Better query plan viewing experience (#4285)

* stuff * stuff including batch_upsert function * stuff * do things * stuff * different timestamps * stuff * Revert changes to comment.rs * Update comment.rs * Update comment.rs * Update post_view.rs * Update utils.rs * Update up.sql * Update up.sql * Update down.sql * Update up.sql * Update main.rs * use anyhow macro * replace get(0) with first() * as_slice * Update series.rs * Update db_perf.sh * Update and rename crates/db_schema/src/utils/series.rs to crates/db_perf/src/series.rs * Update utils.rs * Update main.rs * Update main.rs * Update .woodpecker.yml * fmt main.rs * Update .woodpecker.yml * Instance::delete at end * Update main.rs * Update Cargo.toml --------- Co-authored-by: Nutomic <me@nutomic.com>
2024-10-01 01:36:12 -04:00 · 2024-01-24 08:22:33 -07:00 · 2024-01-24 08:22:33 -07:00 · 759f6d8a9a
commit 759f6d8a9a
parent 8670403a67
15 changed files with 745 additions and 36 deletions
--- a/.woodpecker.yml
+++ b/.woodpecker.yml
@ -135,6 +135,18 @@ steps:
      - diesel migration redo
    when: *slow_check_paths

+  check_db_perf_tool:
+    image: *rust_image
+    environment:
+      LEMMY_DATABASE_URL: postgres://lemmy:password@database:5432/lemmy
+      RUST_BACKTRACE: "1"
+      CARGO_HOME: .cargo_home
+    commands:
+      # same as scripts/db_perf.sh but without creating a new database server
+      - export LEMMY_CONFIG_LOCATION=config/config.hjson
+      - cargo run --package lemmy_db_perf -- --posts 10 --read-post-pages 1
+    when: *slow_check_paths
+
  cargo_clippy:
    image: *rust_image
    environment:
--- a/Cargo.lock
+++ b/Cargo.lock
@ -2653,11 +2653,25 @@ dependencies = [
 "uuid",
 ]

+[[package]]
+name = "lemmy_db_perf"
+version = "0.19.0"
+dependencies = [
+ "clap",
+ "diesel",
+ "diesel-async",
+ "lemmy_db_schema",
+ "lemmy_db_views",
+ "lemmy_utils",
+ "tokio",
+]
+
 [[package]]
 name = "lemmy_db_schema"
 version = "0.19.3"
 dependencies = [
 "activitypub_federation",
+ "anyhow",
 "async-trait",
 "bcrypt",
 "chrono",
--- a/Cargo.toml
+++ b/Cargo.toml
@ -54,6 +54,7 @@ members = [
  "crates/api_common",
  "crates/apub",
  "crates/utils",
+  "crates/db_perf",
  "crates/db_schema",
  "crates/db_views",
  "crates/db_views_actor",
@ -156,6 +157,7 @@ tokio-postgres = "0.7.10"
 tokio-postgres-rustls = "0.10.0"
 enum-map = "2.7"
 moka = { version = "0.12.1", features = ["future"] }
+clap = { version = "4.4.11", features = ["derive"] }
 pretty_assertions = "1.4.0"

 [dependencies]
@ -193,7 +195,7 @@ futures-util = { workspace = true }
 chrono = { workspace = true }
 prometheus = { version = "0.13.3", features = ["process"] }
 serial_test = { workspace = true }
-clap = { version = "4.4.11", features = ["derive"] }
+clap = { workspace = true }
 actix-web-prom = "0.7.0"

