From: Dessalines <dessalines@users.noreply.github.com>
Date: Mon, 23 Jan 2023 14:59:25 +0000 (-0500)
Subject: Speeding up comment-ltree migration, fixing index creation. Fixes #2664 (#2670)
X-Git-Url: http://these/git/%22https:/nerdica.net/photo/contact/80/%24%7Bargs.pageFn.jump%20n%7D?a=commitdiff_plain;h=70c549dad8320a2bd2f46bb39521b571423bef36;p=lemmy.git

Speeding up comment-ltree migration, fixing index creation. Fixes #2664 (#2670)

* Speeding up comment-ltree migration, fixing index creation. Fixes #2664

* Adding some logging lines, fixing for missing posts.

* Adding more postgres config
---

diff --git a/Cargo.lock b/Cargo.lock
index 67a8b5cb..83a9494d 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -2488,6 +2488,8 @@ dependencies = [
  "strum",
  "strum_macros",
  "tokio",
+ "tracing",
+ "tracing-error",
  "typed-builder",
  "url",
 ]
diff --git a/crates/db_schema/Cargo.toml b/crates/db_schema/Cargo.toml
index d996e2ac..83a32ba3 100644
--- a/crates/db_schema/Cargo.toml
+++ b/crates/db_schema/Cargo.toml
@@ -40,6 +40,8 @@ typed-builder = { workspace = true }
 async-trait = { workspace = true }
 tokio = { workspace = true }
 bb8 = { version = "0.8.0", optional = true }
+tracing = { workspace = true }
+tracing-error = { workspace = true }
 
 [dev-dependencies]
 serial_test = { workspace = true }
diff --git a/crates/db_schema/src/utils.rs b/crates/db_schema/src/utils.rs
index f6accb60..cf96b5e7 100644
--- a/crates/db_schema/src/utils.rs
+++ b/crates/db_schema/src/utils.rs
@@ -26,6 +26,7 @@ use lemmy_utils::{error::LemmyError, settings::structs::Settings};
 use once_cell::sync::Lazy;
 use regex::Regex;
 use std::{env, env::VarError};
+use tracing::info;
 use url::Url;
 
 const FETCH_LIMIT_DEFAULT: i64 = 10;
@@ -154,9 +155,11 @@ pub fn run_migrations(db_url: &str) {
   // Needs to be a sync connection
   let mut conn =
     PgConnection::establish(db_url).unwrap_or_else(|_| panic!("Error connecting to {}", db_url));
+  info!("Running Database migrations (This may take a long time)...");
   let _ = &mut conn
     .run_pending_migrations(MIGRATIONS)
     .unwrap_or_else(|_| panic!("Couldn't run DB Migrations"));
+  info!("Database migrations complete.");
 }
 
 pub async fn build_db_pool(settings: &Settings) -> Result<DbPool, LemmyError> {
diff --git a/docker/dev/docker-compose.yml b/docker/dev/docker-compose.yml
index e38d0ca1..639f798a 100644
--- a/docker/dev/docker-compose.yml
+++ b/docker/dev/docker-compose.yml
@@ -104,17 +104,18 @@ services:
       # "-c", "effective_cache_size=9GB",
       # "-c", "maintenance_work_mem=768MB",
       # "-c", "checkpoint_completion_target=0.9",
+      # "-c", "checkpoint_timeout=86400",
       # "-c", "wal_buffers=16MB",
       # "-c", "default_statistics_target=100",
       # "-c", "random_page_cost=4",
       # "-c", "effective_io_concurrency=2",
-      # "-c", "work_mem=7864kB",
+      # "-c", "work_mem=4GB",
       # "-c", "min_wal_size=1GB",
-      # "-c", "max_wal_size=4GB",
+      # "-c", "max_wal_size=30GB",
       # "-c", "max_worker_processes=4",
       # "-c", "max_parallel_workers_per_gather=2",
       # "-c", "max_parallel_workers=4",
-      # "-c", "max_parallel_maintenance_workers=2",
+      # "-c", "max_parallel_maintenance_workers=2"
     ]
     networks:
       - lemmyinternal
diff --git a/migrations/2022-07-07-182650_comment_ltrees/up.sql b/migrations/2022-07-07-182650_comment_ltrees/up.sql
index 401b1fb6..fde9e1b3 100644
--- a/migrations/2022-07-07-182650_comment_ltrees/up.sql
+++ b/migrations/2022-07-07-182650_comment_ltrees/up.sql
@@ -1,4 +1,3 @@
-
 -- Remove the comment.read column, and create a new comment_reply table,
 -- similar to the person_mention table. 
 -- 
@@ -60,6 +59,19 @@ FROM    q
 ORDER BY
 	breadcrumb;
 
+-- Remove indexes and foreign key constraints, and disable triggers for faster updates
+alter table comment disable trigger all;
+
+alter table comment drop constraint if exists comment_creator_id_fkey;
+alter table comment drop constraint if exists comment_parent_id_fkey;
+alter table comment drop constraint if exists comment_post_id_fkey;
+alter table comment drop constraint if exists idx_comment_ap_id;
+
+drop index if exists idx_comment_creator;
+drop index if exists idx_comment_parent;
+drop index if exists idx_comment_post;
+drop index if exists idx_comment_published;
+
 -- Add the ltree column
 update comment c 
 set path = ct.ltree_path
@@ -75,9 +87,32 @@ from (
 ) as c2
 where ca.comment_id = c2.id;
 
+-- Delete comments at a depth of > 150, otherwise the index creation below will fail
+delete from comment where nlevel(path) > 150;
+
+-- Delete from comment where there is a missing post
+delete from comment c where not exists (
+  select from post p where p.id = c.post_id
+);
+
+-- Delete from comment where there is a missing creator_id
+delete from comment c where not exists (
+  select from person p where p.id = c.creator_id
+);
+
+-- Re-enable old constraints and indexes
+alter table comment add constraint "comment_creator_id_fkey" FOREIGN KEY (creator_id) REFERENCES person(id) ON UPDATE CASCADE ON DELETE CASCADE;
+alter table comment add constraint "comment_post_id_fkey" FOREIGN KEY (post_id) REFERENCES post(id) ON UPDATE CASCADE ON DELETE CASCADE;
+alter table comment add constraint "idx_comment_ap_id" unique (ap_id);
+
+create index idx_comment_creator on comment (creator_id);
+create index idx_comment_post on comment (post_id);
+create index idx_comment_published on comment (published desc);
+
 -- Create the index
 create index idx_path_gist on comment using gist (path);
 
 -- Drop the parent_id column
 alter table comment drop column parent_id cascade;
 
+alter table comment enable trigger all;