OXYGENHUB
# 9aa399044d9d oxygen / crates / oxy-core / src / repo / gc.rs
6.7 KiB · Rust · last changed gc repacks everything into one pack, compressed harder 1 month ago source blame history raw
1 
//! Garbage collection (spec §21): delete objects
2 
//! nothing reachable refers to.
3 
//!
4 
//! Roots: the current view, the views of every operation newer than
5 
//! `core.op-retention`, reference attestations, and the policy; receipts
6 
//! live as long as the changes of their state. Operation
7 
//! objects are always kept so the log stays walkable, but an operation older
8 
//! than the retention window no longer keeps its states alive, so undoing
9 
//! that far back may fail once its objects are gone.
10
11 
use super::*;
12 
use crate::store::{PackReport, PruneReport};
13 
use std::collections::HashSet;
14
15 
#[derive(Debug, Clone, Default, PartialEq, Eq)]
16 
pub struct GcReport {
17 
    pub packed: PackReport,
18 
    pub pruned: PruneReport,
19 
    pub reachable: usize,
20 
    /// Packs folded into the one a full repack writes.
21 
    pub merged_packs: usize,
22 
}
23
24 
impl Repo {
25 
    /// Prune unreachable objects older than the grace period (`now`
26 
    /// ignores it), then repack everything into one pack (spec §13.4).
27 
    pub fn gc(&self, now: bool) -> Result<GcReport> {
28 
        let keep = self.reachable()?;
29 
        let grace = if now {
30 
            std::time::Duration::ZERO
31 
        } else {
32 
            std::time::Duration::from_secs(self.config.gc_grace_days * 86_400)
33 
        };
34 
        let pruned = self.store.prune(&keep, grace)?;
35 
        // Then everything left in one pack, compressed harder than a
36 
        // commit's (spec §13.4), as `git gc` repacks.
37 
        let (packed, merged_packs) = self.store.repack_all(crate::store::REPACK_ZSTD_LEVEL)?;
38 
        Ok(GcReport {
39 
            packed,
40 
            pruned,
41 
            reachable: keep.len(),
42 
            merged_packs,
43 
        })
44 
    }
45
46 
    /// Every object that must survive garbage collection.
47 
    pub fn reachable(&self) -> Result<HashSet<ObjectRef>> {
48 
        let mut keep: HashSet<ObjectRef> = HashSet::new();
49 
        let mut states: Vec<ObjectRef> = Vec::new();
50 
        let mut changes: Vec<ChangeHash> = Vec::new();
51 
        let mut attests: Vec<ObjectRef> = Vec::new();
52 
        let mut policies: Vec<ObjectRef> = Vec::new();
53
54 
        // Operations: always kept; views within retention are roots.
55 
        let cutoff = Hlc::now()
56 
            .wall_ms
57 
            .saturating_sub(self.config.op_retention_days * 86_400_000);
58 
        for (i, (r, op)) in self.op_log()?.into_iter().enumerate() {
59 
            keep.insert(r);
60 
            if i == 0 || op.hlc.wall_ms >= cutoff {
61 
                states.extend(op.view.bookmarks.values().copied());
62 
                states.extend(op.view.remotes.values().flat_map(|b| b.values().copied()));
63 
                changes.extend(op.view.workspaces.values().copied());
64 
                policies.extend(op.view.policy);
65 
            }
66 
        }
67 
        // Checkpoints: the object, its payload blob, and the blob's chunks.
68 
        for r in self.checkpoint_refs()? {
69 
            if let Ok(bytes) = self.store.get_typed(&r, ObjectType::Checkpoint) {
70 
                keep.insert(r);
71 
                let cp = crate::objects::Checkpoint::decode(&bytes)?;
72 
                keep.insert(cp.payload);
73 
                if let Ok(b) = self.store.get_typed(&cp.payload, ObjectType::Blob) {
74 
                    keep.extend(Blob::decode(&b)?.chunks);
75 
                }
76 
                // A version 2 payload's graph blobs and their chunks.
