Ask Your Question

Eduardo Gonzalez's profile - activity

2019-01-10 03:01:30 -0600 answered a question [kolla-ansible] Dimensions in globals.yml not applied

Hi, the variable used is incorrect, here is the proper variable:

neutron_openvswitch_agent_dimensions:
    mem_limit: "3g"
    memswap_limit: "4g"
    mem_reservation: "2g"

Note that ansible don't support variables with (-) symbol.

Regards

2019-01-10 02:57:54 -0600 commented question Kolla-ansible -- Error trying to deploy ceph on nodes.

Hi, whats the command used to deploy? Kolla-ansible version (master guess), kolla images version ?

2019-01-04 05:10:27 -0600 answered a question Cannot Deploy kolla-ansible telegraf

Hi, is a bug on this line of the template https://github.com/openstack/kolla-an...

is an easy fix, here is the bug tracking this bug https://bugs.launchpad.net/kolla-ansi...

Thanks for raising it

2019-01-04 05:04:15 -0600 edited question Cannot Deploy kolla-ansible telegraf

Hi! I'm trying to deploy a development opentstack in a virtual environment. I passed de the bootstrap and prechecks, but i got an error in the deployment. i get an error from ansible trying to copy configuration file of telegraf on each node.

TASK [telegraf : Copying over telegraf config file] ************************************************************************************************
failed: [control01] (item=/home/vagrant/kolla-ansible/ansible/roles/telegraf/templates/telegraf.conf.j2) => {"changed": false, "item": "/home/vagrant/kolla-ansible/ansible/roles/telegraf/templates/telegraf.conf.j2", "msg": "AnsibleError: template error while templating string: expected token 'end of print statement', got ':'. String: [global_tags]\n[agent]\n  interval = \"10s\"\n  round_interval = true\n  metric_batch_size = 1000\n  metric_buffer_limit = 10000\..........................

Using ansible 2.7 with python2

Thanks. Mariano

2018-12-19 04:55:21 -0600 commented question unable to get packet from source using networking-sfc

I use echo 1 > /proc/sys/net/ipv4/ip_forward on the vnf with cloud-init instead of changing systctl.conf. May be this?

2018-12-19 04:51:54 -0600 edited question unable to get packet from source using networking-sfc

Hello, I'm doing sfc using networking-sfc right now, and I have some problem about the routing I have three VM in openstack, client, chain-vm and server After creating the port chain with the following command:

openstack sfc flow classifier create \
      --ethertype IPv4 \
      --source-ip-prefix 10.0.0.122/32 \
      --destination-ip-prefix 10.0.0.80/32 \
      --protocol tcp \
      --source-port 51818:51818 \
      --destination-port 8080:8080 \
      --logical-source-port daab6bcc-39c9-4963-802b-f944d3d0f58f \
      --logical-destination-port fc6a92f6-ce28-43ad-801e-471c19e69983 FC1

openstack sfc port pair create --ingress f63a75ff-5fff-4b80-8176-989c11076263 --egress f63a75ff-5fff-4b80-8176-989c11076263  PP1

openstack sfc port pair group create --port-pair PP1 PG1

openstack sfc port chain create --port-pair-group PG1 --flow-classifier FC1 --chain-parameters symmetric=true PC1

from client I use curl --local-port 51818 10.0.0.80:8080, and in chain-vm I can capture the packet from client to server using tcpdump but server didn't receive it, so there's no response I tried to enable the ipv4 routing by setting the net.ipv4.ip_forward=1 in /etc/sysctl.conf in chain-vm(ubuntu) But still not working, can somebody help me? Thank you.

2018-12-19 04:49:52 -0600 answered a question How to Install Tacker on Openstack?

Hi,

Tacker is MANO service, it will connect to the VIMs registered on it and call the endpoints to create resources on it like instances, alarms, etc. Being this said.

Tacker could be installed standalone outside OpenStack and register any VIM (OpenStack and k8s supported at this moment).

