I was pointed about https://github.com/openshift/cluster-network-operator/pull/697 . This prevents from doing upgrades, as the hybrid overlay is not executed. Need to monitor the PR, per Dan Williams this will only land for 4.5.1 .
Fix landed on 4.5.2 and performed an upgrade to make sure it works: [ricky@localhost openshift-installer]$ oc get clusterversion NAME VERSION AVAILABLE PROGRESSING SINCE STATUS version 4.4.13 True True 46m Working towards 4.5.2: 87% complete [ricky@localhost openshift-installer]$ oc get clusterversion NAME VERSION AVAILABLE PROGRESSING SINCE STATUS version 4.5.2 True False 7m54s Cluster version is 4.5.2 [ricky@localhost openshift-installer]$ oc -n openshift-ovn-kubernetes rsh -c ovnkube-node ovnkube-node-d8zqc ovs-ofctl dump-flows br-ext | head -n10 cookie=0x0, duration=16.513s, table=0, n_packets=0, n_bytes=0, priority=100,arp,in_port=ext,arp_tpa=10.130.0.3,arp_op=1 actions=move:NXM_OF_ETH_SRC[]->NXM_OF_ETH_DST[],mod_dl_src:0a:58:0a:82:00:03,load:0x2->NXM_OF_ARP_OP[],move:NXM_NX_ARP_SHA[]->NXM_NX_ARP_THA[],move:NXM_OF_ARP_SPA[]->NXM_OF_ARP_TPA[],load:0xa580a820003->NXM_NX_ARP_SHA[],load:0xa820003->NXM_OF_ARP_SPA[],IN_PORT,resubmit(,1) cookie=0x0, duration=1466.149s, table=0, n_packets=0, n_bytes=0, priority=100,ip,in_port="ext-vxlan",dl_dst=0a:58:0a:82:00:03,nw_dst=10.130.0.0/23 actions=resubmit(,10) cookie=0x0, duration=1466.149s, table=0, n_packets=0, n_bytes=0, priority=10,arp,in_port="ext-vxlan",arp_tpa=10.130.0.0/23,arp_op=1 actions=resubmit(,2) cookie=0x0, duration=1466.149s, table=0, n_packets=2, n_bytes=240, priority=0 actions=drop cookie=0x0, duration=1466.149s, table=1, n_packets=0, n_bytes=0, priority=0 actions=drop cookie=0x0, duration=16.513s, table=2, n_packets=0, n_bytes=0, priority=100,arp,in_port="ext-vxlan",arp_tpa=10.130.0.0/23,arp_op=1 actions=move:NXM_NX_TUN_IPV4_SRC[]->NXM_NX_TUN_IPV4_DST[],load:0x1001->NXM_NX_TUN_ID[0..31],move:NXM_OF_ETH_SRC[]->NXM_OF_ETH_DST[],mod_dl_src:0a:58:0a:82:00:03,load:0x2->NXM_OF_ARP_OP[],move:NXM_NX_ARP_SHA[]->NXM_NX_ARP_THA[],load:0xa580a820003->NXM_NX_ARP_SHA[],move:NXM_OF_ARP_TPA[]->NXM_NX_REG0[],move:NXM_OF_ARP_SPA[]->NXM_OF_ARP_TPA[],move:NXM_NX_REG0[]->NXM_OF_ARP_SPA[],IN_PORT cookie=0x0, duration=1466.149s, table=2, n_packets=0, n_bytes=0, priority=0 actions=drop cookie=0xa820008, duration=1465.979s, table=10, n_packets=0, n_bytes=0, priority=100,ip,nw_dst=10.130.0.8 actions=mod_dl_src:0a:58:0a:82:00:03,mod_dl_dst:ea:58:76:82:00:09,output:ext cookie=0xa820004, duration=1465.979s, table=10, n_packets=0, n_bytes=0, priority=100,ip,nw_dst=10.130.0.4 actions=mod_dl_src:0a:58:0a:82:00:03,mod_dl_dst:ea:58:76:82:00:05,output:ext cookie=0xa820006, duration=1465.706s, table=10, n_packets=0, n_bytes=0, priority=100,ip,nw_dst=10.130.0.6 actions=mod_dl_src:0a:58:0a:82:00:03,mod_dl_dst:ea:58:76:82:00:07,output:ext
I'd say it's fixed. The bug description was about br-ext not having any flows after upgrade, which was caused by a broken daemonset script that didn't run hybrid overlay options. I verified the flows are there after upgrade now (fix is on 4.5.2), but I think Yurii wanted to do more testing on F5 to verify the traffic was going out.
Unassigning as I'm on long vacation.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (OpenShift Container Platform 4.5.5 bug fix update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2020:3188
The needinfo request[s] on this closed bug have been removed as they have been unresolved for 1000 days