mautrix-telegram completely obliterated my synapse instance

[nappa85]

Description

I know this sounds like a support request, but I've already written in #synapse:matrix.org and they suggested me to open a bug anyway, and I really want to help triaging this problem.

As the title suggests, synapse container become literally unresponsive, it can't be killed, not even with SIGKILL, after I started bridging a quite huge telegram account. I was already successfully bridging some whatsapp and signal accounts.

Container cannot be inspected (command waits forever), and ps -o pid,stat,wchan,cmd -p <docker-pid> returns

  PID STAT WCHAN  CMD
17955 Ssl  futex_ /var/packages/ContainerManager/target/usr/bin/dockerd --config-file /var/packages/ContainerManager/etc/dockerd.json

All other containers in the machine are running without problems, machine is running with load average 0.95, the only problematic container is synapse.

Steps to reproduce

• install synapse
• install mautrix-telegram
• bridge a beefy telegram account

Homeserver

nappa85.synology.me

Synapse Version

container latest sha-1bfcc9a

Installation Method

Docker (matrixdotorg/synapse)

Database

started from a clean postgres:15-alpine container

Workers

Single process

Platform

synapse and mautrix bridges are running in a docker compose inside a Synology NAS, so some things are a bit opaque or don't work exactly like a normal linux server

Configuration

no experimental or undocumented features, matrix has been installed less than one week ago and I was still getting to know the system

Relevant log output

I can't get logs from `docker logs <synapse-container>` (it simply waits forever, like trying to kill the container), but I can see logs from Synology container manager, but I'm not sure they're all, I suspect they probably stopped flowing at some point.

From the logs I can see, there are mainly postgres errors like 

synapse.storage.txn - 827 - WARNING - PUT-212 - [TXN OPERROR] {set_profile_avatar_url-20b} could not serialize access due to concurrent update

Anything else that would be useful to know?

No response

[nappa85]

I've been able to remove the container by rebooting the machine and then doing docker-compose down as soon as possible.

Not knowing where or what to check, I tried restating the docker compose, that now seems to run correctly, the only difference is that I increased max_connections to 200 in postgres

I still see logs like

2025-12-15 09:25:57,054 - synapse.storage.txn - 827 - WARNING - PUT-529 - [TXN OPERROR] {set_profile_avatar_url-df1} could not serialize access due to concurrent update

2025-12-15 09:25:57,065 - synapse.http.server - 147 - ERROR - PUT-508 - Failed handle request via 'ProfileFieldRestServlet': <XForwardedForRequest at 0x7ffb743d7b10 method='PUT' uri='/_matrix/client/v3/profile/%40telegram_1180591789%3Anappa85.synology.me/avatar_url?user_id=@telegram_1180591789:nappa85.synology.me' clientproto='HTTP/1.1' site='8008'>

If you need anything to better understand the situation, just ask me, otherwise feel free to close the issue