romanprog

Results 24 comments of romanprog

@zhangzujian > What's the destination ethernet address of the echo reply? If it's wrong, you can monitor both ICMP and ARP packets. Destination address isnt wrong, its always onv-eip's mac...

@zbb88888 > distributed ovn-fip the mac is always pod mac? no, its always oeip mac > did you use the arp proxy in the os? no, k8s + kube-ovn +...

> Are you using distributed ovn-fip? In this case we use OSNAT, but we also checked with OFIP. During network failure - all OSNAT OFIP in VPC do not work....

I created OFIP: ``` kind: OvnEip apiVersion: kubeovn.io/v1 metadata: name: oeip-test spec: externalSubnet: external-def type: nat --- kind: OvnFip apiVersion: kubeovn.io/v1 metadata: name: ofip-test spec: ovnEip: oeip-test vpc: shalb-demo v4Ip:...

I changed the option and recreated OFIP. And add exteral network monitoring on fip pod. ``` k ko nbctl list nat _uuid : 45258255-e09e-4885-9d1b-3fb8a5a6049a allowed_ext_ips : [] exempted_ext_ips : []...

@zbb88888 and we still have same network flapping on OFIP pod

ICMPs from a ofip pod placed on a `kube-dc-worker-1` come out of the `kube-dc-master-1` node's external interface, just like before changing the gateway type.

Let's clarify. I have 2 pods. On different nodes. I created an ofip for both pods. Added ping debugger and monitoring for both pods. All packets from both pods go...

Ok. Lets go step-by-step: - All outgoing packets pass the internal flow and are sent to the outside without problems (i have running tcpdump on all bridges and interfaces, using...