zhuangyan

VxLAN OpenFlow If dpdk-based ovs switch supports L4LB, why using dpvs?

> DPVS crashed after a 2-hours stress test using this patch. Traces show mbuf is tainted. > > ``` > #0 0x00007ffff6b315f7 in raise () from /lib64/libc.so.6 > #1 0x00007ffff6b32ce8...

Okay. We will duplicate this issue and fix it.

It seems related to the below configuration and it is day 1 issue. Have you tested without the patch of this PR? # persistence_timeout 15

I have figured out some solution which will be sent out soon.

With this patch, TCP CPS can be increased by 10% with 14 CPU cores used.

> It seems that many members of dp_vs_conn are still zerod with `memset`. Are you sure the latest patch make 10% performance enhancement? > > Besides, function pointer members (such...

Please have a check on the latest patch.

You might try disable fast_xmit_close in dpvs.conf to try fast xmit feature.

I just tested the config within static_ipaddress{} without the config within vrrp_instance{}, for vrrp_instance{} is not used in our deployment. ``` static_ipaddress { 172.21.33.4/30 dev dpdk0 } ```