 [dev-dependencies]
--- a/crates/db_perf/Cargo.toml
+++ b/crates/db_perf/Cargo.toml
@ -0,0 +1,23 @@
+[package]
+name = "lemmy_db_perf"
+version.workspace = true
+edition.workspace = true
+description.workspace = true
+license.workspace = true
+homepage.workspace = true
+documentation.workspace = true
+repository.workspace = true
+
+
+[lints]
+workspace = true
+
+[dependencies]
+anyhow = { workspace = true }
+clap = { workspace = true }
+diesel = { workspace = true }
+diesel-async = { workspace = true }
+lemmy_db_schema = { workspace = true }
+lemmy_db_views = { workspace = true, features = ["full"] }
+lemmy_utils = { workspace = true }
+tokio = { workspace = true }
--- a/crates/db_perf/src/main.rs
+++ b/crates/db_perf/src/main.rs
@ -0,0 +1,179 @@
+mod series;
+
+use crate::series::ValuesFromSeries;
+use anyhow::Context;
+use clap::Parser;
+use diesel::{
+  dsl::{self, sql},
+  sql_types,
+  ExpressionMethods,
+  IntoSql,
+};
+use diesel_async::{RunQueryDsl, SimpleAsyncConnection};
+use lemmy_db_schema::{
+  schema::post,
+  source::{
+    community::{Community, CommunityInsertForm},
+    instance::Instance,
+    person::{Person, PersonInsertForm},
+  },
+  traits::Crud,
+  utils::{build_db_pool, get_conn, now},
+  SortType,
+};
+use lemmy_db_views::{post_view::PostQuery, structs::PaginationCursor};
+use lemmy_utils::error::{LemmyErrorExt2, LemmyResult};
+use std::num::NonZeroU32;
+
+#[derive(Parser, Debug)]
+struct CmdArgs {
+  #[arg(long, default_value_t = 3.try_into().unwrap())]
+  communities: NonZeroU32,
+  #[arg(long, default_value_t = 3.try_into().unwrap())]
+  people: NonZeroU32,
+  #[arg(long, default_value_t = 100000.try_into().unwrap())]
+  posts: NonZeroU32,
+  #[arg(long, default_value_t = 0)]
+  read_post_pages: u32,
+  #[arg(long)]
+  explain_insertions: bool,
+}
+
+#[tokio::main]
+async fn main() -> anyhow::Result<()> {
+  let mut result = try_main().await.into_anyhow();
+  if let Ok(path) = std::env::var("PGDATA") {
+    result = result.with_context(|| {
+      format!("Failed to run lemmy_db_perf (more details might be available in {path}/log)")
+    });
+  }
+  result
+}
+
+async fn try_main() -> LemmyResult<()> {
+  let args = CmdArgs::parse();
+  let pool = &build_db_pool().await?;
+  let pool = &mut pool.into();
+  let conn = &mut get_conn(pool).await?;
+
+  if args.explain_insertions {
+    // log_nested_statements is enabled to log trigger execution
+    conn
+      .batch_execute(
+        "SET auto_explain.log_min_duration = 0; SET auto_explain.log_nested_statements = on;",
+      )
+      .await?;
+  }
+
+  let instance = Instance::read_or_create(&mut conn.into(), "reddit.com".to_owned()).await?;
+
+  println!("🫃 creating {} people", args.people);
+  let mut person_ids = vec![];
+  for i in 0..args.people.get() {
+    let form = PersonInsertForm::builder()
+      .name(format!("p{i}"))
+      .public_key("pubkey".to_owned())
+      .instance_id(instance.id)
+      .build();
+    person_ids.push(Person::create(&mut conn.into(), &form).await?.id);
+  }
+
+  println!("🌍 creating {} communities", args.communities);
+  let mut community_ids = vec![];
+  for i in 0..args.communities.get() {
+    let form = CommunityInsertForm::builder()
+      .name(format!("c{i}"))
+      .title(i.to_string())
+      .instance_id(instance.id)
+      .build();
+    community_ids.push(Community::create(&mut conn.into(), &form).await?.id);
+  }
+
+  let post_batches = args.people.get() * args.communities.get();
+  let posts_per_batch = args.posts.get() / post_batches;
+  let num_posts = post_batches * posts_per_batch;
+  println!(
+    "📜 creating {} posts ({} featured in community)",
+    num_posts, post_batches
+  );
+  let mut num_inserted_posts = 0;
+  // TODO: progress bar
+  for person_id in &person_ids {
+    for community_id in &community_ids {
+      let n = dsl::insert_into(post::table)
+        .values(ValuesFromSeries {
+          start: 1,
+          stop: posts_per_batch.into(),
+          selection: (
+            "AAAAAAAAAAA".into_sql::<sql_types::Text>(),
+            person_id.into_sql::<sql_types::Integer>(),
+            community_id.into_sql::<sql_types::Integer>(),
+            series::current_value.eq(1),
+            now()
+              - sql::<sql_types::Interval>("make_interval(secs => ")
+                .bind::<sql_types::BigInt, _>(series::current_value)
+                .sql(")"),
+          ),
+        })
+        .into_columns((
+          post::name,
+          post::creator_id,
+          post::community_id,
+          post::featured_community,
+          post::published,
+        ))
+        .execute(conn)
+        .await?;
+      num_inserted_posts += n;
