OXYGENHUB
# 028545dfa7a4 oxygen / crates / oxy-core / src / repo / gc.rs
8.6 KiB · Rust · last changed The issuer countersigns when it accepted a certified update (keyless step 4c) 19 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`, 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 
        self.gc_with(now, &mut |_| {})
29 
    }
30
31 
    /// [`Repo::gc`], reporting each phase and the packing's progress
32 
    /// (oxygen#66), for a spinner or a log.
33 
    pub fn gc_with(&self, now: bool, progress: crate::store::Progress) -> Result<GcReport> {
34 
        self.gc_keeping(now, &[], progress)
35 
    }
36
37 
    /// [`Repo::gc_with`], keeping `extra` blobs and their chunks as well:
38 
    /// objects a host stored for itself and names from outside the
39 
    /// repository's history, such as a forge's job artifacts
40 
    /// (oxygenhub#126).
41 
    pub fn gc_keeping(
42 
        &self,
43 
        now: bool,
44 
        extra: &[ObjectRef],
45 
        progress: crate::store::Progress,
46 
    ) -> Result<GcReport> {
47 
        progress("finding what is reachable");
48 
        let mut keep = self.reachable()?;
49 
        for r in extra {
50 
            if !self.store.has(r) || !keep.insert(*r) {
51 
                continue;
52 
            }
53 
            if let Ok(body) = self.store.get_typed(r, ObjectType::Blob) {
54 
                if let Ok(blob) = Blob::decode(&body) {
55 
                    keep.extend(blob.chunks);
56 
                }
57 
            }
58 
        }
59 
        let grace = if now {
60 
            std::time::Duration::ZERO
61 
        } else {
62 
            std::time::Duration::from_secs(self.config.gc_grace_days * 86_400)
63 
        };
64 
        let pruned = self.store.prune_with(&keep, grace, progress)?;
65 
        // Then everything left in one pack, compressed harder than a
66 
        // commit's (spec §13.4), as `git gc` repacks.
67 
        let (packed, merged_packs) = self
68 
            .store
69 
            .repack_all_with(crate::store::REPACK_ZSTD_LEVEL, progress)?;
70 
        Ok(GcReport {
71 
            packed,
72 
            pruned,
73 
            reachable: keep.len(),
74 
            merged_packs,
75 
        })
76 
    }
77
78 
    /// Every object that must survive garbage collection.
79 
    pub fn reachable(&self) -> Result<HashSet<ObjectRef>> {
80 
        let mut keep: HashSet<ObjectRef> = HashSet::new();
81 
        let mut states: Vec<ObjectRef> = Vec::new();
82 
        let mut changes: Vec<ChangeHash> = Vec::new();
83 
        let mut attests: Vec<ObjectRef> = Vec::new();
84 
        let mut policies: Vec<ObjectRef> = Vec::new();
85
86 
        // Operations: always kept; views within retention are roots.
87 
        let cutoff = Hlc::now()
88 
            .wall_ms
89 
            .saturating_sub(self.config.op_retention_days * 86_400_000);
90 
        for (i, (r, op)) in self.op_log()?.into_iter().enumerate() {
91 
            keep.insert(r);
92 
            if i == 0 || op.hlc.wall_ms >= cutoff {
93 
                states.extend(op.view.bookmarks.values().copied());
94 
                states.extend(op.view.remotes.values().flat_map(|b| b.values().copied()));
95 
                changes.extend(op.view.workspaces.values().copied());
96 
                policies.extend(op.view.policy);
97 
            }
98 
        }
99 
        // Checkpoints: the object, its payload blob, and the blob's chunks.
100 
        for r in self.checkpoint_refs()? {
101 
            if let Ok(bytes) = self.store.get_typed(&r, ObjectType::Checkpoint) {
102 
                keep.insert(r);
103 
                let cp = crate::objects::Checkpoint::decode(&bytes)?;
104 
                keep.insert(cp.payload);
105 
                if let Ok(b) = self.store.get_typed(&cp.payload, ObjectType::Blob) {
106 
                    keep.extend(Blob::decode(&b)?.chunks);
107 
                }
108 
                // A version 2 payload's graph blobs and their chunks.
109 
                for g in self.checkpoint_graph_blobs(&cp).unwrap_or_default() {
110 
                    keep.insert(g);
111 
                    if let Ok(b) = self.store.get_typed(&g, ObjectType::Blob) {
112 
                        keep.extend(Blob::decode(&b)?.chunks);
113 
                    }
114 
                }
115 
            }
116 
        }
117 
        // Tags: their states and attestations.
