OXYGENHUB
# 519c7c061bcb oxygen / crates / oxy-core / src / repo / gc.rs
6.6 KiB · Rust · last changed Store large inserts as text blobs (change format 2) 1 month ago source blame history raw
1 
//! Garbage collection (spec §21): pack loose objects, then 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 one because there were more than `MAX_PACKS`.
21 
    pub merged_packs: usize,
22 
}
23
24 
impl Repo {
25 
    /// Pack loose objects and prune unreachable ones older than the grace
26 
    /// period (`now` ignores the grace period).
27 
    pub fn gc(&self, now: bool) -> Result<GcReport> {
28 
        let packed = self.store.pack_loose()?;
29 
        let keep = self.reachable()?;
30 
        let grace = if now {
31 
            std::time::Duration::ZERO
32 
        } else {
33 
            std::time::Duration::from_secs(self.config.gc_grace_days * 86_400)
34 
        };
35 
        let pruned = self.store.prune(&keep, grace)?;
36 
        let merged_packs = self.store.merge_packs()?;
37 
        Ok(GcReport {
38 
            packed,
39 
            pruned,
40 
            reachable: keep.len(),
41 
            merged_packs,
42 
        })
43 
    }
44
45 
    /// Every object that must survive garbage collection.
46 
    pub fn reachable(&self) -> Result<HashSet<ObjectRef>> {
47 
        let mut keep: HashSet<ObjectRef> = HashSet::new();
48 
        let mut states: Vec<ObjectRef> = Vec::new();
49 
        let mut changes: Vec<ChangeHash> = Vec::new();
50 
        let mut attests: Vec<ObjectRef> = Vec::new();
51 
        let mut policies: Vec<ObjectRef> = Vec::new();
52
53 
        // Operations: always kept; views within retention are roots.
54 
        let cutoff = Hlc::now()
55 
            .wall_ms
56 
            .saturating_sub(self.config.op_retention_days * 86_400_000);
57 
        for (i, (r, op)) in self.op_log()?.into_iter().enumerate() {
58 
            keep.insert(r);
59 
            if i == 0 || op.hlc.wall_ms >= cutoff {
60 
                states.extend(op.view.bookmarks.values().copied());
61 
                states.extend(op.view.remotes.values().flat_map(|b| b.values().copied()));
62 
                changes.extend(op.view.workspaces.values().copied());
63 
                policies.extend(op.view.policy);
64 
            }
65 
        }
66 
        // Checkpoints: the object, its payload blob, and the blob's chunks.
67 
        for r in self.checkpoint_refs()? {
68 
            if let Ok(bytes) = self.store.get_typed(&r, ObjectType::Checkpoint) {
69 
                keep.insert(r);
70 
                let cp = crate::objects::Checkpoint::decode(&bytes)?;
71 
                keep.insert(cp.payload);
72 
                if let Ok(b) = self.store.get_typed(&cp.payload, ObjectType::Blob) {
73 
                    keep.extend(Blob::decode(&b)?.chunks);
74 
                }
75 
                // A version 2 payload's graph blobs and their chunks.
76 
                for g in self.checkpoint_graph_blobs(&cp).unwrap_or_default() {
77 
                    keep.insert(g);
78 
                    if let Ok(b) = self.store.get_typed(&g, ObjectType::Blob) {
79 
                        keep.extend(Blob::decode(&b)?.chunks);
80 
                    }
81 
                }
82 
            }
83 
        }
84 
        // Tags: their states and attestations.
85 
        let view_now = self.view()?;
86 
        states.extend(view_now.tags.values().copied());
87 
        for name in view_now.tags.keys() {
88 
            if let Some((r, _)) = self.tag_attest(name)? {
89 
                attests.push(r);
90 
            }
91 
        }
92 
        // Reference attestations.
93 
        for name in self.view()?.bookmarks.keys() {
94 
            if let Some((r, _)) = self.latest_ref_attest(name)? {
95 
                attests.push(r);
96 
            }
97 
        }
98 
        // A historical view may name objects pruned under an earlier, shorter
99 
        // retention; there is nothing to keep for those.
100 
        for st in states {
101 
            if keep.insert(st) {
102 
                match self.store.get_state(&st) {
103 
                    Ok(s) => changes.extend(s.heads),
104 
                    Err(StoreError::NotFound(_)) => continue,
105 
                    Err(e) => return Err(e.into()),
106 
                }
107 
            }
108 
        }
109 
        for p in policies {
110 
            if keep.insert(p) {
111 
                let bytes = match self.store.get_typed(&p, ObjectType::Policy) {
112 
                    Ok(b) => b,
113 
                    Err(StoreError::NotFound(_)) => continue,
114 
                    Err(e) => return Err(e.into()),
115 
                };
116 
                let policy = crate::policy::Policy::decode(&bytes)?;
117 
                for id in &policy.identities {
118 
                    keep.insert(id.hash());
119 
                }
120 
                for sig in self.store.signatures(&p)? {
121 
                    keep.insert(sig.key);
122 
                }
123 
            }
124 
        }
125 
        // Changes: their closure, identities, blobs, chunks, seals.
126 
        let mut stack = changes;
127 
        while let Some(h) = stack.pop() {
128 
            if !keep.insert(h) {
129 
                continue;
130 
            }
131 
            // Beyond the shallow boundary, or pruned under an earlier
132 
            // retention, there is nothing to keep.
133 
            let c = match self.store.get_change_parts(&h) {
134 
                Ok((c, texts)) => {
135 
                    keep.extend(texts);
136 
                    c
137 
                }
138 
                Err(StoreError::NotFound(_)) => continue,
139 
                Err(e) => return Err(e.into()),
140 
            };
141 
            stack.extend(c.deps.iter().copied());
142 
            if !c.author.is_zero() {
143 
                keep.insert(c.author);
144 
            }
145 
            for sig in self.store.signatures(&h)? {
146 
                keep.insert(sig.key);
147 
            }
148 
            for e in &c.edits {
149 
                if let Edit::Replace { old, new, .. } = e {
150 
                    for b in [old, new].into_iter().flatten() {
151 
                        if keep.insert(*b) {
152 
                            if let Ok(bytes) = self.store.get_typed(b, ObjectType::Blob) {
153 
                                keep.extend(Blob::decode(&bytes)?.chunks);
154 
                            }
155 
                        }
156 
                    }
157 
                }
158 
            }
159 
            attests.extend(self.store.seals_for(&h)?);
160 
        }
161 
        self.reachable_receipts(&mut keep)?;
162 
        for a in attests {
163 
            if keep.insert(a) {
164 
                for sig in self.store.signatures(&a)? {
165 
                    keep.insert(sig.key);
166 
                }
167 
            }
168 
        }
169 
        Ok(keep)
170 
    }
171 
}