gc.rs 91.3%

126 of 138 lines run, 6 of 12 functions · run #268 on main · run missed

//! Garbage collection (spec §21): delete objects
//! nothing reachable refers to.
//!
//! Roots: the current view, the views of every operation newer than
//! `core.op-retention`, reference attestations, and the policy; receipts
//! live as long as the changes of their state. Operation
//! objects are always kept so the log stays walkable, but an operation older
//! than the retention window no longer keeps its states alive, so undoing
//! that far back may fail once its objects are gone.

use super::*;
use crate::store::{PackReport, PruneReport};
use std::collections::HashSet;

#[derive(Debug, Clone, Default, PartialEq, Eq)]
pub struct GcReport {
    pub packed: PackReport,
    pub pruned: PruneReport,
    pub reachable: usize,
    /// Packs folded into the one a full repack writes.
    pub merged_packs: usize,
}

impl Repo {
    /// Prune unreachable objects older than the grace period (`now`
    /// ignores it), then repack everything into one pack (spec §13.4).
    pub fn gc(&self, now: bool) -> Result<GcReport> {
        self.gc_with(now, &mut |_| {})
    }

    /// [`Repo::gc`], reporting each phase and the packing's progress
    /// (oxygen#66), for a spinner or a log.
    pub fn gc_with(&self, now: bool, progress: crate::store::Progress) -> Result<GcReport> {
        self.gc_keeping(now, &[], progress)
    }

    /// [`Repo::gc_with`], keeping `extra` blobs and their chunks as well:
    /// objects a host stored for itself and names from outside the
    /// repository's history, such as a forge's job artifacts
    /// (oxygenhub#126).
    pub fn gc_keeping(
        &self,
        now: bool,
        extra: &[ObjectRef],
        progress: crate::store::Progress,
    ) -> Result<GcReport> {
        progress("finding what is reachable");
        let mut keep = self.reachable()?;
        for r in extra {
            if !self.store.has(r) || !keep.insert(*r) {
                continue;
            }
            if let Ok(body) = self.store.get_typed(r, ObjectType::Blob) {
                if let Ok(blob) = Blob::decode(&body) {
                    keep.extend(blob.chunks);
                }
            }
        }
        let grace = if now {
            std::time::Duration::ZERO
        } else {
            std::time::Duration::from_secs(self.config.gc_grace_days * 86_400)
        };
        let pruned = self.store.prune_with(&keep, grace, progress)?;
        // Then everything left in one pack, compressed harder than a
        // commit's (spec §13.4), as `git gc` repacks.
        let (packed, merged_packs) = self
            .store
            .repack_all_with(crate::store::REPACK_ZSTD_LEVEL, progress)?;
        Ok(GcReport {
            packed,
            pruned,
            reachable: keep.len(),
            merged_packs,
        })
    }

    /// Every object that must survive garbage collection.
    pub fn reachable(&self) -> Result<HashSet<ObjectRef>> {
        let mut keep: HashSet<ObjectRef> = HashSet::new();
        let mut states: Vec<ObjectRef> = Vec::new();
        let mut changes: Vec<ChangeHash> = Vec::new();
        let mut attests: Vec<ObjectRef> = Vec::new();
        let mut policies: Vec<ObjectRef> = Vec::new();

