Merge branch 'bnewbold-fill-changelog-gaps' into 'master'

rust: fill changelog gap helper tool

See merge request webgroup/fatcat!142

3 files changed, 138 insertions, 1 deletions

diff --git a/notes/2022-08-12_changelog_gap.md b/notes/2022-08-12_changelog_gap.md
new file mode 100644
index 00000000..48572973
--- /dev/null
+++ b/notes/2022-08-12_changelog_gap.md
@@ -0,0 +1,41 @@
+
+On 2022-08-11, realized that we had a "gap" in the changelog: after a VM
+reboot, the postgresql primary key sequence for the 'changelog' table had been
+incremented, but rows were not inserted (transaction hadn't finished).
+
+This was a known potential problem (naively relying on the sequence to
+increment with no gaps).
+
+As a work-around, implemented a simple "gap filler" which will create
+empty/dummy editgroups and changelog entries.
+
+This gap extends from 6153703 to 6153721, so just a couple dozen entries. The
+fixup command was:
+
+    ./target/release/fatcat-doctor backfill-changelog-gap 6153702 6153721
+    Inserted changelog: 6153703
+    Inserted changelog: 6153704
+    Inserted changelog: 6153705
+    Inserted changelog: 6153706
+    Inserted changelog: 6153707
+    Inserted changelog: 6153708
+    Inserted changelog: 6153709
+    Inserted changelog: 6153710
+    Inserted changelog: 6153711
+    Inserted changelog: 6153712
+    Inserted changelog: 6153713
+    Inserted changelog: 6153714
+    Inserted changelog: 6153715
+    Inserted changelog: 6153716
+    Inserted changelog: 6153717
+    Inserted changelog: 6153718
+    Inserted changelog: 6153719
+    Inserted changelog: 6153720
+    Inserted changelog: 6153721
+
+After that the changelog worker was happy:
+
+    Aug 13 02:41:59 wbgrp-svc502.us.archive.org fatcat-worker[386037]: Most recent changelog index in Kafka seems to be 6153702
+    Aug 13 02:41:59 wbgrp-svc502.us.archive.org fatcat-worker[386037]: Fetching changelogs from 6153703 through 6158547
+    Aug 13 02:43:12 wbgrp-svc502.us.archive.org fatcat-worker[386037]: Sleeping 5.0 seconds...
+
diff --git a/rust/.gitignore b/rust/.gitignore
index 03e50598..f237ae31 100644
--- a/rust/.gitignore
+++ b/rust/.gitignore
@@ -1,6 +1,5 @@
 .env
 target/
-bin/
 fatcat-*.tar.gz
 
 !.cargo
diff --git a/rust/src/bin/fatcat-doctor.rs b/rust/src/bin/fatcat-doctor.rs
new file mode 100644
index 00000000..6e869634
--- /dev/null
+++ b/rust/src/bin/fatcat-doctor.rs
@@ -0,0 +1,97 @@
+//! Database cleanup tool
+
+use clap::{value_t_or_exit, App, SubCommand};
+
+use fatcat::database_models::*;
+use fatcat::database_schema::*;
+use fatcat::errors::Result;
+use fatcat::identifiers::FatcatId;
+use fatcat::server;
+use fatcat::server::DbConn;
+use std::process;
+use std::str::FromStr;
+
+use diesel;
+use diesel::prelude::*;
+
+fn backfill_changelog_gap(conn: &DbConn, last_good: i64, max_index: i64) -> Result<()> {
+    // sanity check arguments against database
+    assert!(last_good > 0);
+    assert!(max_index > 0);
+    assert!(last_good < max_index);
+    let highest_row: ChangelogRow = changelog::table.order(changelog::id.desc()).first(conn)?;
+    assert!(highest_row.id >= max_index);
+
+    // default values
+    // 'root' editor_id is aaaaaaaaaaaabkvkaaaaaaaaae
+    // 'admin' editor_id is aaaaaaaaaaaabkvkaaaaaaaaai
+    let editor_id = FatcatId::from_str("aaaaaaaaaaaabkvkaaaaaaaaai").unwrap();
+    let description = "Backfill of missing changelog entries due to database id gap";
+
+    // fetch the last entry before the gap, to re-use the timestamp
+    let existing_row: ChangelogRow = changelog::table.find(last_good).first(conn)?;
+
+    for index in last_good + 1..max_index + 1 {
+        // ensure this index is actually a gap
+        let count: i64 = changelog::table
+            .filter(changelog::id.eq(index))
+            .count()
+            .get_result(conn)?;
+        if count != 0 {
+            println!("Found existing changelog: {}", index);
+            return Ok(());
+        }
+
+        // create dummy empty editgroup, then add a changelog entry
+        let eg_row: EditgroupRow = diesel::insert_into(editgroup::table)
+            .values((
+                editgroup::editor_id.eq(editor_id.to_uuid()),
+                editgroup::created.eq(existing_row.timestamp),
+                editgroup::is_accepted.eq(true),
+                editgroup::description.eq(Some(description)),
+            ))
+            .get_result(conn)?;
+        let _entry_row: ChangelogRow = diesel::insert_into(changelog::table)
+            .values((
+                changelog::id.eq(index),
+                changelog::editgroup_id.eq(eg_row.id),
+                changelog::timestamp.eq(existing_row.timestamp),
+            ))
+            .get_result(conn)?;
+        println!("Inserted changelog: {}", index);
+    }
+    Ok(())
+}
+
+fn main() -> Result<()> {
+    let m = App::new("fatcat-doctor")
+        .version(env!("CARGO_PKG_VERSION"))
+        .author("Bryan Newbold <bnewbold@archive.org>")
+        .about("Database cleanup / fixup tool")
+        .subcommand(
+            SubCommand::with_name("backfill-changelog-gap")
+                .about("Inserts dummy changelog entries and editgroups for gap")
+                .args_from_usage(
+                    "<start> 'changelog index of entry just before gap'
+                    <end> 'highest changelog index to backfill'",
+                ),
+        )
+        .get_matches();
+
+    let db_conn = server::database_worker_pool()?
+        .get()
+        .expect("database pool");
+    match m.subcommand() {
+        ("backfill-changelog-gap", Some(subm)) => {
+            let last_good: i64 = value_t_or_exit!(subm.value_of("start"), i64);
+            let max_index: i64 = value_t_or_exit!(subm.value_of("end"), i64);
+            backfill_changelog_gap(&db_conn, last_good, max_index)?;
+        }
+        _ => {
+            println!("Missing or unimplemented command!");
+            println!("{}", m.usage());
+            process::exit(-1);
+        }
+    }
+    Ok(())
+}