+    }
+  }
+  // Make sure the println above shows the correct amount
+  assert_eq!(num_inserted_posts, num_posts as usize);
+
+  // Enable auto_explain
+  conn
+    .batch_execute(
+      "SET auto_explain.log_min_duration = 0; SET auto_explain.log_nested_statements = off;",
+    )
+    .await?;
+
+  // TODO: show execution duration stats
+  let mut page_after = None;
+  for page_num in 1..=args.read_post_pages {
+    println!(
+      "👀 getting page {page_num} of posts (pagination cursor used: {})",
+      page_after.is_some()
+    );
+
+    // TODO: include local_user
+    let post_views = PostQuery {
+      community_id: community_ids.as_slice().first().cloned(),
+      sort: Some(SortType::New),
+      limit: Some(20),
+      page_after,
+      ..Default::default()
+    }
+    .list(&mut conn.into())
+    .await?;
+
+    if let Some(post_view) = post_views.into_iter().last() {
+      println!("👀 getting pagination cursor data for next page");
+      let cursor_data = PaginationCursor::after_post(&post_view)
+        .read(&mut conn.into())
+        .await?;
+      page_after = Some(cursor_data);
+    } else {
+      println!("👀 reached empty page");
+      break;
+    }
+  }
+
+  // Delete everything, which might prevent problems if this is not run using scripts/db_perf.sh
+  Instance::delete(&mut conn.into(), instance.id).await?;
+
+  if let Ok(path) = std::env::var("PGDATA") {
+    println!("🪵 query plans written in {path}/log");
+  }
+
+  Ok(())
+}
--- a/crates/db_perf/src/series.rs
+++ b/crates/db_perf/src/series.rs
@ -0,0 +1,98 @@
+use diesel::{
+  dsl,
+  expression::{is_aggregate, ValidGrouping},
+  pg::Pg,
+  query_builder::{AsQuery, AstPass, QueryFragment},
+  result::Error,
+  sql_types,
+  AppearsOnTable,
+  Expression,
+  Insertable,
+  QueryId,
+  SelectableExpression,
+};
+
+/// Gererates a series of rows for insertion.
+///
+/// An inclusive range is created from `start` and `stop`. A row for each number is generated using `selection`, which can be a tuple.
+/// [`current_value`] is an expression that gets the current value.
+///
+/// For example, if there's a `numbers` table with a `number` column, this inserts all numbers from 1 to 10 in a single statement:
+///
+/// ```
+/// dsl::insert_into(numbers::table)
+///   .values(ValuesFromSeries {
+///     start: 1,
+///     stop: 10,
+///     selection: series::current_value,
+///   })
+///   .into_columns(numbers::number)
+/// ```
+#[derive(QueryId)]
+pub struct ValuesFromSeries<S> {
+  pub start: i64,
+  pub stop: i64,
+  pub selection: S,
+}
+
+impl<S: QueryFragment<Pg>> QueryFragment<Pg> for ValuesFromSeries<S> {
+  fn walk_ast<'b>(&'b self, mut out: AstPass<'_, 'b, Pg>) -> Result<(), Error> {
+    self.selection.walk_ast(out.reborrow())?;
+    out.push_sql(" FROM generate_series(");
+    out.push_bind_param::<sql_types::BigInt, _>(&self.start)?;
+    out.push_sql(", ");
+    out.push_bind_param::<sql_types::BigInt, _>(&self.stop)?;
+    out.push_sql(")");
+
+    Ok(())
+  }
+}
+
+impl<S: Expression> Expression for ValuesFromSeries<S> {
+  type SqlType = S::SqlType;
+}
+
+impl<T, S: AppearsOnTable<current_value>> AppearsOnTable<T> for ValuesFromSeries<S> {}
+
+impl<T, S: SelectableExpression<current_value>> SelectableExpression<T> for ValuesFromSeries<S> {}
+
+impl<T, S: SelectableExpression<current_value>> Insertable<T> for ValuesFromSeries<S>
+where
+  dsl::BareSelect<Self>: AsQuery + Insertable<T>,
+{
+  type Values = <dsl::BareSelect<Self> as Insertable<T>>::Values;
+
+  fn values(self) -> Self::Values {
+    dsl::select(self).values()
+  }
+}
+
+impl<S: ValidGrouping<(), IsAggregate = is_aggregate::No>> ValidGrouping<()>
+  for ValuesFromSeries<S>
+{
+  type IsAggregate = is_aggregate::No;
+}
+
+#[allow(non_camel_case_types)]
+#[derive(QueryId, Clone, Copy, Debug)]
+pub struct current_value;
+
+impl QueryFragment<Pg> for current_value {
+  fn walk_ast<'b>(&'b self, mut out: AstPass<'_, 'b, Pg>) -> Result<(), Error> {
+    out.push_identifier("generate_series")?;
+
+    Ok(())
+  }
+}
+
+impl Expression for current_value {
+  type SqlType = sql_types::BigInt;
+}
+
+impl AppearsOnTable<current_value> for current_value {}
+
+impl SelectableExpression<current_value> for current_value {}
+
+impl ValidGrouping<()> for current_value {
+  type IsAggregate = is_aggregate::No;
+}
--- a/crates/db_schema/Cargo.toml
+++ b/crates/db_schema/Cargo.toml
@ -76,6 +76,7 @@ tokio-postgres = { workspace = true, optional = true }
 tokio-postgres-rustls = { workspace = true, optional = true }
 rustls = { workspace = true, optional = true }
 uuid = { workspace = true, features = ["v4"] }
+anyhow = { workspace = true }

