High CPU usage after sleep about oban HOT 5 CLOSED

@tomtaylor Thanks for opening this up. On the one hand, I don't know if there is anything that can be done to prevent timer messages from backing up when the system is sleeping. On the other hand, that number of messages seemed excessively high and it led me to check the producer module.

There were two timers running per producer, which would have effectively doubled the number of messages to handle and duplicated the polling queries. I've pushed up a fix to address this and will be releasing v0.6.0 shortly.

from oban.

I remember running into something like this with rabbitmq a while ago. I think I ended up turning off rabbitmq before I put my laptop to sleep :(

@sorentwo could you describe a little more what the problem is? I think I'd like to start a discussion on it in on the ElixirForum (on the general topic of sleeping and timers).

from oban.

could you describe a little more what the problem is? I think I'd like to start a discussion on it in on the ElixirForum (on the general topic of sleeping and timers).

Sure, I can explain to the best of my ability.

The producer uses :timer.send_interval/2 to repeatedly send the producer a poll message every second. The :timer module uses a separate process to manage the timer state and it will keep sending the producer the :poll message regardless of whether the producer has handled the previous message. Apaprently, and here I'm hypothesizing, the timer process either doesn't sleep when the laptop is suspended, or it tries to catch up rapidly when the laptop comes back from being suspended.

After researching the underlying issue a bit more (thanks for prompting me) I'm going to switch to using recursive send_after instead of utilities through the :timer module. That will be more efficient in overall and will also prevent this issue entirely.

from oban.

Oh nice, that makes sense as a fix, especially given how this loop is being used 👍

My laptop thanks you!

from oban.

Thanks @sorentwo!

from oban.