Description of problem: Error - dropped over-mtu packet: 1500 > 1450 when accessing VIP created by LBaaS. The network interfaces in qlbaas namespaces uses 1500 as their default MTU setting. This causes trouble when communicating with instances that have different MTU set (the default vlaue for MTU in vxlan tenant network is 1450) There should be a setting in LBaas config files that allow to change the default MTU size to something else like 1540 or 8950 (in case of jumbo frames) Also if the value is not set it should be referencing the value set up in other part of Neutron instead of defaulting to 1500. Version-Release number of selected component (if applicable): Encountered on RH OSP 9 - LbaaS v1. However I believe this affects all version of OSP and LBaaS V1 and V2 How reproducible: Every time when accessing content that exceeds 1400 bytes via LBaaS VIP. Tested with curl Steps to Reproduce: 1. Configure RH OSP with LBaaS 2. Define a Pool of webservers that share content over 1400 bytes and assign VIP to it 3. curl the content of the vip to local machine Actual results: curl never finishes downloading the content. /var/log/messages receive error - dropped over-mtu packet: 1500 > 1450 Expected results: curl finishes with the content of the page Additional info: We were able to work this around by manually changing the mtu setting to lbaas port: for i in `ip netns | grep lbaas`; do echo $i; ip netns exec $i ip a; done ip netns exec qlbaas-82c190b6-538d-4e02-a217-5d7277c8cbbc ip link set tapfd14b090-36 mtu 1450 There are some abandoned bugs out there describing the same behavior: https://bugs.launchpad.net/neutron/+bug/1376446 https://review.openstack.org/#/c/268321/
Ihar, can you work with Nir to figure out if this is relevant for LBaaS v2 with haproxy and Octavia, and if so, send out some fixes?
We will need to backport https://review.openstack.org/#/q/I6c29cf8dd4164d2024b64a86108b1e22a66f966a I am moving the bug to Nir since he is our LBaaS guy.
Verified : [root@controller-0 ~]# rpm -qa |grep neutron openstack-neutron-bigswitch-agent-2015.3.8-1.el7ost.noarch python-neutron-lbaas-8.1.0-2.el7ost.noarch openstack-neutron-openvswitch-8.1.2-14.el7ost.noarch python-neutron-8.1.2-14.el7ost.noarch openstack-neutron-8.1.2-14.el7ost.noarch openstack-neutron-ml2-8.1.2-14.el7ost.noarch python-neutronclient-4.1.1-2.el7ost.noarch python-neutron-lib-0.0.2-1.el7ost.noarch openstack-neutron-bigswitch-lldp-2015.3.8-1.el7ost.noarch openstack-neutron-common-8.1.2-14.el7ost.noarch openstack-neutron-lbaas-8.1.0-2.el7ost.noarch openstack-neutron-metering-agent-8.1.2-14.el7ost.noarch
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://rhn.redhat.com/errata/RHBA-2017-0232.html