chroma_types/
log.rs

1use crate::CollectionUuid;
2
3////////////////////////////////////// constants and variables /////////////////////////////////////
4
5pub fn dirty_log_path_from_hostname(hostname: &str) -> String {
6    format!("dirty-{}", hostname)
7}
8
9//////////////////////////////////////////// DirtyMarker ///////////////////////////////////////////
10
11/// Markers for tracking collection compaction state changes.
12///
13/// DirtyMarker represents state transitions in the compaction lifecycle of collections.
14/// The enum is designed for forwards/backwards compatibility - new variants can be added
15/// and handled independently while maintaining compatibility with older code.
16#[derive(Clone, Debug, Eq, PartialEq, serde::Serialize, serde::Deserialize)]
17// NOTE(rescrv):  This is intentionally an enum for easy forwards/backwards compatibility.  Add a
18// new variant, handle both variants, cycle logs, stop handling old variant.
19// TODO(rescrv):  Dedupe with log-service crate.
20pub enum DirtyMarker {
21    /// Marks a collection as needing compaction due to new records.
22    #[serde(rename = "mark_dirty")]
23    MarkDirty {
24        /// The collection requiring compaction.
25        collection_id: CollectionUuid,
26        /// The position in the write-ahead log where this marker was created.
27        log_position: u64,
28        /// The total number of records in the collection.
29        num_records: u64,
30        /// The number of times this collection has been reinserted into the heap.
31        reinsert_count: u64,
32        /// The epoch time in microseconds when this collection was first marked dirty.
33        initial_insertion_epoch_us: u64,
34    },
35    /// Removes all compaction scheduling for a collection.
36    #[serde(rename = "purge")]
37    Purge {
38        /// The collection to purge from the compaction heap.
39        collection_id: CollectionUuid,
40    },
41    // A Cleared marker is a no-op.  It exists so that a log consisting of mark-dirty markers that
42    // map onto purge markers will be cleared and can be erased.
43    /// A no-op marker used for log compaction.
44    ///
45    /// When a log contains mark-dirty markers that have been purged, those entries
46    /// can be replaced with Cleared markers to allow log truncation.
47    #[serde(rename = "clear")]
48    Cleared,
49}
50
51impl DirtyMarker {
52    /// The collection ID for a given dirty marker.
53    pub fn collection_id(&self) -> CollectionUuid {
54        match self {
55            DirtyMarker::MarkDirty { collection_id, .. } => *collection_id,
56            DirtyMarker::Purge { collection_id } => *collection_id,
57            DirtyMarker::Cleared => CollectionUuid::default(),
58        }
59    }
60
61    /// Increment any reinsert counter on the variant.
62    pub fn reinsert(&mut self) {
63        if let DirtyMarker::MarkDirty {
64            collection_id: _,
65            log_position: _,
66            num_records: _,
67            reinsert_count,
68            initial_insertion_epoch_us: _,
69        } = self
70        {
71            *reinsert_count = reinsert_count.saturating_add(1);
72        }
73    }
74}
chroma_types/log.rs

chroma_types/
log.rs