        // Operations: always kept; views within retention are roots.
        let cutoff = Hlc::now()
            .wall_ms
            .saturating_sub(self.config.op_retention_days * 86_400_000);
        for (i, (r, op)) in self.op_log()?.into_iter().enumerate() {
            keep.insert(r);
            if i == 0 || op.hlc.wall_ms >= cutoff {
                states.extend(op.view.bookmarks.values().copied());
                states.extend(op.view.remotes.values().flat_map(|b| b.values().copied()));
                changes.extend(op.view.workspaces.values().copied());
                policies.extend(op.view.policy);
            }
        }
        // Checkpoints: the object, its payload blob, and the blob's chunks.
        for r in self.checkpoint_refs()? {
            if let Ok(bytes) = self.store.get_typed(&r, ObjectType::Checkpoint) {
                keep.insert(r);
                let cp = crate::objects::Checkpoint::decode(&bytes)?;
                keep.insert(cp.payload);
                if let Ok(b) = self.store.get_typed(&cp.payload, ObjectType::Blob) {
                    keep.extend(Blob::decode(&b)?.chunks);
                }
                // A version 2 payload's graph blobs and their chunks.
                for g in self.checkpoint_graph_blobs(&cp).unwrap_or_default() {
                    keep.insert(g);
                    if let Ok(b) = self.store.get_typed(&g, ObjectType::Blob) {
                        keep.extend(Blob::decode(&b)?.chunks);
                    }
                }
            }
        }
        // Tags: their states and attestations.
        let view_now = self.view()?;
        states.extend(view_now.tags.values().copied());
        for name in view_now.tags.keys() {
            if let Some((r, _)) = self.tag_attest(name)? {
                attests.push(r);
            }
        }
        // Reference attestations.
        for name in self.view()?.bookmarks.keys() {
            if let Some((r, _)) = self.latest_ref_attest(name)? {
                attests.push(r);
            }
        }
        // A historical view may name objects pruned under an earlier, shorter
        // retention; there is nothing to keep for those.
        for st in states {
            if keep.insert(st) {
                match self.store.get_state(&st) {
                    Ok(s) => changes.extend(s.heads),
                    Err(StoreError::NotFound(_)) => continue,
                    Err(e) => return Err(e.into()),
                }
            }
        }
        for p in policies {
            if keep.insert(p) {
                let bytes = match self.store.get_typed(&p, ObjectType::Policy) {
                    Ok(b) => b,
                    Err(StoreError::NotFound(_)) => continue,
                    Err(e) => return Err(e.into()),
                };
                let policy = crate::policy::Policy::decode(&bytes)?;
                for id in &policy.identities {
                    keep.insert(id.hash());
                }
                for sig in self.store.signatures(&p)? {
                    keep.insert(sig.key);
                }
            }
        }
        // Changes: their closure, identities, blobs, chunks, seals.
        let mut stack = changes;
        while let Some(h) = stack.pop() {
            if !keep.insert(h) {
                continue;
            }
            // Beyond the shallow boundary, or pruned under an earlier
            // retention, there is nothing to keep.
            let c = match self.store.get_change_parts(&h) {
                Ok((c, texts)) => {
                    keep.extend(texts);
                    c
                }
                Err(StoreError::NotFound(_)) => continue,
                Err(e) => return Err(e.into()),
            };
            stack.extend(c.deps.iter().copied());
            if !c.author.is_zero() {
                keep.insert(c.author);
            }
            for sig in self.store.signatures(&h)? {
                keep.insert(sig.key);
            }
            for e in &c.edits {
                if let Edit::Replace { old, new, .. } = e {
                    for b in [old, new].into_iter().flatten() {
                        if keep.insert(*b) {
                            if let Ok(bytes) = self.store.get_typed(b, ObjectType::Blob) {
                                keep.extend(Blob::decode(&bytes)?.chunks);
                            }
                        }
                    }
                }
            }
            attests.extend(self.store.seals_for(&h)?);
        }
        self.reachable_receipts(&mut keep)?;
        for a in attests {
            if keep.insert(a) {
                for sig in self.store.signatures(&a)? {
                    keep.insert(sig.key);
                }
            }
        }
        Ok(keep)
    }
}

`/`	Search
`g` `c`	Code
`g` `h`	History
`g` `b`	Bookmarks
`g` `t`	Trust
`y`	Permalink to this state (files and trees)
`?`	This sheet

