switch to use hostnames like nova breaks upgrades of l3-agent

Bug #1236439 reported by James Page
32
This bug affects 7 people
Affects Status Importance Assigned to Milestone
Release Notes for Ubuntu
Fix Released
Undecided
James Page
Ubuntu Cloud Archive
Won't Fix
High
Unassigned
neutron
Won't Fix
Medium
Mark McClain
neutron (Ubuntu)
Won't Fix
High
James Page
Saucy
Won't Fix
High
James Page

Bug Description

Commit https://github.com/openstack/neutron/commit/140029ebd006c116ee684890dd70e13b7fc478ec switch to using socket.gethostname() for the name of neutron agents; this has the unfortunate side effect with the l3-agent that all router services are no longer scheduled on an active agent, resulting in floating ip and access outages.

Looks like this will effect upgrades from grizzly->havana as well:

ubuntu@churel:/etc/maas$ quantum agent-list
+--------------------------------------+--------------------+--------------------------+-------+----------------+
| id | agent_type | host | alive | admin_state_up |
+--------------------------------------+--------------------+--------------------------+-------+----------------+
| 02ad1175-209c-4125-889a-e390a15ecd50 | Open vSwitch agent | caipora.1ss.qa.lexington | xxx | True |
| 191d4757-05f6-4170-a78d-d6a3c1b9265e | Open vSwitch agent | canaima | :-) | True |
| 306cbfbb-8879-4d64-ac26-db007f9113a9 | DHCP agent | cofgod.1ss.qa.lexington | xxx | True |
| 32081821-1e94-4274-993b-b0bf2714e5ac | Open vSwitch agent | ciguapa.1ss.qa.lexington | xxx | True |
| 5697a23a-712e-4de3-a218-2a6c177bf555 | Open vSwitch agent | chakora | :-) | True |
| 5ea5e207-1da0-47e3-9a7e-984589b11300 | Open vSwitch agent | cuegle.1ss.qa.lexington | xxx | True |
| 71e31354-76e7-4640-9a5b-368678bc22d0 | Open vSwitch agent | canaima.1ss.qa.lexington | xxx | True |
| 7267e3d2-d9bf-4e57-8d19-803aab636f36 | Open vSwitch agent | chakora.1ss.qa.lexington | xxx | True |
| 75ff2563-f5a5-4df3-aa19-fe8310146c10 | Open vSwitch agent | cuegle | :-) | True |
| 875de52e-d6c3-4e82-8cbd-269831ff00bc | Open vSwitch agent | cofgod | :-) | True |
| 9afaf6f2-2756-4863-b5d0-7faba502e878 | L3 agent | cofgod | :-) | True |
| a81ac370-a318-42e4-9279-eef2b6141644 | Open vSwitch agent | cofgod.1ss.qa.lexington | xxx | True |
| d6e6332e-822a-438e-8613-16013da825e0 | L3 agent | cofgod.1ss.qa.lexington | xxx | True |
| d9712755-03b3-4326-99c1-3bf66c878dc6 | Open vSwitch agent | ciguapa | :-) | True |
| dadf284c-ac8f-4dc1-9ba4-73182e5f1911 | DHCP agent | cofgod | :-) | True |
| ed07ff1a-dcca-4bbd-b026-1296bb90f89b | Open vSwitch agent | caipora | :-) | True |
+--------------------------------------+--------------------+--------------------------+-------+----------------+

James Page (james-page)
Changed in neutron (Ubuntu Saucy):
importance: Undecided → High
description: updated
James Page (james-page)
summary: - switch to use not fully qualified hostnames breaks upgrades of l3-agent
+ switch to use hostnames live nova breaks upgrades of l3-agent
summary: - switch to use hostnames live nova breaks upgrades of l3-agent
+ switch to use hostnames like nova breaks upgrades of l3-agent
James Page (james-page)
tags: added: havana-rc-potential
Changed in neutron:
status: New → Triaged
importance: Undecided → Critical
Changed in neutron:
assignee: nobody → Mark McClain (markmcclain)
James Page (james-page)
Changed in neutron (Ubuntu Saucy):
assignee: nobody → James Page (james-page)
James Page (james-page)
Changed in ubuntu-release-notes:
assignee: nobody → James Page (james-page)
James Page (james-page)
Changed in ubuntu-release-notes:
status: New → Fix Released
James Page (james-page)
Changed in neutron (Ubuntu Saucy):
status: New → Triaged
Thierry Carrez (ttx)
tags: added: havana-backport-potential
removed: havana-rc-potential
Revision history for this message
James Page (james-page) wrote :

@Mark

I note that this issue was included in the Havana release notes. Was there a plan to provide a helper for upgrader's to move services between renamed agents and tidy?

Revision history for this message
Aaron Rosen (arosen) wrote :

I don't think this issue will be that simple to fix. In order to provide a helper script we would need to have the script access all of the machines to determine their hostname and fqdn.

If one wants to avoid this one could write a script to gather this data and then update the db with this data. That said I think it will be hard for us to provide a shrink warp script to solve this issue as unfortunate as it is. Also fwiw on all of my machines my hostnames just happen to be the same as the fqdn.

If someone does want to work on this though I definitely would support that effort.

Changed in neutron:
importance: Critical → High
James Page (james-page)
Changed in cloud-archive:
status: New → Triaged
importance: Undecided → High
Revision history for this message
yong sheng gong (gongysh) wrote :

I think we should close this bug now.

Revision history for this message
Eugene Nikanorov (enikanorov) wrote :

Decreasing importance and marking as incomplete as 2 releases has passed since the change was introduced.

Changed in neutron:
status: Triaged → Incomplete
importance: High → Medium
Revision history for this message
Rolf Leggewie (r0lf) wrote :

saucy has seen the end of its life and is no longer receiving any updates. Marking the saucy task for this ticket as "Won't Fix".

Changed in neutron (Ubuntu Saucy):
status: Triaged → Won't Fix
Revision history for this message
James Page (james-page) wrote :

As we've had no further bug reports about this feature, marking 'Won't Fix' for Ubuntu and the UCA

Changed in neutron (Ubuntu):
status: Triaged → Won't Fix
Changed in cloud-archive:
status: Triaged → Won't Fix
Changed in neutron:
status: Incomplete → Won't Fix
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.