Hi Willy,
Am 13.05.2016 um 17:01 schrieb Willy Tarreau:
Hi Sebastian,
On Thu, May 12, 2016 at 09:58:22AM +0200, Sebastian Heid wrote:
Hi Lukas,
starting from around 200mbit/s in&out, haproxy processes (nbproc 6) are hitting
100% cpu regularly (noticed up to 3 processes at the same time with 100%), but
recover again on its own after some time.
stracing such a process yesterday showed the following:
epoll_wait(0, {}, 200, 0) = 0
epoll_wait(0, {}, 200, 0) = 0
epoll_wait(0, {}, 200, 0) = 0
epoll_wait(0, {}, 200, 0) = 0
epoll_wait(0, {}, 200, 0) = 0
Unfortunately I can't do any more debugging in this setup. HAproxy 1.5.14 is never near to 10% cpu usage with way higher bandwidth.
So we still have another situation making haproxy think it can receive while
it cannot. There's a report of another issue possibly related to this on 1.6,
so let me try to understand it then I'll check if it may cause this behaviour
on 1.5 (assuming it's present there as well). Don't take risks for now.
Not sure if that's what you meant by the other issue, but if there are
still buffer issues it may also caused the reported crash in zlib (since
1.6.4 but also affects 1.6.5), that would be thread "Crash with kernel
error", where we are waiting for the OP to provide a stack trace.
cheers,
Lukas
