create-interaction-response! blocks after a few consecutive calls

[Avetixz]

Describe the bug
I'm calling discljord.messaging/create-interaction-response! in response to a button being clicked and after around 4 clicks, this function blocks and interactions become very slow. The bot then displays "This interaction failed".

If I don't use discljord's create-interaction-response! function for this and instead just send a plain HTTP request, this same problem doesn't occur.

Version Information

• Clojure Version: 1.10.1
• Discljord Version: 1.3.1
• OS bot is hosted on: Arch Linux

To Reproduce
Using my minimal reproduction case:

1. Use the /open command to make the bot create a message with a button
2. Press the button a few times
3. The bot will stop responding

Minimal reproduction case:
https://github.com/Avetixz/minimal-tournament-bot

Expected behavior
The bot should not block so quickly after just a few interactions.

Additional context
In my minimal reproduction case, I wrote a private function create-interaction-response! which sends a plain HTTP request. If you replace discord-rest/create-interaction-response! with create-interaction-response! in line 66, you'll see that the same problem doesn't occur.

[JohnnyJayJay]

I set up a small test myself on the latest develop changes, I didn't get this problem. I also tried your exact reproduction case (thank you very much for providing that, btw!). I could not reproduce the issue – I can press the button as soon as it comes up without issues.

Would you mind enabling trace logging for discljord.messaging.impl and sharing the log output (make sure to redact the bot token)? You can do this via the following logback.xml configuration (with logback-classic), for example:

<?xml version="1.0" encoding="UTF-8" ?>
<!DOCTYPE configuration>

<configuration scan="true" scanPeriod="15 seconds">
  <import class="ch.qos.logback.classic.encoder.PatternLayoutEncoder"/>
  <import class="ch.qos.logback.core.ConsoleAppender"/>

  <appender name="STDOUT" class="ConsoleAppender">
    <encoder class="PatternLayoutEncoder">
      <pattern>%d{HH:mm:ss.SSS} [%thread] %-5level %logger{36} -%kvp- %msg%n</pattern>
    </encoder>
  </appender>

  <logger name="discljord.messaging.impl" level="trace" />
  <root level="info">
    <appender-ref ref="STDOUT"/>
  </root>
</configuration>

Hopefully we get more insights that way.

[JohnnyJayJay]

Getting back to this rather late, but I have a growing suspicion that this is a system clock issue. I experienced something very similar yesterday and the cause ended up being that my NTP service (on Linux) didn't use a remote server to sync the system clock, so the rate limiter ended up computing incorrect reset times.

It's just speculation without a closer look at the trace logs, but it's an obscure enough cause that barely anyone else will catch it.