77 
                for g in self.checkpoint_graph_blobs(&cp).unwrap_or_default() {
78 
                    keep.insert(g);
79 
                    if let Ok(b) = self.store.get_typed(&g, ObjectType::Blob) {
80 
                        keep.extend(Blob::decode(&b)?.chunks);
81 
                    }
82 
                }
83 
            }
84 
        }
85 
        // Tags: their states and attestations.
86 
        let view_now = self.view()?;
87 
        states.extend(view_now.tags.values().copied());
88 
        for name in view_now.tags.keys() {
89 
            if let Some((r, _)) = self.tag_attest(name)? {
90 
                attests.push(r);
91 
            }
92 
        }
93 
        // Reference attestations.
94 
        for name in self.view()?.bookmarks.keys() {
95 
            if let Some((r, _)) = self.latest_ref_attest(name)? {
96 
                attests.push(r);
97 
            }
98 
        }
99 
        // A historical view may name objects pruned under an earlier, shorter
100 
        // retention; there is nothing to keep for those.
101 
        for st in states {
102 
            if keep.insert(st) {
103 
                match self.store.get_state(&st) {
104 
                    Ok(s) => changes.extend(s.heads),
105 
                    Err(StoreError::NotFound(_)) => continue,
106 
                    Err(e) => return Err(e.into()),
107 
                }
108 
            }
109 
        }
110 
        for p in policies {
111 
            if keep.insert(p) {
112 
                let bytes = match self.store.get_typed(&p, ObjectType::Policy) {
113 
                    Ok(b) => b,
114 
                    Err(StoreError::NotFound(_)) => continue,
115 
                    Err(e) => return Err(e.into()),
116 
                };
117 
                let policy = crate::policy::Policy::decode(&bytes)?;
118 
                for id in &policy.identities {
119 
                    keep.insert(id.hash());
120 
                }
121 
                for sig in self.store.signatures(&p)? {
122 
                    keep.insert(sig.key);
123 
                }
124 
            }
125 
        }
126 
        // Changes: their closure, identities, blobs, chunks, seals.
127 
        let mut stack = changes;
128 
        while let Some(h) = stack.pop() {
129 
            if !keep.insert(h) {
130 
                continue;
131 
            }
132 
            // Beyond the shallow boundary, or pruned under an earlier
133 
            // retention, there is nothing to keep.
134 
            let c = match self.store.get_change_parts(&h) {
135 
                Ok((c, texts)) => {
136 
                    keep.extend(texts);
137 
                    c
138 
                }
139 
                Err(StoreError::NotFound(_)) => continue,
140 
                Err(e) => return Err(e.into()),
141 
            };
142 
            stack.extend(c.deps.iter().copied());
143 
            if !c.author.is_zero() {
144 
                keep.insert(c.author);
145 
            }
146 
            for sig in self.store.signatures(&h)? {
147 
                keep.insert(sig.key);
148 
            }
149 
            for e in &c.edits {
150 
                if let Edit::Replace { old, new, .. } = e {
151 
                    for b in [old, new].into_iter().flatten() {
152 
                        if keep.insert(*b) {
153 
                            if let Ok(bytes) = self.store.get_typed(b, ObjectType::Blob) {
154 
                                keep.extend(Blob::decode(&bytes)?.chunks);
155 
                            }
156 
                        }
157 
                    }
158 
                }
159 
            }
160 
            attests.extend(self.store.seals_for(&h)?);
161 
        }
162 
        self.reachable_receipts(&mut keep)?;
163 
        for a in attests {
164 
            if keep.insert(a) {
165 
                for sig in self.store.signatures(&a)? {
166 
                    keep.insert(sig.key);
167 
                }
168 
            }
169 
        }
170 
        Ok(keep)
171 
    }
172 
}