chalans's profile - activity

2020-07-04 17:57:26 -0600 commented question How can i rebalance my network nodes

Hello, I think you can do it with openstack network agent add router and network agent agent remove router A shell script can do the job

2020-06-03 13:20:57 -0600 answered a question separate storage node

To complete the answers :


LVM use ISCSI portals on storage Node usually with DAS Storage . Then Cinder expose it over bloc mode on instance.


If you have already a storage Array, better to use the constructor Driver (Netapp, Dell-EMC, Hitachi) for a direct FC access to your Lun => best performance and fiability . LVM was for testing and should not be use in a production environnement .

Best storage with OpenStack is CEPH (my opinion ;)

2020-06-02 17:09:21 -0600 commented question Supported array

Try to use HBSD driver ..

2020-06-02 17:06:30 -0600 received badge  Notable Question (source)
2020-06-01 08:16:02 -0600 commented question Supported array

Your Hitachi array was also old. We use the driver for Hitachi vSP Gx00 actually (G1000 array) and it works on Queens/Rocky and Stein

2020-05-29 16:11:18 -0600 commented answer Openstack on KVM

Not sure if i undestrand correctly your question but you can virtualize Network Node but not Compute Node

2020-05-28 11:01:47 -0600 answered a question Openstack on KVM

Hello, For information we used the following configuration for Production environnement :

  • 3 CTRL VMs for API/Keystone/Horizon
  • 3 MON VMs for ELK/Kibana/Monitoring
  • 3 BDD/RabbitMq VMs
  • 2 physicals servers for Network Node
  • 3 or more compute node physical Server (for this never use VM in a production environnement)
2020-05-27 16:27:35 -0600 commented question Cannot launch instances anymore due to RabbitMQ errors

What about the status of the cluster : rabbitmqctl cluster_status ? And queue list rabbitmqctl list_queue ?

2020-05-26 15:19:05 -0600 answered a question user to availability zone mapping

Hello,

Maybe you can use HostAggregate for this, so you can set the provisionning on specific host on specific AZ

Greetings,

Yes, Create your Aggregate as Admin. https://docs.openstack.org/nova/lates...

openstack  aggregate create --property key=value1  agg1_name
openstack aggregate add host agg1_name Host1

Then set the key,value on specific flavor for you project

openstack flavor set --property  aggregate_instance_extra_specs:key=value1 my_flavor1

Do the same thing with the other HostAggre and create another Flavor

And after just assign specific Flavor to land your instance on specific Host

Ah dont forget to set this line on nova.conf (append to your current configuration)

[filter_scheduler]
enabled_filters=AggregateInstanceExtraSpecsFilter,AvailabilityZoneFilter,RamFilter,ServerGroupAntiAffinityFilter,ServerGroupAffinityFilte
2020-05-26 01:08:36 -0600 received badge  Popular Question (source)
2020-05-25 17:25:52 -0600 asked a question Octavia how to recreate deleted Amphora

Hello,

OpenStack Stein version with Octavia with amphora/haproxy That is a question with an answer but i want to know if you have a better solution than me ;) My issue was the following : After a network outage some amphora lost their IPs. One of the solution was to stop/start them, but bad idea the instance was deleted ... So we are in the situation that the LoadBalancer is active but no amphora When we do an

openstack loadbalancer amphora list | grep {id_Lb}

=> nothing return ... It seems there are no command to retrieve , the only way was to go in database Octavia . Type

select * from load_balancer where id = 'Id_lb' ;
select * from amphora where load_balancer_id = 'id_lb' ;

That should return lines with id of amphora with status DELETED like this

| e8df1c33-566c-4ee0-91e0-35392d0f7e23 | 5a67ff45-b384-4e41-8eab-e3ccff6aebaf | DELETED | 49ab80bc-50a7-4c86-a173-1e25b3b94165 | 10.x.xxx | 10.x.xxx | 10.x.xxx | e55e68cb-4f55-4a96-a750-922a20502819 | dd00d6f6-09e2-4ecd-bf30-41a5c9fb5c30 | STANDALONE | 2022-05-04 08:05:31 |         0 | NULL           |       1 |          NULL | nova        | 20231 | 20:21 | 33b0ec25-b2d1-4d17-bb51-c1fa50495273 | octavia        |

Retrieve the id of the amphora Type

update amphora set status ='ERROR' where id = 'id_amphora' ;

Set the loadbalancer provisoning to ACTIVE for remove immuable

 update load_balancer set provisioning_status = 'ACTIVE' where id = 'id_lb' ;

The amphora should be present now and you can retrieve the id for a Failover to put it on Allocated status

openstack loadbalancer amphora list | grep id_lb

openstack loadbalancer amphora failover id_amphora

The amphora was recreated with the right information and you LB was ready to serve !

