Improve user in-guild sync process

Previously we set all users in_guild to False, and relied on users being set back to in_guild when iterating through guild.members

However, this caused two problems
1. For a short window a users in_guild status was incorrect
2. It required an update for all users in_guild to be sent to postgres to update in_guild back to True.

This diff changes that, so instead only users who are not found in the guild have in_guild set to False.

The bottleneck for this query is the number of users that are currently in_guild=False.
Testing locally, with 360k users off guild, this took 7.4s to query out, and 0.5s to process & 15.1 s to commit.

1 files changed, 15 insertions, 6 deletions

diff --git a/metricity/exts/event_listeners/startup_sync.py b/metricity/exts/event_listeners/startup_sync.py
index 0f6264f..017ea1f 100644
--- a/metricity/exts/event_listeners/startup_sync.py
+++ b/metricity/exts/event_listeners/startup_sync.py
@@ -5,7 +5,7 @@ import math
 import discord
 from discord.ext import commands
 from pydis_core.utils import logging, scheduling
-from sqlalchemy import column, update
+from sqlalchemy import column, select
 from sqlalchemy.dialects.postgresql import insert
 
 from metricity import models
@@ -35,10 +35,6 @@ class StartupSyncer(commands.Cog):
         await _syncer_utils.sync_thread_archive_state(guild)
 
         log.info("Beginning user synchronisation process")
-        async with async_session() as sess:
-            await sess.execute(update(models.User).values(in_guild=False))
-            await sess.commit()
-
         users = (
             {
                 "id": str(user.id),
@@ -85,7 +81,6 @@ class StartupSyncer(commands.Cog):
                 ))
 
                 objs = list(res)
-
                 created += [obj[0] == 0 for obj in objs].count(True)
                 updated += [obj[0] != 0 for obj in objs].count(True)
 
@@ -95,6 +90,20 @@ class StartupSyncer(commands.Cog):
             await sess.commit()
 
         log.info("User upsert complete")
+        log.info("Beginning user in_guild sync")
+
+        users_updated = 0
+        guild_member_ids = {str(member.id) for member in guild.members}
+        async with async_session() as sess:
+            res = await sess.execute(select(models.User).filter_by(in_guild=True))
+            in_guild_users = res.scalars()
+            for user in in_guild_users:
+                if user.id not in guild_member_ids:
+                    users_updated += 1
+                    user.in_guild = False
+            await sess.commit()
+        log.info("User in_guild sync updated  %d users to be off guild", users_updated)
+        log.info("User sync complete")
 
         self.bot.sync_process_complete.set()