 [dev-dependencies]
 serial_test = { workspace = true }
--- a/crates/db_schema/src/impls/actor_language.rs
+++ b/crates/db_schema/src/impls/actor_language.rs
@ -96,16 +96,18 @@ impl LocalUserLanguage {
            .execute(conn)
            .await?;

-          for l in lang_ids {
-            let form = LocalUserLanguageForm {
+          let forms = lang_ids
+            .into_iter()
+            .map(|l| LocalUserLanguageForm {
              local_user_id: for_local_user_id,
              language_id: l,
-            };
-            insert_into(local_user_language)
-              .values(form)
-              .get_result::<Self>(conn)
-              .await?;
-          }
+            })
+            .collect::<Vec<_>>();
+
+          insert_into(local_user_language)
+            .values(forms)
+            .execute(conn)
+            .await?;
          Ok(())
        }) as _
      })
@ -164,16 +166,18 @@ impl SiteLanguage {
            .execute(conn)
            .await?;

-          for l in lang_ids {
-            let form = SiteLanguageForm {
+          let forms = lang_ids
+            .into_iter()
+            .map(|l| SiteLanguageForm {
              site_id: for_site_id,
              language_id: l,
-            };
-            insert_into(site_language)
-              .values(form)
-              .get_result::<Self>(conn)
-              .await?;
-          }
+            })
+            .collect::<Vec<_>>();
+
+          insert_into(site_language)
+            .values(forms)
+            .get_result::<Self>(conn)
+            .await?;

          CommunityLanguage::limit_languages(conn, instance_id).await?;

--- a/crates/db_schema/src/utils.rs
+++ b/crates/db_schema/src/utils.rs
@ -6,6 +6,7 @@ use crate::{
  SortType,
 };
 use activitypub_federation::{fetch::object_id::ObjectId, traits::Object};
+use anyhow::Context;
 use chrono::{DateTime, Utc};
 use deadpool::Runtime;
 use diesel::{
@ -13,6 +14,8 @@ use diesel::{
  deserialize::FromSql,
  helper_types::AsExprOf,
  pg::Pg,
+  query_builder::{Query, QueryFragment},
+  query_dsl::methods::LimitDsl,
  result::{ConnectionError, ConnectionResult, Error as DieselError, Error::QueryBuilderError},
  serialize::{Output, ToSql},
  sql_types::{Text, Timestamptz},
@ -150,6 +153,67 @@ macro_rules! try_join_with_pool {
  }};
 }

