Bridges take very long time to answer (~20min-2h)

Description

Hello! I’m running an homeserver on Debian11 Proxmox LXC with 2 CPUs and 4GB RAM (never peaked out until now). On the homeserver I have Discord, Signal, Telegram and WhatsApp Bridge running. When no one interacted with bridge bot (no matter which one) for a while and initiates a conversation (invite or message) the bot takes a really long time for the first answer to appear or for it to join the chat. But after the first reply of the bot and the user answers thereafter, it works without problems. When not contacting the bot for a while, the problem repeats. Weirdly, the bridging itself works the whole time. Since this issue is with every bot I thougt it should be a problem with synapse.

I looked into the logs, but didn’t find anything, in my opinion relevant. It looks like the messages take a long to be sent to the bridge bot, rebooting the bridge doesn’t help, but after rebooting synapse, the bots are answering instantly.

Steps to reproduce

1. Install Synapse via deb package manager
2. Install bridges according to official instructions
3. Enable double puppeting according to official instructions
4. Write message to bot and wait

Homeserver

https://chat.flowerhouse.at

Synapse Version

{“server_version”:“1.63.1”,“python_version”:“3.9.2”}

Installation Method

Debian packages from packages.matrix.org

Platform

Debian11 ProxMox LXC

Relevant log output

Didn't find anything relevant in the logs

Anything else that would be useful to know?

No response

[H-Shay]

A few questions: when you say

bot is dead and never online

are you determining this via presence or are there other indications that the bot is not working?

Also just to make sure this is clear, you removed localhost from the ::1 line and it solved the issue with the slow bot responses, but now heisenbridge (which was working before) is no longer working?

Finally, you mention writing messages to heisenbridge around 1:28, but the last timestamp on the logs provided is 1:15:02, could you provide logs that cover sending messages to heisenbridge?

[reivilibre]

@Tropaion

I’m definitely not sure if this is the problem or not, but you have two entries for localhost:

127.0.0.1       localhost
::1             localhost ip6-localhost ip6-loopback

The ::1 entry is for IPv6. IPv6 triggers an immediate suspicion alert for me here. My local machine has this in its hosts file:

127.0.0.1       localhost

# The following lines are desirable for IPv6 capable hosts
::1     ip6-localhost ip6-loopback

In other words, it doesn’t call ::1 localhost.

Try removing localhost from the ::1 line and then trying again. I don’t think you need to restart Synapse, but you can try in case something is stuck because it can’t contact your application service.

Let me know how you get on.