Gunnar Þór Magnússon

> It would also be nice to add opentracing support to the carbonserver (and to the carbonapi). Sure. I've never actually worked on a system with tracing support, so I'd...

I think this one's my fault, introduced while working on [issue 237](https://github.com/lomik/go-carbon/issues/237). The problem we had was that before that series of commits, all targets in a fetch request would...

Thanks, I'll have a look in a bit.

At Booking.com we're running: - carbonapi 0.9.2 - carbonzipper 0.73 - go-carbon commit b1e6baf (current HEAD of master) - carbon-c-relay 2.6 We can vouch for this being a good combination....

> As about version at booking.com they are using 2.6 as it's the most stable one. Up to you if you want to try newer. We're on 2.6 mostly because...

> How does the delay of cancellation manifest? Is there an easy way to > provoke the issue? I'm not sure what you mean by the delay of cancellation. If...

You actually don't want a FIFO queue. Under normal, happy conditions it doesn't matter what type of queue you have because almost everything gets served. When you're under heavy load...

I can prove this with an Etch-A-Sketch (so to speak): https://github.com/gunnihinn/queuesim

> This model does not include the case described in the first post. Indeed; I only wanted to point out that FIFO queues are maybe not the best option, and...

Thanks, that's fair enough. If you'd like some help with or review of any of those prototypes or their design or implementation, I'd be happy to be of assistance.