[Openstack-operators] Experimenting with DHCP Agent redundancy

Alvise Dorigo alvise.dorigo at pd.infn.it
Wed Mar 12 09:42:50 UTC 2014


Hi,
I've installed 2 Network nodes running only neutron agents (the neutron server is located in the controller node).
I see this:

# neutron agent-list
+--------------------------------------+--------------------+-----------------------------+-------+----------------+
| id                                   | agent_type         | host                        | alive | admin_state_up |
+--------------------------------------+--------------------+-----------------------------+-------+----------------+
| 215324e6-f19a-41ce-b967-40089388f9ed | Open vSwitch agent | network-02.cloud.pd.infn.it | :-)   | True           |
| 55d8c998-2557-4cac-abbc-4a85d9f190a0 | DHCP agent         | network-01.cloud.pd.infn.it | :-)   | True           |
| 5c03007b-6b02-4260-9f5c-7f1e951cf1b0 | Open vSwitch agent | compute.cloud.pd.infn.it    | :-)   | True           |
| 97ec9951-5c7b-45e0-9de6-40d4efeccd6b | L3 agent           | network-01.cloud.pd.infn.it | :-)   | True           |
| c96ab5a0-2ef4-4f44-90bb-686bbaef4f55 | DHCP agent         | network-02.cloud.pd.infn.it | :-)   | True           |
| cbe4e6bc-0dda-46df-b170-93f6d324a0a5 | Open vSwitch agent | network-01.cloud.pd.infn.it | :-)   | True           |
| dedc11e7-c1c5-4db3-be90-d218f210a00a | L3 agent           | network-02.cloud.pd.infn.it | :-)   | True           |
+--------------------------------------+--------------------+-----------------------------+-------+----------------+

I've created 2 networks (one is external -"esterna"-, the other one is not - "interna"-):

# neutron net-list
+--------------------------------------+---------+------------------------------------------------------+
| id                                   | name    | subnets                                              |
+--------------------------------------+---------+------------------------------------------------------+
| 6cd46839-c434-42e2-9b9e-a3b2a7fae470 | esterna | 08c33f86-c34a-4fa3-8295-5f9866d9527d 192.135.16.0/24 |
| e1cd28e4-83da-4dc8-bae8-f50600baeff9 | interna | 7ec2317c-40c5-4daf-9997-1242fd4f78b0 192.168.70.0/24 |
+--------------------------------------+---------+------------------------------------------------------+

I've added the net "interna" to both dhcp-agents:

# neutron dhcp-agent-list-hosting-net interna
+--------------------------------------+-----------------------------+----------------+-------+
| id                                   | host                        | admin_state_up | alive |
+--------------------------------------+-----------------------------+----------------+-------+
| 55d8c998-2557-4cac-abbc-4a85d9f190a0 | network-01.cloud.pd.infn.it | True           | :-)   |
| c96ab5a0-2ef4-4f44-90bb-686bbaef4f55 | network-02.cloud.pd.infn.it | True           | :-)   |
+--------------------------------------+-----------------------------+----------------+-------+

VM creation goes smoothly well (I'm using the cirros 0.3.1).

Now, if I disable one of the two dhcp agents (using the command "neutron agent-update 55d8c998-2557-4cac-abbc-4a85d9f190a0 --admin_state_up=false") I'd expect that the second dhcp agent can take care of the (new) VMs asking for dhcp address... 
I disabled the first (network-01) dhcp agent.
If I create a new VM it will never get its IP address. So it seems to me that the second dhcp agent (running on network-02) is not actually doing its job... 
If I re-enable the first dhcp agent, new VMs get their IPs.

What could I have missed in the procedure ?

thanks,

  Alvise
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 1756 bytes
Desc: not available
URL: <http://lists.openstack.org/pipermail/openstack-operators/attachments/20140312/108837ff/attachment.bin>


More information about the OpenStack-operators mailing list