1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
//! A thread local context for gathering performance counter efficiently
//! and transparently.

use std::fmt;

use rocks_sys as ll;

/// A thread local context for gathering performance counter efficiently
/// and transparently.
///
/// Use `SetPerfLevel(PerfLevel::kEnableTime)` to enable time stats.
#[derive(Debug)]
#[repr(C)]
pub struct PerfContext {
    /// total number of user key comparisons
    pub user_key_comparison_count: u64,
    /// total number of block cache hits
    pub block_cache_hit_count: u64,
    /// total number of block reads (with IO)
    pub block_read_count: u64,
    /// total number of bytes from block reads
    pub block_read_byte: u64,
    /// total nanos spent on block reads
    pub block_read_time: u64,
    /// total nanos spent on block checksum
    pub block_checksum_time: u64,
    /// total nanos spent on block decompression
    pub block_decompress_time: u64,
    /// total number of internal keys skipped over during iteration.
    ///
    /// There are several reasons for it:
    /// 1. when calling `Next()`, the iterator is in the position of the previous
    ///    key, so that we'll need to skip it. It means this counter will always
    ///    be incremented in `Next()`.
    /// 2. when calling `Next()`, we need to skip internal entries for the previous
    ///    keys that are overwritten.
    /// 3. when calling `Next()`, `Seek()` or `SeekToFirst()`, after previous key
    ///    before calling `Next()`, the seek key in `Seek()` or the beginning for
    ///    `SeekToFirst()`, there may be one or more deleted keys before the next
    ///    valid key that the operation should place the iterator to. We need
    ///    to skip both of the tombstone and updates hidden by the tombstones. The
    ///    tombstones are not included in this counter, while previous updates
    ///    hidden by the tombstones will be included here.
    /// 4. symmetric cases for `Prev()` and `SeekToLast()`
    ///
    /// `internal_recent_skipped_count` is not included in this counter.
    pub internal_key_skipped_count: u64,
    /// Total number of deletes and single deletes skipped over during iteration
    ///
    /// When calling `Next()`, `Seek()` or `SeekToFirst()`, after previous position
    /// before calling `Next()`, the seek key in `Seek()` or the beginning for
    /// `SeekToFirst()`, there may be one or more deleted keys before the next valid
    /// key. Every deleted key is counted once. We don't recount here if there are
    /// still older updates invalidated by the tombstones.
    pub internal_delete_skipped_count: u64,
    /// How many times iterators skipped over internal keys that are more recent
    /// than the snapshot that iterator is using.
    pub internal_recent_skipped_count: u64,
    /// How many values were fed into merge operator by iterators.
    pub internal_merge_count: u64,

    /// total nanos spent on getting snapshot
    pub get_snapshot_time: u64,
    /// total nanos spent on querying memtables
    pub get_from_memtable_time: u64,
    /// number of mem tables queried
    pub get_from_memtable_count: u64,
    /// total nanos spent after `Get()` finds a key
    pub get_post_process_time: u64,
    /// total nanos reading from output files
    pub get_from_output_files_time: u64,
    /// total nanos spent on seeking memtable
    pub seek_on_memtable_time: u64,
    /// number of seeks issued on memtable
    /// (including `SeekForPrev` but not `SeekToFirst` and `SeekToLast`)
    pub seek_on_memtable_count: u64,
    /// number of `Next()`s issued on memtable
    pub next_on_memtable_count: u64,
    /// number of `Prev()`s issued on memtable
    pub prev_on_memtable_count: u64,
    /// total nanos spent on seeking child iters
    pub seek_child_seek_time: u64,
    /// number of seek issued in child iterators
    pub seek_child_seek_count: u64,
    /// total nanos spent on the merge min heap
    pub seek_min_heap_time: u64,
    /// total nanos spent on the merge max heap
    pub seek_max_heap_time: u64,
    /// total nanos spent on seeking the internal entries
    pub seek_internal_seek_time: u64,
    /// total nanos spent on iterating internal entries to find the next user entry
    pub find_next_user_entry_time: u64,

    /// total nanos spent on writing to WAL
    pub write_wal_time: u64,
    /// total nanos spent on writing to mem tables
    pub write_memtable_time: u64,
    /// total nanos spent on delaying write
    pub write_delay_time: u64,
    /// total nanos spent on writing a record, excluding the above three times
    pub write_pre_and_post_process_time: u64,

    /// time spent on acquiring DB mutex.
    pub db_mutex_lock_nanos: u64,
    /// Time spent on waiting with a condition variable created with DB mutex.
    pub db_condition_wait_nanos: u64,
    /// Time spent on merge operator.
    pub merge_operator_time_nanos: u64,

    /// Time spent on reading index block from block cache or SST file
    pub read_index_block_nanos: u64,
    /// Time spent on reading filter block from block cache or SST file
    pub read_filter_block_nanos: u64,
    /// Time spent on creating data block iterator
    pub new_table_block_iter_nanos: u64,
    /// Time spent on creating a iterator of an SST file.
    pub new_table_iterator_nanos: u64,
    /// Time spent on seeking a key in data/index blocks
    pub block_seek_nanos: u64,
    /// Time spent on finding or creating a table reader
    pub find_table_nanos: u64,
    /// total number of mem table bloom hits
    pub bloom_memtable_hit_count: u64,
    /// total number of mem table bloom misses
    pub bloom_memtable_miss_count: u64,
    /// total number of SST table bloom hits
    pub bloom_sst_hit_count: u64,
    /// total number of SST table bloom misses
    pub bloom_sst_miss_count: u64,
}

impl PerfContext {
    /// PerfContext for current thread
    pub fn current() -> &'static mut PerfContext {
        unsafe {
            let ptr = ll::rocks_get_perf_context() as *mut PerfContext;
            ptr.as_mut().unwrap()
        }
    }

    /// reset all performance counters to zero
    pub fn reset(&mut self) {
        unsafe {
            let ptr = self as *mut PerfContext as *mut ll::rocks_perf_context_t;
            ll::rocks_perf_context_reset(ptr);
        }
    }
}

impl fmt::Display for PerfContext {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        let mut s = String::new();
        let ptr = self as *const PerfContext as *const ll::rocks_perf_context_t;
        let exclude_zero_counters = false;
        unsafe {
            ll::rocks_perf_context_to_string(ptr, exclude_zero_counters as u8, &mut s as *mut String as *mut _);
        }
        write!(f, "{}", s)
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use super::super::rocksdb::*;

    #[test]
    fn perf_context() {
        set_perf_level(PerfLevel::EnableTime);

        let tmp_dir = ::tempdir::TempDir::new_in(".", "rocks").unwrap();
        let db = DB::open(Options::default().map_db_options(|db| db.create_if_missing(true)), &tmp_dir).unwrap();

        assert!(
            db.put(&Default::default(), b"long-key", vec![b'A'; 1024 * 1024].as_ref())
                .is_ok()
        );
        assert!(db.put(&Default::default(), b"a", b"1").is_ok());
        assert!(db.put(&Default::default(), b"b", b"2").is_ok());
        assert!(db.put(&Default::default(), b"c", b"3").is_ok());

        assert!(db.compact_range(&Default::default(), ..).is_ok());

        assert!(db.get(&Default::default(), b"long-key").is_ok());

        let stat = PerfContext::current();

        assert!(stat.block_read_count > 0);
        assert!(stat.user_key_comparison_count > 0);

        assert!(stat.to_string().len() > 200);

        println!("dbg => {:?}", stat);
        println!("show => {}", stat);

        stat.reset();
        assert_eq!(stat.user_key_comparison_count, 0);
    }
}