Bot used max memory + CPU

[tolik518]

https://grafana.r0.fyi/d/behftjz4mgk5ce/server-overview?orgId=1&from=now-6h&to=now&timezone=browser

Not sure yet what happened, but when I checked the container with docker logs, it started to normalize right after. Might be a docker issue and not a code issue.

Starting time around: 12:50:52

[Aras14HD]

The post that was replied to immediately after was this, a large tower with !all. Definitely a code issue. 100% just cloning all those large numbers.

I was already investigating and fixing an issue resulting in 15s+ calc times. I think I will cap the number of calculations after a number of steps, 10 makes sense.

[tolik518]

Makes sense, I'm just very confused why it was hang for so long and continue when I look at the docker logs, I'll investigate on the devops side aswell

[tolik518]

docker logs could have fixed the issue because it flushed the stdout buffer maybe? I think the bot has calculated everything but the output was just too long because of !all

[Aras14HD]

Might have even been more the info!("Comment -> {comment:?}"); which tried to print that vector with 3983 Calculation elements, all with probably large numbers to be formatted.

[tolik518]

Additionally: Even after it posted the number, the bot still used nearly 20GB memory, looks like something isn't cleared up?

[Aras14HD]

That's probably the default, gmp might delay cleanup. (like how Vec::pop/remove never deallocate)