OXYGENHUB
# 62cdf34daec1 oxygenhub / crates / hub / src / maintenance.rs
4.3 KiB · Rust · last changed Reports retention: files beyond the newest 50 runs off the default bookmark expire 24 days ago source blame history raw
1 
//! Housekeeping that runs on its own (hub spec §17.2): the nightly
2 
//! garbage collection of every repository, one at a time under its
3 
//! write lock, so a push never races a repack. `oxygenhub gc` runs the
4 
//! same pass by hand.
5
6 
use crate::db;
7 
use crate::repos::RepoName;
8 
use crate::Shared;
9
10 
/// What one repository's gc did, for the log and `oxygenhub gc`.
11 
#[derive(Clone, Debug)]
12 
pub struct Done {
13 
    pub repo: String,
14 
    pub failed: bool,
15 
    pub summary: String,
16 
}
17
18 
impl std::fmt::Display for Done {
19 
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
20 
        write!(f, "{}: {}", self.repo, self.summary)
21 
    }
22 
}
23
24 
/// Garbage-collect every repository, in turn. A repository that fails is
25 
/// reported and the rest still run.
26 
pub async fn gc_all(app: &Shared) -> Vec<Done> {
27 
    // Report files past retention first (#127), so this run prunes them.
28 
    match crate::reports::expire(&app.db).await {
29 
        Ok(0) => {}
30 
        Ok(n) => tracing::info!("gc: {n} reports' files expired under retention"),
31 
        Err(e) => tracing::warn!("gc: expiring report files: {e}"),
32 
    }
33 
    let rows = match db::repos::list(&app.db).await {
34 
        Ok(rows) => rows,
35 
        Err(e) => {
36 
            tracing::warn!("gc: listing repositories: {e}");
37 
            return Vec::new();
38 
        }
39 
    };
40 
    let mut done = Vec::with_capacity(rows.len());
41 
    for row in rows {
42 
        let slug = format!("{}/{}", row.owner, row.name);
43 
        let Some(name) = RepoName::new(&row.owner, &row.name) else { continue };
44 
        if !app.repos.exists(&name) {
45 
            continue;
46 
        }
47 
        let label = slug.clone();
48 
        // Flow's artifacts and reports' files are named only in the
49 
        // database: kept as roots, or gc would prune them after the grace
50 
        // period (#126, #127). A listing that fails skips this repository
51 
        // rather than risk them.
52 
        let listed = match db::flow::artifact_blobs(&app.db, row.id).await {
53 
            Ok(mut blobs) => crate::reports::file_blobs(&app.db, row.id).await.map(|files| {
54 
                blobs.extend(files);
55 
                blobs
56 
            }),
57 
            Err(e) => Err(e),
58 
        };
59 
        let keep: Vec<oxy_core::hash::ObjectRef> = match listed {
60 
            Ok(blobs) => blobs.iter().filter_map(|b| oxy_core::hash::ObjectRef::from_hex(b)).collect(),
61 
            Err(e) => {
62 
                tracing::warn!("gc: {slug}: listing its artifacts: {e}; skipped");
63 
                continue;
64 
            }
65 
        };
66 
        let started = std::time::Instant::now();
67 
        let result = app
68 
            .repos
69 
            .write(&name, move |repo, _| Ok(repo.gc_keeping(false, &keep, &mut |phase| tracing::debug!("gc {label}: {phase}"))?))
70 
            .await;
71 
        let d = match result {
72 
            Ok(r) => Done {
73 
                summary: format!(
74 
                    "{} reachable; pruned {} ({} bytes){}; packed {} into one pack in {:.1} s",
75 
                    r.reachable,
76 
                    r.pruned.objects,
77 
                    r.pruned.bytes,
78 
                    if r.pruned.deferred > 0 { format!(", {} deferred by the grace period", r.pruned.deferred) } else { String::new() },
79 
                    r.packed.objects,
80 
                    started.elapsed().as_secs_f64()
81 
                ),
82 
                repo: slug,
83 
                failed: false,
84 
            },
85 
            Err(e) => Done { summary: format!("failed: {e}"), repo: slug, failed: true },
86 
        };
87 
        if d.failed {
88 
            tracing::warn!("gc {}", d);
89 
        } else {
90 
            tracing::info!("gc {}", d);
91 
        }
92 
        done.push(d);
93 
    }
94 
    done
95 
}
96
97 
/// Run [`gc_all`] once a day at `retention.gc_hour` UTC.
98 
pub async fn nightly(app: Shared) {
99 
    let hour = app.config.retention.gc_hour;
100 
    if hour >= 24 {
101 
        tracing::info!("gc: the nightly run is off (retention.gc_hour = {hour})");
102 
        return;
103 
    }
104 
    let mut last_day = None;
105 
    loop {
106 
        tokio::time::sleep(std::time::Duration::from_secs(60)).await;
107 
        let now = time::OffsetDateTime::now_utc();
108 
        if u32::from(now.hour()) != hour {
109 
            continue;
110 
        }
111 
        let day = now.date();
112 
        if last_day == Some(day) {
113 
            continue;
114 
        }
115 
        last_day = Some(day);
116 
        let done = gc_all(&app).await;
117 
        tracing::info!("gc: nightly run over {} repositories, {} failed", done.len(), done.iter().filter(|d| d.failed).count());
118 
    }
119 
}