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	12×	pub fn gc(&self, now: bool) -> Result<GcReport> {
28	66×	self.gc_with(now, &mut \|_\| {})
29	12×	}
30
31		/// [`Repo::gc`], reporting each phase and the packing's progress
32		/// (oxygen#66), for a spinner or a log.
33	13×	pub fn gc_with(&self, now: bool, progress: crate::store::Progress) -> Result<GcReport> {
34	13×	self.gc_keeping(now, &[], progress)
35	13×	}
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	14×	pub fn gc_keeping(
42	14×	&self,
43	14×	now: bool,
44	14×	extra: &[ObjectRef],
45	14×	progress: crate::store::Progress,
46	14×	) -> Result<GcReport> {
47	14×	progress("finding what is reachable");
48	14×	let mut keep = self.reachable()?;
49	14×	for r in extra {
50	1×	if !self.store.has(r) \|\| !keep.insert(*r) {
51	0×	continue;
52	1×	}
53	1×	if let Ok(body) = self.store.get_typed(r, ObjectType::Blob) {
54	1×	if let Ok(blob) = Blob::decode(&body) {
55	1×	keep.extend(blob.chunks);
56	1×	}
57	0×	}
58		}
59	14×	let grace = if now {
60	12×	std::time::Duration::ZERO
61		} else {
62	2×	std::time::Duration::from_secs(self.config.gc_grace_days * 86_400)
63		};
64	14×	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	14×	let (packed, merged_packs) = self
68	14×	.store
69	14×	.repack_all_with(crate::store::REPACK_ZSTD_LEVEL, progress)?;
70	14×	Ok(GcReport {
71	14×	packed,
72	14×	pruned,
73	14×	reachable: keep.len(),
74	14×	merged_packs,
75	14×	})
76	14×	}
77
78		/// Every object that must survive garbage collection.
79	14×	pub fn reachable(&self) -> Result<HashSet<ObjectRef>> {
80	14×	let mut keep: HashSet<ObjectRef> = HashSet::new();
81	14×	let mut states: Vec<ObjectRef> = Vec::new();
82	14×	let mut changes: Vec<ChangeHash> = Vec::new();
83	14×	let mut attests: Vec<ObjectRef> = Vec::new();
84	14×	let mut policies: Vec<ObjectRef> = Vec::new();
85
86		// Operations: always kept; views within retention are roots.
87	14×	let cutoff = Hlc::now()
88	14×	.wall_ms
89	14×	.saturating_sub(self.config.op_retention_days * 86_400_000);
90	108×	for (i, (r, op)) in self.op_log()?.into_iter().enumerate() {
91	108×	keep.insert(r);
92	108×	if i == 0 \|\| op.hlc.wall_ms >= cutoff {
93	70×	states.extend(op.view.bookmarks.values().copied());
94	70×	states.extend(op.view.remotes.values().flat_map(\|b\| b.values().copied()));
95	70×	changes.extend(op.view.workspaces.values().copied());
96	70×	policies.extend(op.view.policy);
97	38×	}
98		}
99		// Checkpoints: the object, its payload blob, and the blob's chunks.
100	14×	for r in self.checkpoint_refs()? {
101	2×	if let Ok(bytes) = self.store.get_typed(&r, ObjectType::Checkpoint) {
102	2×	keep.insert(r);
103	2×	let cp = crate::objects::Checkpoint::decode(&bytes)?;
104	2×	keep.insert(cp.payload);
105	2×	if let Ok(b) = self.store.get_typed(&cp.payload, ObjectType::Blob) {
106	2×	keep.extend(Blob::decode(&b)?.chunks);
107	0×	}
108		// A version 2 payload's graph blobs and their chunks.
109	2×	for g in self.checkpoint_graph_blobs(&cp).unwrap_or_default() {
110	1×	keep.insert(g);
111	1×	if let Ok(b) = self.store.get_typed(&g, ObjectType::Blob) {
112	1×	keep.extend(Blob::decode(&b)?.chunks);
113	0×	}
114		}
115	0×	}
116		}
117		// Tags: their states and attestations.
118	14×	let view_now = self.view()?;
119	14×	states.extend(view_now.tags.values().copied());
120	14×	for name in view_now.tags.keys() {
121	3×	if let Some((r, _)) = self.tag_attest(name)? {
122	3×	attests.push(r);
123	3×	}
124		}
125		// Reference attestations.
126	14×	for name in self.view()?.bookmarks.keys() {
127	14×	if let Some((r, _)) = self.latest_ref_attest(name)? {
128	0×	attests.push(r);
129	14×	}
130		}
131		// A historical view may name objects pruned under an earlier, shorter
132		// retention; there is nothing to keep for those.
133	89×	for st in states {
134	89×	if keep.insert(st) {
135	43×	match self.store.get_state(&st) {
136	40×	Ok(s) => changes.extend(s.heads),
137	3×	Err(StoreError::NotFound(_)) => continue,
138	0×	Err(e) => return Err(e.into()),
139		}
140	46×	}
141		}
142	14×	for p in policies {
143	4×	if keep.insert(p) {
144	1×	let bytes = match self.store.get_typed(&p, ObjectType::Policy) {
145	1×	Ok(b) => b,
146	0×	Err(StoreError::NotFound(_)) => continue,
147	0×	Err(e) => return Err(e.into()),
148		};
149	1×	let policy = crate::policy::Policy::decode(&bytes)?;
150	1×	for id in &policy.identities {
151	1×	keep.insert(id.hash());
152	1×	}
153	1×	for sig in self.store.signatures(&p)? {
154	1×	keep.insert(sig.key);
155	1×	}
156	3×	}
157		}
158		// Changes: their closure, identities, blobs, chunks, seals.
159	14×	let mut stack = changes;
160	228×	while let Some(h) = stack.pop() {
161	214×	if !keep.insert(h) {
162	109×	continue;
163	105×	}
164		// Beyond the shallow boundary, or pruned under an earlier
165		// retention, there is nothing to keep.
166	105×	let c = match self.store.get_change_parts(&h) {
167	95×	Ok((c, texts)) => {
168	95×	keep.extend(texts);
169	95×	c
170		}
171	10×	Err(StoreError::NotFound(_)) => continue,
172	0×	Err(e) => return Err(e.into()),
173		};
174	95×	stack.extend(c.deps.iter().copied());
175	95×	if !c.author.is_zero() {
176	30×	keep.insert(c.author);
177	65×	}
178	95×	for sig in self.store.signatures(&h)? {
179	13×	keep.insert(sig.key);
180	13×	}
181	108×	for e in &c.edits {
182	108×	if let Edit::Replace { old, new, .. } = e {
183	17×	for b in [old, new].into_iter().flatten() {
184	17×	if keep.insert(*b) {
185	17×	if let Ok(bytes) = self.store.get_typed(b, ObjectType::Blob) {
186	17×	keep.extend(Blob::decode(&bytes)?.chunks);
187	0×	}
188	0×	}
189		}
190	91×	}
191		}
192	95×	attests.extend(self.store.seals_for(&h)?);
193		}
194	14×	self.reachable_receipts(&mut keep)?;
195	14×	for a in attests {
196	5×	if keep.insert(a) {
197	4×	for sig in self.store.signatures(&a)? {
198	4×	keep.insert(sig.key);
199	4×	}
200	1×	}
201		}
202	14×	Ok(keep)
203	14×	}
204		}

Coverage / crates/oxy-core/src/repo/gc.rs 91.3%

Keyboard