OXYGENHUB
# b349e15a3eab oxygenhub / crates / hub / src / maintenance.rs
4.6 KiB · Rust · last changed Drops: per-user gists, each a small hidden Oxygen repository (#279) 16 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 
    // Drops are repositories to keep tidy too (#279).
34 
    let rows = match db::repos::list_with_drops(&app.db).await {
35 
        Ok(rows) => rows,
36 
        Err(e) => {
37 
            tracing::warn!("gc: listing repositories: {e}");
38 
            return Vec::new();
39 
        }
40 
    };
41 
    let mut done = Vec::with_capacity(rows.len());
42 
    for row in rows {
43 
        let slug = format!("{}/{}", row.owner, row.name);
44 
        let Some(name) = RepoName::new(&row.owner, &row.name) else { continue };
45 
        if !app.repos.exists(&name) {
46 
            continue;
47 
        }
48 
        let label = slug.clone();
49 
        // Flow's artifacts and reports' files are named only in the
50 
        // database: kept as roots, or gc would prune them after the grace
51 
        // period (#126, #127). A listing that fails skips this repository
52 
        // rather than risk them.
53 
        let listed = match db::flow::artifact_blobs(&app.db, row.id).await {
54 
            Ok(mut blobs) => crate::reports::file_blobs(&app.db, row.id).await.map(|files| {
55 
                blobs.extend(files);
56 
                blobs
57 
            }),
58 
            Err(e) => Err(e),
59 
        };
60 
        let keep: Vec<oxy_core::hash::ObjectRef> = match listed {
61 
            Ok(blobs) => blobs.iter().filter_map(|b| oxy_core::hash::ObjectRef::from_hex(b)).collect(),
62 
            Err(e) => {
63 
                tracing::warn!("gc: {slug}: listing its artifacts: {e}; skipped");
64 
                continue;
65 
            }
66 
        };
67 
        let started = std::time::Instant::now();
68 
        let result = app
69 
            .repos
70 
            .write(&name, move |repo, _| Ok(repo.gc_keeping(false, &keep, &mut |phase| tracing::debug!("gc {label}: {phase}"))?))
71 
            .await;
72 
        let d = match result {
73 
            Ok(r) => Done {
74 
                summary: format!(
75 
                    "{} reachable; pruned {} ({} bytes){}; packed {} into one pack in {:.1} s",
76 
                    r.reachable,
77 
                    r.pruned.objects,
78 
                    r.pruned.bytes,
79 
                    if r.pruned.deferred > 0 { format!(", {} deferred by the grace period", r.pruned.deferred) } else { String::new() },
80 
                    r.packed.objects,
81 
                    started.elapsed().as_secs_f64()
82 
                ),
83 
                repo: slug,
84 
                failed: false,
85 
            },
86 
            Err(e) => Done { summary: format!("failed: {e}"), repo: slug, failed: true },
87 
        };
88 
        if d.failed {
89 
            tracing::warn!("gc {}", d);
90 
        } else {
91 
            tracing::info!("gc {}", d);
92 
        }
93 
        done.push(d);
94 
    }
95 
    done
96 
}
97
98 
/// Run [`gc_all`] once a day at `retention.gc_hour` UTC.
99 
pub async fn nightly(app: Shared) {
100 
    let hour = app.config.retention.gc_hour;
101 
    if hour >= 24 {
102 
        tracing::info!("gc: the nightly run is off (retention.gc_hour = {hour})");
103 
        return;
104 
    }
105 
    let mut last_day = None;
106 
    loop {
107 
        tokio::time::sleep(std::time::Duration::from_secs(60)).await;
108 
        let now = time::OffsetDateTime::now_utc();
109 
        if u32::from(now.hour()) != hour {
110 
            continue;
111 
        }
112 
        let day = now.date();
113 
        if last_day == Some(day) {
114 
            continue;
115 
        }
116 
        last_day = Some(day);
117 
        match crate::db::sessions::sweep(&app.db).await {
118 
            Ok(n) if n > 0 => tracing::info!("sessions: forgot {n} that had ended"),
119 
            Ok(_) => {}
120 
            Err(e) => tracing::warn!("sessions: sweep failed: {e}"),
121 
        }
122 
        let done = gc_all(&app).await;
123 
        tracing::info!("gc: nightly run over {} repositories, {} failed", done.len(), done.iter().filter(|d| d.failed).count());
124 
    }
125 
}