OXYGENHUB
◆ switch-says oxygen / crates / oxy-core / src / repo / gc.rs
8.4 KiB · Rust · last changed Snapshots: large new files left out and named; past working copies expire 24 days 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` (their working copies only within the shorter
6 
//! `core.snapshot-retention`), reference attestations, and the policy; receipts
7 
//! live as long as the changes of their state. Operation
8 
//! objects are always kept so the log stays walkable, but an operation older
9 
//! than the retention window no longer keeps its states alive, so undoing
10 
//! that far back may fail once its objects are gone.
11
12 
use super::*;
13 
use crate::store::{PackReport, PruneReport};
14 
use std::collections::HashSet;
15
16 
#[derive(Debug, Clone, Default, PartialEq, Eq)]
17 
pub struct GcReport {
18 
    pub packed: PackReport,
19 
    pub pruned: PruneReport,
20 
    pub reachable: usize,
21 
    /// Packs folded into the one a full repack writes.
22 
    pub merged_packs: usize,
23 
}
24
25 
impl Repo {
26 
    /// Prune unreachable objects older than the grace period (`now`
27 
    /// ignores it), then repack everything into one pack (spec §13.4).
28 
    pub fn gc(&self, now: bool) -> Result<GcReport> {
29 
        self.gc_with(now, &mut |_| {})
30 
    }
31
32 
    /// [`Repo::gc`], reporting each phase and the packing's progress
33 
    /// (oxygen#66), for a spinner or a log.
34 
    pub fn gc_with(&self, now: bool, progress: crate::store::Progress) -> Result<GcReport> {
35 
        self.gc_keeping(now, &[], progress)
36 
    }
37
38 
    /// [`Repo::gc_with`], keeping `extra` blobs and their chunks as well:
39 
    /// objects a host stored for itself and names from outside the
40 
    /// repository's history, such as a forge's job artifacts
41 
    /// (oxygenhub#126).
42 
    pub fn gc_keeping(
43 
        &self,
44 
        now: bool,
45 
        extra: &[ObjectRef],
46 
        progress: crate::store::Progress,
47 
    ) -> Result<GcReport> {
48 
        progress("finding what is reachable");
49 
        let mut keep = self.reachable()?;
50 
        for r in extra {
51 
            if !self.store.has(r) || !keep.insert(*r) {
52 
                continue;
53 
            }
54 
            if let Ok(body) = self.store.get_typed(r, ObjectType::Blob) {
55 
                if let Ok(blob) = Blob::decode(&body) {
56 
                    keep.extend(blob.chunks);
57 
                }
58 
            }
59 
        }
60 
        let grace = if now {
61 
            std::time::Duration::ZERO
62 
        } else {
63 
            std::time::Duration::from_secs(self.config.gc_grace_days * 86_400)
64 
        };
65 
        let pruned = self.store.prune_with(&keep, grace, progress)?;
66 
        // Then everything left in one pack, compressed harder than a
67 
        // commit's (spec §13.4), as `git gc` repacks.
68 
        let (packed, merged_packs) = self
69 
            .store
70 
            .repack_all_with(crate::store::REPACK_ZSTD_LEVEL, progress)?;
71 
        Ok(GcReport {
72 
            packed,
73 
            pruned,
74 
            reachable: keep.len(),
75 
            merged_packs,
76 
        })
77 
    }
78
79 
    /// Every object that must survive garbage collection.
80 
    pub fn reachable(&self) -> Result<HashSet<ObjectRef>> {
81 
        let mut keep: HashSet<ObjectRef> = HashSet::new();
82 
        let mut states: Vec<ObjectRef> = Vec::new();
83 
        let mut changes: Vec<ChangeHash> = Vec::new();
84 
        let mut attests: Vec<ObjectRef> = Vec::new();
85 
        let mut policies: Vec<ObjectRef> = Vec::new();
86
87 
        // Operations: always kept; views within retention are roots. Past
88 
        // working copies only within the shorter `core.snapshot-retention`
89 
        // (oxygen#98): a file only ever snapshotted, such as a build output
90 
        // a status took in, goes once they expire.
91 
        let now_ms = Hlc::now().wall_ms;
92 
        let cutoff = now_ms.saturating_sub(self.config.op_retention_days * 86_400_000);
93 
        let wc_cutoff = now_ms
94 
            .saturating_sub(self.config.snapshot_retention_days * 86_400_000)
95 
            .max(cutoff);
96 
        for (i, (r, op)) in self.op_log()?.into_iter().enumerate() {
97 
            keep.insert(r);
98 
            if i == 0 || op.hlc.wall_ms >= cutoff {
99 
                states.extend(op.view.bookmarks.values().copied());
100 
                states.extend(op.view.remotes.values().flat_map(|b| b.values().copied()));
101 
                policies.extend(op.view.policy);
102 
            }
103 
            if i == 0 || (self.config.snapshot_retention_days > 0 && op.hlc.wall_ms >= wc_cutoff) {
104 
                changes.extend(op.view.workspaces.values().copied());
105 
            }
106 
        }
107 
        // Checkpoints: the object, its payload blob, and the blob's chunks.
