {
+            data: EntryData {
+                json: key.to_string(),
+                pid: Some("worker-0_0".to_string()),
+            },
+            meta: EntryMetadata {
+                multiprocess_mode: Symbol::new("all"),
+                type_: StaticSymbol::new("gauge"),
+                value: 1.0,
+            },
+        };
+
+        let matching_entry = FileEntry {
+            data: EntryData {
+                json: key.to_string(),
+                pid: Some("worker-0_0".to_string()),
+            },
+            meta: EntryMetadata {
+                multiprocess_mode: Symbol::new("all"),
+                type_: StaticSymbol::new("gauge"),
+                value: 5.0,
+            },
+        };
+
+        let same_key_different_worker = FileEntry {
+            data: EntryData {
+                json: key.to_string(),
+                pid: Some("worker-1_0".to_string()),
+            },
+            meta: EntryMetadata {
+                multiprocess_mode: Symbol::new("all"),
+                type_: StaticSymbol::new("gauge"),
+                value: 100.0,
+            },
+        };
+
+        let unmatched_entry = FileEntry {
+            data: EntryData {
+                json: "another key".to_string(),
+                pid: Some("worker-0_0".to_string()),
+            },
+            meta: EntryMetadata {
+                multiprocess_mode: Symbol::new("all"),
+                type_: StaticSymbol::new("gauge"),
+                value: 1.0,
+            },
+        };
+
+        let mut map = EntryMap::new();
+
+        map.0
+            .insert(starting_entry.data.clone(), starting_entry.meta.clone());
+
+        map.merge_or_store(matching_entry.data, matching_entry.meta);
+
+        assert_eq!(
+            5.0,
+            map.0.get(&starting_entry.data).unwrap().value,
+            "value updated"
+        );
+        assert_eq!(1, map.0.len(), "no entry added");
+
+        map.merge_or_store(
+            same_key_different_worker.data,
+            same_key_different_worker.meta,
+        );
+
+        assert_eq!(
+            5.0,
+            map.0.get(&starting_entry.data).unwrap().value,
+            "value unchanged"
+        );
+
+        assert_eq!(2, map.0.len(), "additional entry added");
+
+        map.merge_or_store(unmatched_entry.data, unmatched_entry.meta);
+
+        assert_eq!(
+            5.0,
+            map.0.get(&starting_entry.data).unwrap().value,
+            "value unchanged"
+        );
+        assert_eq!(3, map.0.len(), "entry added");
+    }
+
+    #[test]
+    fn test_process_buffer() {
+        struct TestCase {
+            name: &'static str,
+            json: &'static [&'static str],
+            values: &'static [f64],
+            used: Option<u32>,
+            expected_ct: usize,
+            expected_err: Option<MmapError>,
+        }
+
+        let _cleanup = unsafe { magnus::embed::init() };
+        let ruby = magnus::Ruby::get().unwrap();
+        crate::init(&ruby).unwrap();
+
+        let tc = vec![
+            TestCase {
+                name: "single entry",
+                json: &[
+                    r#"["first_family","first_name",["label_a","label_b"],["value_a","value_b"]]"#,
+                ],
+                values: &[1.0],
+                used: None,
+                expected_ct: 1,
+                expected_err: None,
+            },
+            TestCase {
+                name: "multiple entries",
+                json: &[
+                    r#"["first_family","first_name",["label_a","label_b"],["value_a","value_b"]]"#,
+                    r#"["second_family","first_name",["label_a","label_b"],["value_a","value_b"]]"#,
+                ],
+                values: &[1.0, 2.0],
+                used: None,
+                expected_ct: 2,
+                expected_err: None,
+            },
+            TestCase {
+                name: "empty",
+                json: &[],
+                values: &[],
+                used: None,
+                expected_ct: 0,
+                expected_err: None,
+            },
+            TestCase {
+                name: "used too long",
+                json: &[
+                    r#"["first_family","first_name",["label_a","label_b"],["value_a","value_b"]]"#,
+                ],
+                values: &[1.0],
+                used: Some(9999),
+                expected_ct: 0,
+                expected_err: Some(MmapError::PromParsing(String::new())),
+            },
+            TestCase {
+                name: "used too short",
+                json: &[
+                    r#"["first_family","first_name",["label_a","label_b"],["value_a","value_b"]]"#,
+                ],
+                values: &[1.0],
+                used: Some(15),
+                expected_ct: 0,
+                expected_err: Some(MmapError::out_of_bounds(88, 7)),
+            },
+        ];
+
+        for case in tc {
+            let name = case.name;
+
+            let input_bytes = testhelper::entries_to_db(case.json, case.values, case.used);
+
+            let TestFile {
+                file,
+                path,
+                dir: _dir,
+            } = TestFile::new(&input_bytes);
+
+            let info = FileInfo {
+                file,
+                path,
+                len: case.json.len(),
+                multiprocess_mode: Symbol::new("max"),
+                type_: StaticSymbol::new("gauge"),
+                pid: "worker-1".to_string(),
+            };
+
+            let mut map = EntryMap::new();
+            let result = map.process_buffer(info, &input_bytes);
+
+            assert_eq!(case.expected_ct, map.0.len(), "test case: {name} - count");
+
+            if let Some(expected_err) = case.expected_err {
+                // Validate we have the right enum type for the error. Error
+                // messages contain the temp dir path and can't be predicted
+                // exactly.