+/// Includes an SQL comment before `T`, which can be used to label auto_explain output
+#[derive(QueryId)]
+pub struct Commented<T> {
+  comment: String,
+  inner: T,
+}
+
+impl<T> Commented<T> {
+  pub fn new(inner: T) -> Self {
+    Commented {
+      comment: String::new(),
+      inner,
+    }
+  }
+
+  /// Adds `text` to the comment if `condition` is true
+  pub fn text_if(mut self, text: &str, condition: bool) -> Self {
+    if condition {
+      if !self.comment.is_empty() {
+        self.comment.push_str(", ");
+      }
+      self.comment.push_str(text);
+    }
+    self
+  }
+
+  /// Adds `text` to the comment
+  pub fn text(self, text: &str) -> Self {
+    self.text_if(text, true)
+  }
+}
+
+impl<T: Query> Query for Commented<T> {
+  type SqlType = T::SqlType;
+}
+
+impl<T: QueryFragment<Pg>> QueryFragment<Pg> for Commented<T> {
+  fn walk_ast<'b>(
+    &'b self,
+    mut out: diesel::query_builder::AstPass<'_, 'b, Pg>,
+  ) -> Result<(), DieselError> {
+    for line in self.comment.lines() {
+      out.push_sql("\n-- ");
+      out.push_sql(line);
+    }
+    out.push_sql("\n");
+    self.inner.walk_ast(out.reborrow())
+  }
+}
+
+impl<T: LimitDsl> LimitDsl for Commented<T> {
+  type Output = Commented<T::Output>;
+
+  fn limit(self, limit: i64) -> Self::Output {
+    Commented {
+      comment: self.comment,
+      inner: self.inner.limit(limit),
+    }
+  }
+}
+
 pub fn fuzzy_search(q: &str) -> String {
  let replaced = q.replace('%', "\\%").replace('_', "\\_").replace(' ', "%");
  format!("%{replaced}%")
@ -275,15 +339,18 @@ impl ServerCertVerifier for NoCertVerifier {

 pub const MIGRATIONS: EmbeddedMigrations = embed_migrations!();

-fn run_migrations(db_url: &str) {
+fn run_migrations(db_url: &str) -> Result<(), LemmyError> {
  // Needs to be a sync connection
  let mut conn =
-    PgConnection::establish(db_url).unwrap_or_else(|e| panic!("Error connecting to {db_url}: {e}"));
+    PgConnection::establish(db_url).with_context(|| format!("Error connecting to {db_url}"))?;
+
  info!("Running Database migrations (This may take a long time)...");
-  let _ = &mut conn
+  conn
    .run_pending_migrations(MIGRATIONS)
-    .unwrap_or_else(|e| panic!("Couldn't run DB Migrations: {e}"));
+    .map_err(|e| anyhow::anyhow!("Couldn't run DB Migrations: {e}"))?;
  info!("Database migrations complete.");
+
+  Ok(())
 }

 pub async fn build_db_pool() -> Result<ActualDbPool, LemmyError> {
@ -304,7 +371,7 @@ pub async fn build_db_pool() -> Result<ActualDbPool, LemmyError> {
    .runtime(Runtime::Tokio1)
    .build()?;

-  run_migrations(&db_url);
+  run_migrations(&db_url)?;

  Ok(pool)
 }
--- a/crates/db_views/src/post_view.rs
+++ b/crates/db_views/src/post_view.rs
@ -43,6 +43,7 @@ use lemmy_db_schema::{
    get_conn,
    limit_and_offset,
    now,
+    Commented,
    DbConn,
    DbPool,
    ListFn,
@ -282,7 +283,10 @@ fn queries<'a>() -> Queries<
          );
      }

-      query.first::<PostView>(&mut conn).await
+      Commented::new(query)
+        .text("PostView::read")
+        .first::<PostView>(&mut conn)
+        .await
    };

  let list = move |mut conn: DbConn<'a>, options: PostQuery<'a>| async move {
@ -564,7 +568,14 @@ fn queries<'a>() -> Queries<

    debug!("Post View Query: {:?}", debug_query::<Pg, _>(&query));

