Bug 1608951
| Summary: | Octavia - floating IP not working with Octavia + OVN + DVR | ||||||
|---|---|---|---|---|---|---|---|
| Product: | Red Hat OpenStack | Reporter: | broskos | ||||
| Component: | python-networking-ovn | Assignee: | Daniel Alvarez Sanchez <dalvarez> | ||||
| Status: | CLOSED ERRATA | QA Contact: | Alexander Stafeyev <astafeye> | ||||
| Severity: | high | Docs Contact: | |||||
| Priority: | urgent | ||||||
| Version: | 13.0 (Queens) | CC: | akaris, amuller, apevec, astafeye, cgoncalves, dalvarez, dmaley, dpeacock, dprince, dsneddon, ekuris, emacchi, gcharot, ihrachys, jamsmith, jsaucier, lhh, lpeer, madgupta, majopela, mhernon, nyechiel, pgrist, pmorey, rlopez, rzaleski, swogatpradhan22, yoliynyk | ||||
| Target Milestone: | z3 | Keywords: | Triaged, ZStream | ||||
| Target Release: | 13.0 (Queens) | ||||||
| Hardware: | x86_64 | ||||||
| OS: | Linux | ||||||
| Whiteboard: | |||||||
| Fixed In Version: | python-networking-ovn-4.0.2-3.el7ost | Doc Type: | If docs needed, set a value | ||||
| Doc Text: | Story Points: | --- | |||||
| Clone Of: | |||||||
| : | 1627838 (view as bug list) | Environment: | |||||
| Last Closed: | 2018-11-13 23:32:56 UTC | Type: | Bug | ||||
| Regression: | --- | Mount Type: | --- | ||||
| Documentation: | --- | CRM: | |||||
| Verified Versions: | Category: | --- | |||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||
| Embargoed: | |||||||
| Bug Depends On: | |||||||
| Bug Blocks: | 1547478, 1627838 | ||||||
| Attachments: |
|
||||||
Created attachment 1470799 [details]
sos report from compute node hosting all instances and lbs
Thank you for bug report and for attaching sos reports. This issue seems to be very similar to rhbz #1608343. Networking-ovn is setting external_mac on the NAT entry for the floating IP associated to unbound Neutron ports. Unsetting external_mac makes it work for the user to reach the load balancer. A possible solution (verified on an OSP13 environment) is to fallback to non-DVR behavior in such cases (FIPs associated to unbound Neutron ports). We also confirmed that one can reach a load balancer via a floating IP on Octavia/OVN-non DVR environments. Thanks a lot to Daniel and Lucas for troubleshooting and brainstorming on possible solutions! Please see this: Reference implentation has similar (not the same, I believe) issues. https://bugs.launchpad.net/neutron/+bug/1774459 They have a discussion point on the PTG, On wednesday 2:15 - 3:30 L3 topics Update permanent ARP entries for allowed_address_pair IPs in DVR Routers (swami / haleyb) Moving it back to ON_DEV until it gets merged upstream. Until bug #1630929 gets in, to test this configuration THT env files must be ordered properly (with the OVN one last) Look on Bernard comment 51 https://bugzilla.redhat.com/show_bug.cgi?id=1608951#c51 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, 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-2018:3614 I am currently hitting this issue, Openstack wallaby | centos 8 My floating ip attached to octavia lb vip is not working. Unable to browse or unable to telnet. Please suggest a solution or workaround if you have any. Wallaby + OVN DVR |
Description of problem: I've deployed Octavia + OVN + DVR and floating IPs assigned to Octavia VIPs are not reachable. Other floating IPs tied to nova instances are working fine. The octavia vip is reachable and working fine from inside on the private network, however the Floating IP is not reachable at all. Checked security groups for tenant, they are fine and allow port 80 - access via internal address to instance port 80 returns page as expected. access via floating ip assigned to instance returns http page as expected. Found LB security group in service tenant, confirmed in has ingress rule for tcp/80 from 0.0.0.0/0 Director deployment using: -e /usr/share/openstack-tripleo-heat-templates/environments/services-docker/octavia.yaml \ -e /usr/share/openstack-tripleo-heat-templates/environments/services-docker/neutron-ovn-dvr-ha.yaml \ Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: create 2 test webservers, configure fake healthcheck page at /healthcheck openstack loadbalancer create --name lb1 --vip-subnet-id admin-subnet openstack loadbalancer listener create --name listener1 --protocol HTTP --protocol-port 80 lb1 openstack loadbalancer pool create --name pool1 --lb-algorithm ROUND_ROBIN --listener listener1 --protocol HTTP openstack loadbalancer healthmonitor create --delay 5 --max-retries 4 --timeout 10 --type HTTP --url-path /healthcheck pool1 openstack loadbalancer member create --subnet-id admin-subnet --address 192.168.99.6 --protocol-port 80 pool1 openstack loadbalancer member create --subnet-id admin-subnet --address 192.168.99.13 --protocol-port 80 pool1 Now attach floating IP Actual results: website is not available via floating ip Expected results: website should be available via floating ip Additional info: