trozet

[root@ovn-worker2 ~]# ovs-dpctl dump-flows
recirc_id(0x48),in_port(2),ct_state(+new-est-rel-rpl-inv+trk),ct_mark(0/0x1),eth(src=66:49:b8:4b:12:3e,dst=0a:58:0a:f4:01:05),eth_type(0x0800),ipv4(src=10.244.1.2,dst=10.244.1.5,frag=no), packets:0, bytes:0, used:never, actions:ct(commit,zone=19,mark=0/0x1,nat(src)),6
recirc_id(0),in_port(3),skb_mark(0),eth(dst=02:42:ac:12:00:03),eth_type(0x0800),ipv4(proto=6,frag=no),tcp(dst=8192/0xe000), packets:797, bytes:70414, used:0.004s, flags:SP., actions:ct(zone=64000,nat),recirc(0x27)
recirc_id(0),in_port(3),eth(dst=02:42:ac:12:00:03),eth_type(0x0800),ipv4(proto=17,frag=no),udp(dst=6081), packets:1, bytes:132, used:0.753s, actions:4
recirc_id(0x46),tunnel(tun_id=0xff0003,src=172.18.0.2,dst=172.18.0.3,geneve({}{}),flags(-df+csum+key)),in_port(5),ct_state(+new-est-rel-rpl-inv+trk),ct_mark(0/0x1),eth(src=0a:58:0a:f4:01:01,dst=00:00:00:00:00:00/01:00:00:00:00:00),eth_type(0x0800),ipv4(src=10.244.0.0/255.255.255.0,frag=no), packets:0, bytes:0, used:never, actions:ct(commit,zone=

trozet

---
apiVersion: v1
kind: Pod
metadata:
  name: client
  labels:
    pod-name: client
    role: webserver
      #app: spk-coredns  
spec:

trozet

[root@ovn-control-plane ~]# ovs-appctl ofproto/trace --ct-next trk,rpl --ct-next trk,rpl br-int in_port=3,tun_id=16711683,tun_metadata0=262147,dl_src=0a:58:2b:22:eb:86,dl_dst=0a:58:92:3f:71:e5,tcp6,tp_src=8080,tp_dst=43434,ipv6_dst=fd00:10:244:1::7,ipv6_src=fc00:f853:ccd:e793::4,nw_ttl=254 | ovn-detrace
Flow: tcp6,tun_id=0xff0003,in_port=3,vlan_tci=0x0000,dl_src=0a:58:2b:22:eb:86,dl_dst=0a:58:92:3f:71:e5,ipv6_src=fc00:f853:ccd:e793::4,ipv6_dst=fd00:10:244:1::7,ipv6_label=0x00000,nw_tos=0,nw_ecn=0,nw_ttl=254,nw_frag=no,tp_src=8080,tp_dst=43434,tcp_flags=0

bridge("br-int")
----------------
0. in_port=3, priority 100
move:NXM_NX_TUN_ID[0..23]->OXM_OF_METADATA[0..23]
-> OXM_OF_METADATA[0..23] is now 0xff0003
move:NXM_NX_TUN_METADATA0[16..30]->NXM_NX_REG14[0..14]
-> NXM_NX_REG14[0..14] is now 0x4

trozet

[root@ovn-control-plane ~]#  ovn-trace  --ct trk,rpl --ct trk,rpl transit_switch  'inport == "tstor-ovn-worker" && eth.src ==0a:58:2b:22:eb:86 && eth.dst==0a:58:92:3f:71:e5 && ip6  && ip.ttl==64 && ip6.src==fc00:f853:ccd:e793::4 && ip6.dst==fd00:10:244:1::7 && tcp && tcp.src == 8080 && tcp.dst ==43434'
2024-04-17T17:08:05Z|00001|ovntrace|WARN|ct.new && !ct.rel && ip6 && ip6.dst == ^NODEIP_IPv6_0 && tcp && tcp.dst == 30926: parsing expression failed (Syntax error at `NODEIP_IPv6_0' expecting constant.)
2024-04-17T17:08:05Z|00002|ovntrace|WARN|ct.new && ip6.dst == ^NODEIP_IPv6_0 && tcp.dst == 30926: parsing expression failed (Syntax error at `NODEIP_IPv6_0' expecting constant.)
2024-04-17T17:08:05Z|00003|ovntrace|WARN|reg0[2] == 1 && ip6.dst == ^NODEIP_IPv6_0 && tcp.dst == 30926: parsing expression failed (Syntax error at `NODEIP_IPv6_0' expecting constant.)
2024-04-17T17:08:05Z|00004|ovntrace|WARN|ip && ip6.dst == ^NODEIP_IPv6_0: parsing expression failed (Syntax error at `NODEIP_IPv6_0' expecting constant.)
#

trozet

### ebpf
#include <linux/bpf.h>
#include <linux/if_ether.h>
#include <linux/ip.h>
#include <linux/in.h>
#include <linux/tcp.h>

struct data_t {
    __u32 src_ip;
    __u32 dst_ip;

trozet

[trozet@fedora 250-nodetracker]$ cat must-gather.local.3877710284032856534/inspect.local.4295100335563262757/namespaces/openshift-ovn-kubernetes/pods/ovnkube-node-5ksf5/ovnkube-controller/ovnkube-controller/logs/current.log | grep -E 'Starting controller|router changed, syncing services| Node tracker sync took|Full service sync requested'
2024-01-16T02:33:18.213324326Z I0116 02:33:18.213311    3961 services_controller.go:163] Starting controller ovn-lb-controller
2024-01-16T02:33:18.213864591Z I0116 02:33:18.213856    3961 node_tracker.go:185] Node ip-10-0-236-9.us-west-2.compute.internal switch + router changed, syncing services
2024-01-16T02:33:18.214366306Z I0116 02:33:18.214357    3961 services_controller.go:513] Full service sync requested
2024-01-16T02:33:18.238191487Z I0116 02:33:18.238179    3961 node_tracker.go:185] Node ip-10-0-149-39.us-west-2.compute.internal switch + router changed, syncing services
2024-01-16T02:33:18.238242960Z I0116 02:33:18.238234    3961 services_controller.go:513] Full serv

trozet

pkt received on worker node:

01:30:41.176263   M 00:07:35:c0:23:cd ethertype IPv6 (0x86dd), length 88: (flowlabel 0x8e949, hlim 255, next-header ICMPv6 (58) payload length: 32) fe80::207:35ff:fec0:23cd > ff02::1: [icmp6 sum ok] ICMP6, neighbor advertisement, length 32, tgt is fd2e:6f44:5dd8:c956::18, Flags [override]
          destination link-address option (2), length 8 (1): 00:07:35:c0:23:cd

datapath flow:
recirc_id(0xd),in_port(1),ct_state(-new-est-rel+trk),ct_mark(0),eth(src=00:07:35:c0:23:cd,dst=33:33:00:00:00:01),eth_type(0x86dd),ipv6(src=fe80::207:35ff:fec0:23cd,dst=ff02::1,proto=58,hlimit=255,frag=no),icmpv6(type=136,code=0), packets:27611, bytes:2374546, used:0.002s, actions:2,check_pkt_len(size=1414,gt(sample(sample=100.0%,actions(meter(3),userspace(pid=4294967295,controller(reason=1,dont_send=0,continuation=0,recirc_id=25194,rule_cookie=0x25862262,controller_id=0,max_len=65535))))),le(drop)


mac is not changing: 

trozet

#### setup, client curling a service with session affinity that is backed by server and server-sdn pods


[trozet@fedora test]$ oc get service
NAME          TYPE           CLUSTER-IP       EXTERNAL-IP                            PORT(S)           AGE
kubernetes    ClusterIP      172.30.0.1       <none>                                 443/TCP           46m
my-service1   ClusterIP      172.30.189.139   <none>                                 1337/UDP,80/TCP   5m5s
openshift     ExternalName   <none>           kubernetes.default.svc.cluster.local   <none>            41m

trozet

## NP with port range

---
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
  name: default-deny-egress
spec:
  podSelector:
  policyTypes:

trozet

0. Use ovn-k master leader pod for the following process

1. Install dependencies (gdb)
In order to be able to generate a core file of the ovnkube-master process, you will need the gcore binary from the gdb package. The easiest way to get this package is by spawning a toolbox on the node in question and by installing gdb:
# toolbox
# yum install gdb -y
One may have to change the toolbox image (https://docs.openshift.com/container-platform/4.10/support/gathering-cluster-data.html?extIdCarryOver=true&sc_cid=701f2000001OH7JAAW#starting-an-alternative-image-with-toolbox_gathering-cluster-data)

2. Generating a gcore of the current running ovnkube-master process