118 
        let view_now = self.view()?;
119 
        states.extend(view_now.tags.values().copied());
120 
        for name in view_now.tags.keys() {
121 
            if let Some((r, _)) = self.tag_attest(name)? {
122 
                attests.push(r);
123 
            }
124 
        }
125 
        // Reference attestations.
126 
        for name in self.view()?.bookmarks.keys() {
127 
            if let Some((r, _)) = self.latest_ref_attest(name)? {
128 
                // With a host's acceptance of it (spec §14.7).
129 
                if let Some(acc) = self.acceptance_of(&r)? {
130 
                    attests.push(acc.attest.hash());
131 
                }
132 
                attests.push(r);
133 
            }
134 
        }
135 
        // A historical view may name objects pruned under an earlier, shorter
136 
        // retention; there is nothing to keep for those.
137 
        for st in states {
138 
            if keep.insert(st) {
139 
                match self.store.get_state(&st) {
140 
                    Ok(s) => changes.extend(s.heads),
141 
                    Err(StoreError::NotFound(_)) => continue,
142 
                    Err(e) => return Err(e.into()),
143 
                }
144 
            }
145 
        }
146 
        for p in policies {
147 
            if keep.insert(p) {
148 
                let bytes = match self.store.get_typed(&p, ObjectType::Policy) {
149 
                    Ok(b) => b,
150 
                    Err(StoreError::NotFound(_)) => continue,
151 
                    Err(e) => return Err(e.into()),
152 
                };
153 
                let policy = crate::policy::Policy::decode(&bytes)?;
154 
                for id in &policy.identities {
155 
                    keep.insert(id.hash());
156 
                }
157 
                for sig in self.store.signatures(&p)? {
158 
                    keep.insert(sig.key);
159 
                }
160 
            }
161 
        }
162 
        // Changes: their closure, identities, blobs, chunks, seals.
163 
        let mut stack = changes;
164 
        while let Some(h) = stack.pop() {
165 
            if !keep.insert(h) {
166 
                continue;
167 
            }
168 
            // Beyond the shallow boundary, or pruned under an earlier
169 
            // retention, there is nothing to keep.
170 
            let c = match self.store.get_change_parts(&h) {
171 
                Ok((c, texts)) => {
172 
                    keep.extend(texts);
173 
                    c
174 
                }
175 
                Err(StoreError::NotFound(_)) => continue,
176 
                Err(e) => return Err(e.into()),
177 
            };
178 
            stack.extend(c.deps.iter().copied());
179 
            if !c.author.is_zero() {
180 
                keep.insert(c.author);
181 
            }
182 
            for sig in self.store.signatures(&h)? {
183 
                self.keep_signer(&mut keep, sig.key)?;
184 
            }
185 
            for e in &c.edits {
186 
                if let Edit::Replace { old, new, .. } = e {
187 
                    for b in [old, new].into_iter().flatten() {
188 
                        if keep.insert(*b) {
189 
                            if let Ok(bytes) = self.store.get_typed(b, ObjectType::Blob) {
190 
                                keep.extend(Blob::decode(&bytes)?.chunks);
191 
                            }
192 
                        }
193 
                    }
194 
                }
195 
            }
196 
            attests.extend(self.store.seals_for(&h)?);
197 
        }
198 
        self.reachable_receipts(&mut keep)?;
199 
        for a in attests {
200 
            if keep.insert(a) {
201 
                for sig in self.store.signatures(&a)? {
202 
                    self.keep_signer(&mut keep, sig.key)?;
203 
                }
204 
            }
205 
        }
206 
        Ok(keep)
207 
    }
208
209 
    /// Keep a signer's identity and, for a short-lived key, its
210 
    /// certificate and the certificate's issuer (spec §14.7).
211 
    pub(crate) fn keep_signer(&self, keep: &mut HashSet<ObjectRef>, key: ObjectRef) -> Result<()> {
212 
        keep.insert(key);
213 
        if let Some(c) = self.store.certificate_of(&key)? {
214 
            keep.insert(c);
215 
            if let Ok(cert) = self.store.get_certificate(&c) {
216 
                keep.insert(cert.cert.issuer);
217 
            }
218 
        }
219 
        Ok(())
220 
    }
221 
}