)]}'
{"/PATCHSET_LEVEL":[{"author":{"_account_id":10342,"name":"Jay Faulkner","display_name":"JayF","email":"jay@jvf.cc","username":"JayF","status":"youtube.com/@oss-gr / podcast.gr-oss.io"},"change_message_id":"0c0a474e2cd9560a0734c5bd3ae4872cd71b5ae0","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":3,"id":"3a7ad55f_8e1372e8","updated":"2024-09-12 12:41:16.000000000","message":"this is a needed CI fix which has been discussed significantly in the qa channel. landing as a single core","commit_id":"510f87a033ce5f76a7aa881f56b2bd9958c8582f"}],"devstack/settings":[{"author":{"_account_id":4146,"name":"Clark Boylan","email":"cboylan@sapwetik.org","username":"cboylan"},"change_message_id":"565cd75b66e5846f1654257f98408bde64557365","unresolved":true,"context_lines":[{"line_number":24,"context_line":"# 50 bytes is overhead for vxlan (which is greater than GRE"},{"line_number":25,"context_line":"# allowing us to use either overlay option with this MTU."},{"line_number":26,"context_line":"# However, if traffic is flowing over IPv6 tunnels, then"},{"line_number":27,"context_line":"# The overhead is essentially another 100 bytes. In order to"},{"line_number":28,"context_line":"# handle both cases, lets go ahead and drop the maximum by"},{"line_number":29,"context_line":"# 100 bytes (but not too low to avoid other problems)."},{"line_number":30,"context_line":"PUBLIC_BRIDGE_MTU\u003d${OVERRIDE_PUBLIC_BRIDGE_MTU:-$((local_mtu - 100))}"}],"source_content_type":"application/octet-stream","patch_set":1,"id":"47e121e4_be144395","line":27,"updated":"2024-09-11 17:31:35.000000000","message":"According to haleyb in discussion over in #openstack-qa geneve + ipv6 overhead is 78 bytes. vxlan + ipv4 is 50 bytes. vxlan + ipv6 is 70 bytes. I think the vlxan + ipv6 70 byte overhead is why we end up with 1372 in the log below since the interface starts with a 1442 mtu:\n\n```\n++ /opt/stack/ironic/devstack/settings:source:6 :   PUBLIC_BRIDGE_MTU\u003d1372\n+++ /opt/stack/ironic/devstack/settings:source:12 :   sort -n\n+++ /opt/stack/ironic/devstack/settings:source:12 :   ip link show\n+++ /opt/stack/ironic/devstack/settings:source:12 :   head -1\n+++ /opt/stack/ironic/devstack/settings:source:12 :   sed -ne \u0027s/.*mtu \\([0-9]\\+\\).*/\\1/p\u0027\n++ /opt/stack/ironic/devstack/settings:source:12 :   local_mtu\u003d1442\n++ /opt/stack/ironic/devstack/settings:source:20 :   \u0027[\u0027 1442 -gt 1372 \u0027]\u0027\n++ /opt/stack/ironic/devstack/settings:source:21 :   local_mtu\u003d1372\n++ /opt/stack/ironic/devstack/settings:source:30 :   PUBLIC_BRIDGE_MTU\u003d1272\n```\n\nAll that to say we can probably adjust this 100 bytes down to 78 bytes and still cover the ipv6 + geneve case? Then we\u0027d end up with 1442 - 70 (for vxlan) - 78 (for vxlan or geneve) \u003d 1294 \u003e 1280 which should avoid the ipv6 min mtu of 1280 problem.\n\nThen you can decide if you still want to set the min mtu to 1280 at the end. Maybe add a comment here that 1280 is the minimum you can configure with ipv6 on an interface as a breakcrumb if that does break in the future?","commit_id":"1fdf5f6dbe6b3159ce7e84767fca432a80c6e01e"},{"author":{"_account_id":1131,"name":"Brian Haley","email":"haleyb.dev@gmail.com","username":"brian-haley"},"change_message_id":"f0bca14db08114663d0084d9f50904732a9ce15e","unresolved":true,"context_lines":[{"line_number":24,"context_line":"# 50 bytes is overhead for vxlan (which is greater than GRE"},{"line_number":25,"context_line":"# allowing us to use either overlay option with this MTU."},{"line_number":26,"context_line":"# However, if traffic is flowing over IPv6 tunnels, then"},{"line_number":27,"context_line":"# The overhead is essentially another 100 bytes. In order to"},{"line_number":28,"context_line":"# handle both cases, lets go ahead and drop the maximum by"},{"line_number":29,"context_line":"# 100 bytes (but not too low to avoid other problems)."},{"line_number":30,"context_line":"PUBLIC_BRIDGE_MTU\u003d${OVERRIDE_PUBLIC_BRIDGE_MTU:-$((local_mtu - 100))}"}],"source_content_type":"application/octet-stream","patch_set":1,"id":"a078952b_062a2e84","line":27,"in_reply_to":"47e121e4_be144395","updated":"2024-09-11 19:28:09.000000000","message":"I would +1 changing to 78, the only thing it doesn\u0027t support is an IPv6 underlay (if necessary) since that would require another 20 bytes. If I did the math right.","commit_id":"1fdf5f6dbe6b3159ce7e84767fca432a80c6e01e"}]}