microcosm.blue
Server tools to backfill, tail, mirror, and verify PLC logs
move data to actual tables
seems a little slow perhaps
bad-example.com
+6
-1
src/backfill.rs
+6
-1
src/backfill.rs
···
8
8
source: impl BundleSource + Send + 'static,
9
9
dest: flume::Sender<ExportPage>,
10
10
source_workers: usize,
11
+
until: Option<Dt>,
11
12
) -> anyhow::Result<()> {
12
13
// queue up the week bundles that should be available
13
-
let weeks = Arc::new(Mutex::new(Week::range(FIRST_WEEK..)));
14
+
let weeks = Arc::new(Mutex::new(
15
+
until
16
+
.map(|u| Week::range(FIRST_WEEK..u.into()))
17
+
.unwrap_or(Week::range(FIRST_WEEK..)),
18
+
));
14
19
weeks.lock().await.reverse();
15
20
16
21
let mut workers: JoinSet<anyhow::Result<()>> = JoinSet::new();
+6
-2
src/bin/allegedly.rs
+6
-2
src/bin/allegedly.rs
···
37
37
/// Pass a postgres connection url like "postgresql://localhost:5432"
38
38
#[arg(long)]
39
39
to_postgres: Option<Url>,
40
+
/// Stop at the week ending before this date
41
+
#[arg(long)]
42
+
until: Option<Dt>,
40
43
},
41
44
/// Scrape a PLC server, collecting ops into weekly bundles
42
45
///
···
88
91
dir,
89
92
source_workers,
90
93
to_postgres,
94
+
until,
91
95
} => {
92
96
let (tx, rx) = flume::bounded(32); // big pages
93
97
tokio::task::spawn(async move {
94
98
if let Some(dir) = dir {
95
99
log::info!("Reading weekly bundles from local folder {dir:?}");
96
-
backfill(FolderSource(dir), tx, source_workers.unwrap_or(1))
100
+
backfill(FolderSource(dir), tx, source_workers.unwrap_or(1), until)
97
101
.await
98
102
.unwrap();
99
103
} else {
100
104
log::info!("Fetching weekly bundles from from {http}");
101
-
backfill(HttpSource(http), tx, source_workers.unwrap_or(4))
105
+
backfill(HttpSource(http), tx, source_workers.unwrap_or(4), until)
102
106
.await
103
107
.unwrap();
104
108
}
+32
-3
src/plc_pg.rs
+32
-3
src/plc_pg.rs
···
92
92
cid text not null,
93
93
operation jsonb not null,
94
94
nullified boolean not null,
95
-
createdAt timestamptz not null
95
+
"createdAt" timestamptz not null
96
96
)"#,
97
97
&[],
98
98
)
···
135
135
log::info!("copied in {n} rows");
136
136
137
137
tx.commit().await?;
138
+
log::info!("copy in time: {:?}", t0.elapsed());
138
139
139
-
let dt = t0.elapsed();
140
-
log::info!("backfill time: {dt:?}");
140
+
log::info!("copying dids into plc table...");
141
+
let n = client
142
+
.execute(
143
+
r#"
144
+
INSERT INTO dids
145
+
SELECT distinct did FROM allegedly_backfill
146
+
ON CONFLICT do nothing"#,
147
+
&[],
148
+
)
149
+
.await?;
150
+
log::info!("{n} inserted; elapsed: {:?}", t0.elapsed());
151
+
152
+
log::info!("copying ops into plc table...");
153
+
let n = client
154
+
.execute(
155
+
r#"
156
+
INSERT INTO operations (did, cid, operation, nullified, "createdAt")
157
+
SELECT did, cid, operation, nullified, "createdAt" FROM allegedly_backfill
158
+
ON CONFLICT do nothing"#,
159
+
&[],
160
+
)
161
+
.await?;
162
+
log::info!("{n} inserted; elapsed: {:?}", t0.elapsed());
163
+
164
+
log::info!("clean up backfill table...");
165
+
client
166
+
.execute(r#"DROP TABLE allegedly_backfill"#, &[])
167
+
.await?;
168
+
169
+
log::info!("total backfill time: {:?}", t0.elapsed());
141
170
142
171
Ok(())
143
172
}