CPU hog about zeromq.node HOT 10 CLOSED

I'll keep plugging on my side, but thought I'd throw this issue up in case I'm hitting a PEBKAC or missing something obvious.

from zeromq.node.

It's my fault, i'm waiting for incoming messages in a very stupid way. I'm going to rework that shortly (hopefully tonight?).

from zeromq.node.

I'm experiencing the same problem. Hoping for a fix soon!

from zeromq.node.

Sorry! I'm hoping to get some time to work on it this weekend, since Wednesday didn't happen.

from zeromq.node.

No worries :) If you could give a quick overview/pointer about how you want the polling done I can take a whack at a patch. I tried some quick & dirty stuff yesterday (like setting a non-zero timeout on zmq_poll()) but only got segfaults.

from zeromq.node.

thanks, I saw in your commit logs that you had a plan to redesign the driver. Is that the case in the future?

from zeromq.node.

Yup. The API shouldn't change (at least not right now), but the 100% CPU thing obviously needs to be fixed.

@tobert - I'm planning on moving reads to a blocking eio_custom thread instead of continuously polling. Sends should stay the same. Go for it if you're feeling brave :)

from zeromq.node.

I did look into this this week, and it's more complicated than I first though due to the thread-safety of sockets (there isn't any). Hopefully I'll have some time to get back to this sometime next week.

from zeromq.node.

This should be fixed with the reworks done by mlc and stephank

from zeromq.node.

awesome thanks

from zeromq.node.