It works but that is a pain to do,we created an ansible playbook, but anyone have a better solution and more clean (i dont like to go in database ) ?

Thank you

Stephane Chalansonnet

2020-02-28 15:57:53 -0600 commented question Restarting neutron l3-agent

Check also on DHCP , we have this error sometime when the process on network node are down. Restart the dhcp service correct the problem. Restart a L3 agent on a network node affect temporaly traffic so do it during the night :)

2020-02-28 15:49:21 -0600 commented question Galera: Recover a complete environment failure

Already happened here , we used the kolla-ansible playbook to recover the cluster . Even if you don't have Kolla, you can follow the playbook : https://gist.github.com/pennycoders/0...

2019-12-07 05:23:41 -0600 received badge  Editor (source)
2019-12-07 05:21:27 -0600 answered a question Stein - Kolla - Neutron router ports are down - TooManyExternalNetworks

Hello,

We have the same problem after an upgrade from Rocky to Stein ... Floating IP port remain down . After waiting for some minutes, the port become Up .. Also with a restart of L3_Agent , the port become up immediatly.

If i create directly on provider network (no vxlan + FIP) i have no problem So you are true the problem was on L3_agent

PS : can you contact me directly on my email adress schalans@gmail.com , maybe we can help us . (if you are french , n'hésite pas à écrire en français ;) )

Greeting

2019-08-23 15:26:40 -0600 commented question What is the rationale behind nova flavors?

Hello, You can use extra properties on flavor => exemple Host-Agregate => cinder storage type etc..

2019-05-23 08:47:06 -0600 received badge  Enthusiast
2019-05-22 16:01:54 -0600 answered a question Howto analyze package loss an ovs br-int br-ext?

Hello,

Maybe you can try to configure sflow on each OVS and send to a collector For our tests we used sFlowTrend (free version limited to 5 device)

You juste need to configure each OVS with this command :

docker exec -it -u 0 neutron_openvswitch_agent ovs-vsctl --id=@sflow create sflow agent=bond_pub.XXX target="\"XXX:6343\"" header=128 sampling=64 polling=10 -- set bridge br-tun sflow=@sflow

2019-05-22 15:41:57 -0600 answered a question kolla-ansible multiple external vlans provider network configuration

Hello,

First you dont need to reconfigure all for this modification . In your deployement node where you have kolla-ansible intalled under the directory /etc/kolla You can create the directory : /etc/kolla/config/neutron create the ml2_conf.ini with these parameters : This configuration assume you have a trunk on your external interface Bond0 (no tag under linux) with vlans 10->200

[ml2_type_vlan]

network_vlan_ranges = physnet1:10:200

[ml2_type_flat]

flat_networks =

Kolla-ansible use all files under config directory for specific configuration. Then you can have a centralised configuration . Juste run (choose the right path for multinode files) kolla-ansible/tools/kolla-ansible -i multinode -t neutron reconfigure

You should have your /etc/kolla/neutron-openvswitch-agent/ml2_conf.ini updated with the new configure

2019-03-13 17:39:03 -0600 commented answer can't ping external gateway or floating IP's from within router namespace

Hi , Your configuration should not be set with a Vxlan Type but Flat . What did you type exactly when you created the network External ?

2019-03-11 17:43:02 -0600 received badge  Teacher (source)
2019-03-10 08:45:41 -0600 answered a question Only VMs on private network not acquiring IP address

I've had the same problem . For me , no port was created on Ovs for the instance , i saw that on the log of the instance (no ethx :().

Check on the ovs log, for my case i had plenty errors and a restart of the Ovs container corrected my problem.

Regard,

2019-03-10 08:45:41 -0600 answered a question can't ping external gateway or floating IP's from within router namespace

Hello,

What is the configuration of you external network ? An openstack network show {Name_of-your-external-network} can help us to determine your configuration .

Regards,

Stef

2019-03-09 18:46:35 -0600 answered a question Neutron network information

Flat or Vlan depending on your physical configuration. Usually we have multiple vlan on a trunk inside your external interfaces so the ml2 provider was configuration on Vlan mode. And yes, that for public network / provider Network for FIP but you can assign directly an public IP on your Instance.

Indeed , Vxlan are for private IP on instance.

Not sure if i answered on your questions ;)

Regards,

Stef