-    query.load::<PostView>(&mut conn).await
+    Commented::new(query)
+      .text("PostQuery::list")
+      .text_if(
+        "getting upper bound for next query",
+        options.community_id_just_for_prefetch,
+      )
+      .load::<PostView>(&mut conn)
+      .await
  };

  Queries::new(read, list)
--- a/migrations/2023-12-19-210053_tolerable-batch-insert-speed/down.sql
+++ b/migrations/2023-12-19-210053_tolerable-batch-insert-speed/down.sql
@ -0,0 +1,88 @@
+CREATE OR REPLACE FUNCTION post_aggregates_post ()
+    RETURNS TRIGGER
+    LANGUAGE plpgsql
+    AS $$
+BEGIN
+    IF (TG_OP = 'INSERT') THEN
+        INSERT INTO post_aggregates (post_id, published, newest_comment_time, newest_comment_time_necro, community_id, creator_id, instance_id)
+        SELECT
+            NEW.id,
+            NEW.published,
+            NEW.published,
+            NEW.published,
+            NEW.community_id,
+            NEW.creator_id,
+            community.instance_id
+        FROM
+            community
+        WHERE
+            NEW.community_id = community.id;
+    ELSIF (TG_OP = 'DELETE') THEN
+        DELETE FROM post_aggregates
+        WHERE post_id = OLD.id;
+    END IF;
+    RETURN NULL;
+END
+$$;
+
+CREATE OR REPLACE TRIGGER post_aggregates_post
+    AFTER INSERT OR DELETE ON post
+    FOR EACH ROW
+    EXECUTE PROCEDURE post_aggregates_post ();
+
+CREATE OR REPLACE TRIGGER community_aggregates_post_count
+    AFTER INSERT OR DELETE OR UPDATE OF removed,
+    deleted ON post
+    FOR EACH ROW
+    EXECUTE PROCEDURE community_aggregates_post_count ();
+
+DROP FUNCTION IF EXISTS community_aggregates_post_count_insert CASCADE;
+
+DROP FUNCTION IF EXISTS community_aggregates_post_update CASCADE;
+
+DROP FUNCTION IF EXISTS site_aggregates_post_update CASCADE;
+
+DROP FUNCTION IF EXISTS person_aggregates_post_insert CASCADE;
+
+CREATE OR REPLACE FUNCTION site_aggregates_post_insert ()
+    RETURNS TRIGGER
+    LANGUAGE plpgsql
+    AS $$
+BEGIN
+    IF (was_restored_or_created (TG_OP, OLD, NEW)) THEN
+        UPDATE
+            site_aggregates sa
+        SET
+            posts = posts + 1
+        FROM
+            site s
+        WHERE
+            sa.site_id = s.id;
+    END IF;
+    RETURN NULL;
+END
+$$;
+
+CREATE OR REPLACE TRIGGER site_aggregates_post_insert
+    AFTER INSERT OR UPDATE OF removed,
+    deleted ON post
+    FOR EACH ROW
+    WHEN (NEW.local = TRUE)
+    EXECUTE PROCEDURE site_aggregates_post_insert ();
+
+CREATE OR REPLACE FUNCTION generate_unique_changeme ()
+    RETURNS text
+    LANGUAGE sql
+    AS $$
+    SELECT
+        'http://changeme.invalid/' || substr(md5(random()::text), 0, 25);
+$$;
+
+CREATE OR REPLACE TRIGGER person_aggregates_post_count
+    AFTER INSERT OR DELETE OR UPDATE OF removed,
+    deleted ON post
+    FOR EACH ROW
+    EXECUTE PROCEDURE person_aggregates_post_count ();
+
+DROP SEQUENCE IF EXISTS changeme_seq;
+
--- a/migrations/2023-12-19-210053_tolerable-batch-insert-speed/up.sql
+++ b/migrations/2023-12-19-210053_tolerable-batch-insert-speed/up.sql
@ -0,0 +1,166 @@
+-- Change triggers to run once per statement instead of once per row
+-- post_aggregates_post trigger doesn't need to handle deletion because the post_id column has ON DELETE CASCADE
+CREATE OR REPLACE FUNCTION post_aggregates_post ()
+    RETURNS TRIGGER
+    LANGUAGE plpgsql
+    AS $$
+BEGIN
+    INSERT INTO post_aggregates (post_id, published, newest_comment_time, newest_comment_time_necro, community_id, creator_id, instance_id)
+    SELECT
+        id,
+        published,
+        published,
+        published,
+        community_id,
+        creator_id,
+        (
+            SELECT
+                community.instance_id
+            FROM
+                community
+            WHERE
+                community.id = community_id
+            LIMIT 1)
+FROM
+    new_post;
+    RETURN NULL;
+END
+$$;
+
+CREATE OR REPLACE FUNCTION community_aggregates_post_count_insert ()
+    RETURNS TRIGGER
+    LANGUAGE plpgsql
+    AS $$
+BEGIN
