Table device_lists_changes_in_room grows in rapid pace, synapse stops with full database

[ABerhorn]

Description

We are using synapse for an internal chat application. The client is build with the matrix-js-sdk (v34.11.1).

The table device_lists_changes_in_room is constantly is growing in rows constantly.
There seems to be no cleanup (see #13043).
Synapse fails with full database errors until we added disk space to postgreSQL.

Here is the top 5 of the "Show top 20 larger tables by storage size"-sql query from the "useful sql for admins"-page:

• public.device_lists_changes_in_room
  423 GB
• public.access_tokens
  1050 MB
• public.devices
  851 MB
• public.device_lists_stream
  727 MB
• public.event_json
  677 MB

Are there any workarounds that might help us?

Steps to reproduce

• constant chat usage of the chat by different users

Homeserver

internal

Synapse Version

1.121.1

Installation Method

Docker (matrixdotorg/synapse)

Database

PostgreSQL

Workers

Single process

Platform

Kubernetes deployment

Configuration

• presence: Is on default setting

Relevant log output

-

Anything else that would be useful to know?

#13043 is related