Hi Maciej,
On Fri, Apr 05, 2019 at 01:33:29PM +0200, Maciej Zdeb wrote:
> I think I found something, please look at the session 0x2110edc0 and
> "calls", it never ends:
>
> socat /run/haproxy/haproxy2.sock - <<< "show sess all"
> 0x2110edc0: [05/Apr/2019:12:03:32.141927] id=14505 proto=tcpv4
> so
I think I found something, please look at the session 0x2110edc0 and
"calls", it never ends:
socat /run/haproxy/haproxy2.sock - <<< "show sess all"
0x2110edc0: [05/Apr/2019:12:03:32.141927] id=14505 proto=tcpv4
source=S.S.S.S:52414
flags=0x4504e, conn_retries=3, srv_conn=0xe2e7fb0, pend_pos=(nil
After process spin up to 100% cpu, I switched traffic to another (backup)
server then I attached gdb and executed "generate-core-file". However I
don't know how to proceed with it further (it's quite big 3,2GB).
When I'm attached with gdb sometimes (not always) "bt full" shows this (no
idea if it's
Hi again,
I verified that local DoS did not cause that high cpu usage. I'm upgrading
from 1.8.19 to 1.9.6.
Please look at strace:
of process with 100% cpu usage:
strace -fp 9790 -c
Process 9790 attached
% time seconds usecs/call callserrors syscall
-- --- --- ---
I'm very sorry, it is probably "real" traffic that haproxy needs to handle.
Some local DDoS occurred just in time when I upgraded to 1.9 (what terribly
bad luck).
Please ignore previous email! Sorry for wasting your time.
czw., 4 kwi 2019 o 14:36 Maciej Zdeb napisał(a):
> Hi,
>
> After haproxy
Hi,
After haproxy starts everything is fine, but after some time one of haproxy
processes uses 100% of CPU. I've managed to dump some "show" commands (I
have also "show fd", "show sess all" and "strace" but if needed I would
prefer to send it privately). If you could give any advice howto debug it
6 matches
Mail list logo