+    UPDATE
+        community_aggregates
+    SET
+        posts = posts + post_group.count
+    FROM (
+        SELECT
+            community_id,
+            count(*)
+        FROM
+            new_post
+        GROUP BY
+            community_id) post_group
+WHERE
+    community_aggregates.community_id = post_group.community_id;
+    RETURN NULL;
+END
+$$;
+
+CREATE OR REPLACE FUNCTION person_aggregates_post_insert ()
+    RETURNS TRIGGER
+    LANGUAGE plpgsql
+    AS $$
+BEGIN
+    UPDATE
+        person_aggregates
+    SET
+        post_count = post_count + post_group.count
+    FROM (
+        SELECT
+            creator_id,
+            count(*)
+        FROM
+            new_post
+        GROUP BY
+            creator_id) post_group
+WHERE
+    person_aggregates.person_id = post_group.creator_id;
+    RETURN NULL;
+END
+$$;
+
+CREATE OR REPLACE TRIGGER post_aggregates_post
+    AFTER INSERT ON post REFERENCING NEW TABLE AS new_post
+    FOR EACH STATEMENT
+    EXECUTE PROCEDURE post_aggregates_post ();
+
+-- Don't run old trigger for insert
+CREATE OR REPLACE TRIGGER community_aggregates_post_count
+    AFTER DELETE OR UPDATE OF removed,
+    deleted ON post
+    FOR EACH ROW
+    EXECUTE PROCEDURE community_aggregates_post_count ();
+
+CREATE OR REPLACE TRIGGER community_aggregates_post_count_insert
+    AFTER INSERT ON post REFERENCING NEW TABLE AS new_post
+    FOR EACH STATEMENT
+    EXECUTE PROCEDURE community_aggregates_post_count_insert ();
+
+CREATE OR REPLACE FUNCTION site_aggregates_post_update ()
+    RETURNS TRIGGER
+    LANGUAGE plpgsql
+    AS $$
+BEGIN
+    IF (was_restored_or_created (TG_OP, OLD, NEW)) THEN
+        UPDATE
+            site_aggregates sa
+        SET
+            posts = posts + 1
+        FROM
+            site s
+        WHERE
+            sa.site_id = s.id;
+    END IF;
+    RETURN NULL;
+END
+$$;
+
+CREATE OR REPLACE FUNCTION site_aggregates_post_insert ()
+    RETURNS TRIGGER
+    LANGUAGE plpgsql
+    AS $$
+BEGIN
+    UPDATE
+        site_aggregates sa
+    SET
+        posts = posts + (
+            SELECT
+                count(*)
+            FROM
+                new_post)
+    FROM
+        site s
+    WHERE
+        sa.site_id = s.id;
+    RETURN NULL;
+END
+$$;
+
+CREATE OR REPLACE TRIGGER site_aggregates_post_update
+    AFTER UPDATE OF removed,
+    deleted ON post
+    FOR EACH ROW
+    WHEN (NEW.local = TRUE)
+    EXECUTE PROCEDURE site_aggregates_post_update ();
+
+CREATE OR REPLACE TRIGGER site_aggregates_post_insert
+    AFTER INSERT ON post REFERENCING NEW TABLE AS new_post
+    FOR EACH STATEMENT
+    EXECUTE PROCEDURE site_aggregates_post_insert ();
+
+CREATE OR REPLACE TRIGGER person_aggregates_post_count
+    AFTER DELETE OR UPDATE OF removed,
+    deleted ON post
+    FOR EACH ROW
+    EXECUTE PROCEDURE person_aggregates_post_count ();
+
+CREATE OR REPLACE TRIGGER person_aggregates_post_insert
+    AFTER INSERT ON post REFERENCING NEW TABLE AS new_post
+    FOR EACH STATEMENT
+    EXECUTE PROCEDURE person_aggregates_post_insert ();
+
+-- Avoid running hash function and random number generation for default ap_id
+CREATE SEQUENCE IF NOT EXISTS changeme_seq AS bigint CYCLE;
+
+CREATE OR REPLACE FUNCTION generate_unique_changeme ()
+    RETURNS text
+    LANGUAGE sql
+    AS $$
+    SELECT
+        'http://changeme.invalid/seq/' || nextval('changeme_seq')::text;
+$$;
+
--- a/scripts/db_perf.sh
+++ b/scripts/db_perf.sh
@ -0,0 +1,20 @@
+#!/usr/bin/env bash
+
+# This script runs crates/lemmy_db_perf/src/main.rs, which lets you see info related to database query performance, such as query plans.
+
+set -e
+
+CWD="$(cd -P -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd -P)"
+
+cd $CWD/../
+
+source scripts/start_dev_db.sh
+
+export LEMMY_CONFIG_LOCATION=config/config.hjson
+export RUST_BACKTRACE=1
+
+cargo run --package lemmy_db_perf -- "$@"
+
+pg_ctl stop --silent
+
+# $PGDATA directory is kept so log can be seen
--- a/scripts/start_dev_db.sh
+++ b/scripts/start_dev_db.sh
@ -2,23 +2,47 @@

