Christopher Faulet

Ok thanks, I'm closing the issue.

Can we close the issue or should we change it to a feature request ?

Thanks for report. Could you provide your configuration and logs corresponding to those requests ? For now, it is not clear if the issue was observed on TCP or HTTP...

Thanks ! I'm surprised that HAProxy has missed the first reset and I'm unable to reproduce the issue with quick tests. Can you tell me from where the capture was...

@infa-sanaik, what is the status of this issue ? Have you tested a up-to-date version ? or maybe a more recent version ?

I guess with 53 peers, you've hit some limits. It is not a dead lock. At least, reading the code, it seems good. but the locking on the peers is...

Ok, nice. I'll try to find a way to not lock all peers from the task dealing with reconnect. It is most probably the root cause. But it is not...

Just to let you know. It is harder than expected, but after a short brainstorming with Emeric, we identified a way to improve the situation by avoiding to lock all...

@idl0r, I finally finished changes to no longer lock all peers at once in the sync task. I've performed some tests of course. But it is pretty hard to validate...

No problem, thanks. The issue is not really new and I guess having so many peers is quite uncommon :) So no stress !