108 
        for r in self.checkpoint_refs()? {
109 
            if let Ok(bytes) = self.store.get_typed(&r, ObjectType::Checkpoint) {
110 
                keep.insert(r);
111 
                let cp = crate::objects::Checkpoint::decode(&bytes)?;
112 
                keep.insert(cp.payload);
113 
                if let Ok(b) = self.store.get_typed(&cp.payload, ObjectType::Blob) {
114 
                    keep.extend(Blob::decode(&b)?.chunks);
115 
                }
116 
                // A version 2 payload's graph blobs and their chunks.
117 
                for g in self.checkpoint_graph_blobs(&cp).unwrap_or_default() {
118 
                    keep.insert(g);
119 
                    if let Ok(b) = self.store.get_typed(&g, ObjectType::Blob) {
120 
                        keep.extend(Blob::decode(&b)?.chunks);
121 
                    }
122 
                }
123 
            }
124 
        }
125 
        // Tags: their states and attestations.
126 
        let view_now = self.view()?;
127 
        states.extend(view_now.tags.values().copied());
128 
        for name in view_now.tags.keys() {
129 
            if let Some((r, _)) = self.tag_attest(name)? {
130 
                attests.push(r);
131 
            }
132 
        }
133 
        // Reference attestations.
134 
        for name in self.view()?.bookmarks.keys() {
135 
            if let Some((r, _)) = self.latest_ref_attest(name)? {
136 
                attests.push(r);
137 
            }
138 
        }
139 
        // A historical view may name objects pruned under an earlier, shorter
140 
        // retention; there is nothing to keep for those.
141 
        for st in states {
142 
            if keep.insert(st) {
143 
                match self.store.get_state(&st) {
144 
                    Ok(s) => changes.extend(s.heads),
145 
                    Err(StoreError::NotFound(_)) => continue,
146 
                    Err(e) => return Err(e.into()),
147 
                }
148 
            }
149 
        }
150 
        for p in policies {
151 
            if keep.insert(p) {
152 
                let bytes = match self.store.get_typed(&p, ObjectType::Policy) {
153 
                    Ok(b) => b,
154 
                    Err(StoreError::NotFound(_)) => continue,
155 
                    Err(e) => return Err(e.into()),
156 
                };
157 
                let policy = crate::policy::Policy::decode(&bytes)?;
158 
                for id in &policy.identities {
159 
                    keep.insert(id.hash());
160 
                }
161 
                for sig in self.store.signatures(&p)? {
162 
                    keep.insert(sig.key);
163 
                }
164 
            }
165 
        }
166 
        // Changes: their closure, identities, blobs, chunks, seals.
167 
        let mut stack = changes;
168 
        while let Some(h) = stack.pop() {
169 
            if !keep.insert(h) {
170 
                continue;
171 
            }
172 
            // Beyond the shallow boundary, or pruned under an earlier
173 
            // retention, there is nothing to keep.
174 
            let c = match self.store.get_change_parts(&h) {
175 
                Ok((c, texts)) => {
176 
                    keep.extend(texts);
177 
                    c
178 
                }
179 
                Err(StoreError::NotFound(_)) => continue,
180 
                Err(e) => return Err(e.into()),
181 
            };
182 
            stack.extend(c.deps.iter().copied());
183 
            if !c.author.is_zero() {
184 
                keep.insert(c.author);
185 
            }
186 
            for sig in self.store.signatures(&h)? {
187 
                keep.insert(sig.key);
188 
            }
189 
            for e in &c.edits {
190 
                if let Edit::Replace { old, new, .. } = e {
191 
                    for b in [old, new].into_iter().flatten() {
192 
                        if keep.insert(*b) {
193 
                            if let Ok(bytes) = self.store.get_typed(b, ObjectType::Blob) {
194 
                                keep.extend(Blob::decode(&bytes)?.chunks);
195 
                            }
196 
                        }
197 
                    }
198 
                }
199 
            }
200 
            attests.extend(self.store.seals_for(&h)?);
201 
        }
202 
        self.reachable_receipts(&mut keep)?;
203 
        for a in attests {
204 
            if keep.insert(a) {
205 
                for sig in self.store.signatures(&a)? {
206 
                    keep.insert(sig.key);
207 
                }
208 
            }
209 
        }
210 
        Ok(keep)
211 
    }
212 
}