 export PGDATA="$PWD/dev_pgdata"
 export PGHOST=$PWD
-export LEMMY_DATABASE_URL="postgresql://lemmy:password@/lemmy?host=$PWD"
+export DATABASE_URL="postgresql://lemmy:password@/lemmy?host=$PWD"
+export LEMMY_DATABASE_URL=$DATABASE_URL

 # If cluster exists, stop the server and delete the cluster
-if [ -d $PGDATA ]
+if [[ -d $PGDATA ]]
 then
-  # Prevent `stop` from failing if server already stopped
-  pg_ctl restart > /dev/null
-  pg_ctl stop
+  # Only stop server if it is running
+  pg_status_exit_code=0
+  (pg_ctl status > /dev/null) || pg_status_exit_code=$?
+  if [[ ${pg_status_exit_code} -ne 3 ]]
+  then
+    pg_ctl stop --silent
+  fi
+
  rm -rf $PGDATA
 fi

-# Create cluster
-initdb --username=postgres --auth=trust --no-instructions
+config_args=(
+  # Only listen to socket in current directory
+  -c listen_addresses=
+  -c unix_socket_directories=$PWD

-# Start server that only listens to socket in current directory
-pg_ctl start --options="-c listen_addresses= -c unix_socket_directories=$PWD" > /dev/null
+  # Write logs to a file in $PGDATA/log
+  -c logging_collector=on
+
+  # Allow auto_explain to be turned on
+  -c session_preload_libraries=auto_explain
+
+  # Include actual row amounts and run times for query plan nodes
+  -c auto_explain.log_analyze=on
+
+  # Don't log parameter values
+  -c auto_explain.log_parameter_max_length=0
+)
+
+# Create cluster
+pg_ctl init --silent --options="--username=postgres --auth=trust --no-instructions"
+
+# Start server
+pg_ctl start --silent --options="${config_args[*]}"

 # Setup database
-psql -c "CREATE USER lemmy WITH PASSWORD 'password' SUPERUSER;" -U postgres
-psql -c "CREATE DATABASE lemmy WITH OWNER lemmy;" -U postgres
+psql --quiet -c "CREATE USER lemmy WITH PASSWORD 'password' SUPERUSER;" -U postgres
+psql --quiet -c "CREATE DATABASE lemmy WITH OWNER lemmy;" -U postgres
--- a/scripts/test.sh
+++ b/scripts/test.sh
@ -27,5 +27,5 @@ cargo test -p lemmy_utils --all-features --no-fail-fast

 # Add this to do printlns: -- --nocapture

-pg_ctl stop
+pg_ctl stop --silent
 rm -rf $PGDATA