+                assert_eq!(
+                    mem::discriminant(&expected_err),
+                    mem::discriminant(&result.unwrap_err()),
+                    "test case: {name} - failure"
+                );
+            } else {
+                assert_eq!(Ok(()), result, "test case: {name} - success");
+
+                assert_eq!(
+                    case.json.len(),
+                    map.0.len(),
+                    "test case: {name} - all entries captured"
+                );
+            }
+        }
+    }
+}

data/ext/fast_mmaped_file_rs/src/mmap.rs

@@ -0,0 +1,151 @@
+use magnus::typed_data::Obj;
+use magnus::value::Fixnum;
+use magnus::{Integer, RArray, RClass, RHash, RString, Value};
+
+use crate::file_entry::FileEntry;
+use crate::map::EntryMap;
+
+/// A Rust struct wrapped in a Ruby object, providing access to a memory-mapped
+/// file used to store, update, and read out Prometheus metrics.
+///
+/// - File format:
+///     - Header:
+///         - 4 bytes: u32 - total size of metrics in file.
+///         - 4 bytes: NUL byte padding.
+///     - Repeating metrics entries:
+///         - 4 bytes: u32 - entry JSON string size.
+///         - `N` bytes: UTF-8 encoded JSON string used as entry key.
+///         - (8 - (4 + `N`) % 8) bytes: 1 to 8 padding space (0x20) bytes to
+///           reach 8-byte alignment.
+///         - 8 bytes: f64 - entry value.
+///
+/// All numbers are saved in native-endian format.
+///
+/// Generated via [luismartingarcia/protocol](https://github.com/luismartingarcia/protocol):
+///
+///
+/// ```
+/// protocol "Used:4,Pad:4,K1 Size:4,K1 Name:4,K1 Value:8,K2 Size:4,K2 Name:4,K2 Value:8"
+///
+/// 0                   1                   2                   3
+/// 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+/// +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+/// |  Used |  Pad  |K1 Size|K1 Name|   K1 Value    |K2 Size|K2 Name|
+/// +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+/// |  K2 Value   |
+/// +-+-+-+-+-+-+-+
+/// ```
+#[derive(Debug, Default)]
+#[magnus::wrap(class = "FastMmapedFileRs", free_immediately, size)]
+pub struct MmapedFile;
+
+impl MmapedFile {
+    /// call-seq:
+    ///   new(file)
+    ///
+    /// create a new Mmap object
+    ///
+    /// * <em>file</em>
+    ///
+    ///
+    ///     Creates a mapping that's shared with all other processes
+    ///     mapping the same area of the file.
+    pub fn new(_klass: RClass, _args: &[Value]) -> magnus::error::Result<Obj<Self>> {
+        Ok(Obj::wrap(Self))
+    }
+
+    /// Initialize a new `FastMmapedFileRs` object. This must be defined in
+    /// order for inheritance to work.
+    pub fn initialize(_rb_self: Obj<Self>, _fname: String) -> magnus::error::Result<()> {
+        unimplemented!();
+    }
+
+    /// Read the list of files provided from Ruby and convert them to a Prometheus
+    /// metrics String.
+    pub fn to_metrics(file_list: RArray) -> magnus::error::Result<String> {
+        let mut map = EntryMap::new();
+        map.aggregate_files(file_list)?;
+
+        let sorted = map.into_sorted()?;
+
+        FileEntry::entries_to_string(sorted).map_err(|e| e.into())
+    }
+
+    /// Document-method: []
+    /// Document-method: slice
+    ///
+    /// call-seq: [](args)
+    ///
+    /// Element reference - with the following syntax:
+    ///
+    ///   self[nth]
+    ///
+    /// retrieve the <em>nth</em> character
+    ///
+    ///   self[start..last]
+    ///
+    /// return a substring from <em>start</em> to <em>last</em>
+    ///
+    ///   self[start, length]
+    ///
+    /// return a substring of <em>lenght</em> characters from <em>start</em>
+    pub fn slice(_rb_self: Obj<Self>, _args: &[Value]) -> magnus::error::Result<RString> {
+        unimplemented!();
+    }
+
+    /// Document-method: msync
+    /// Document-method: sync
+    /// Document-method: flush
+    ///
+    /// call-seq: msync
+    ///
+    /// flush the file
+    pub fn sync(&self, _args: &[Value]) -> magnus::error::Result<()> {
+        unimplemented!();
+    }
+
+    /// Document-method: munmap
+    /// Document-method: unmap
+    ///
+    /// call-seq: munmap
+    ///
+    /// terminate the association
+    pub fn munmap(_rb_self: Obj<Self>) -> magnus::error::Result<()> {
+        unimplemented!();
+    }
+
+    /// Fetch the `used` header from the `.db` file, the length
+    /// in bytes of the data written to the file.
+    pub fn load_used(&self) -> magnus::error::Result<Integer> {
+        unimplemented!();
+    }
+
+    /// Update the `used` header for the `.db` file, the length
+    /// in bytes of the data written to the file.
+    pub fn save_used(_rb_self: Obj<Self>, _used: Fixnum) -> magnus::error::Result<Fixnum> {
+        unimplemented!();
+    }
+
+    /// Fetch the value associated with a key from the mmap.
+    /// If no entry is present, initialize with the default
+    /// value provided.
+    pub fn fetch_entry(
+        _rb_self: Obj<Self>,
+        _positions: RHash,
+        _key: RString,
+        _default_value: f64,
+    ) -> magnus::error::Result<f64> {
+        unimplemented!();
+    }
+
+    /// Update the value of an existing entry, if present. Otherwise create a new entry
+    /// for the key.
+    pub fn upsert_entry(
+        _rb_self: Obj<Self>,
+        _positions: RHash,
+        _key: RString,
+        _value: f64,
+    ) -> magnus::error::Result<f64> {
+        unimplemented!();
+    }
+}