Just need to follow the following steps in https://docs.openstack.org/tacker/latest/install/manual_installation.html. Could be in the same nodes as your controllers or separate nodes, only need access to the APIs and databases.

If want to try standalone or complete OpenStack deployment with tacker, in kolla-ansible we have support to tacker and all other required services. https://docs.openstack.org/kolla-ansible/latest/reference/orchestration-and-nfv/tacker-guide.html

Regards

2018-12-07 02:39:49 -0600 answered a question I want to get involved in OpenStack but don't know what I want to work on.

Hi,

It really depends on what you like and technologies you know.

In example if you like NFV or SDN, may want to find a related project as Tacker, neutron drivers, etc. If you are more from deployment or operation, may want to contribute to deployment projects as Kolla, OSA or infrastructure services as Zuul.

In the other hands depends on your develop knowledge, if in example you are experience with java, Monasca is one project written on it. Most of the others are written in python, and deployment projects with ansible, puppet, chef, etc.

As resume, if I was looking for a project to contribute, I'd start thinking what are my preferences and my knowledge, if you choose a project related to a technology you don't like, may stress you and end up leaving the project.

  • Pro-tip: Smaller projects are easy to contribute and receive feed back from the team, bigger as Neutron and Nova are hardest to get involved.

Hope it help, Regards

2018-11-28 02:47:32 -0600 commented question kolla ansible - instance no internet

8- Modify the public network scheme.

EXT_NET_CIDR='10.10.20.0/24'
EXT_NET_RANGE='start=10.10.20.110,end=10.10.20.254'
EXT_NET_GATEWAY='10.10.20.1'
2018-11-28 02:47:06 -0600 commented question kolla ansible - instance no internet

A correction to the blog used, public interface should be UP but without IP address, in the blog is assigned an IP address. The following config should match to your public IP ranges

2018-11-28 02:44:53 -0600 commented question kolla ansible - instance no internet

Hi, if you tcpdump on enp0s8 traffic appears? If yes, is an issue with physical interfaces or host networks(switches, routers, etc). If not, please share ovs-vsclt show and ofctl-dump flows br-ex within ovs containers

2018-11-19 06:36:38 -0600 commented question Tacker Installation Error

Maybe not enough resources to add more services? Mistral doesn't touch anything on nova/neutron, no reason why should be failing except not enough resources.

2018-10-25 03:28:38 -0600 edited answer How do I support additional backend drivers for Cinder in kolla-ansible?

So, directly in Kolla you could use template_override https://docs.openstack.org/kolla/latest/admin/image-building.html#dockerfile-customisation and use your own Cinder docker images with GPFS driver installed (if it's not in the "usual'' package).

Then add required stanzas in the Cinder config file - and it should work.

But yes - there is no ,"out of the box'' support for GPFS in Kolla-Ansible - but the project would be happy if you could contribute.

2018-09-18 04:19:29 -0600 commented answer Adding compute node to an existing openstack environment

May be an older version, it was introduced in queens if i recall correctly

2018-08-23 02:33:36 -0600 commented question Kolla-Ansible - Can't add compute node using ansible-playbook --limit

eranlevin, could you try with ansible 2.5, maybe something changed in ansible which makes limit not work as expected, with 2.5 for sure it works

2018-08-15 16:09:11 -0600 received badge  Nice Answer (source)
2018-08-13 03:04:51 -0600 commented answer Unknown Router Interface in DVR Mode

No, traffic goes through the q-router, check the traffic flow https://docs.openstack.org/liberty/ne... section "North/south for instances with a fixed IP address"

2018-08-10 02:11:28 -0600 commented answer Unknown Router Interface in DVR Mode

The rfp- port in the q-router namespace.

Could you share ovs-vsctl show output, paste it to paste.openstack.org please

2018-08-09 01:34:07 -0600 commented answer Unknown Router Interface in DVR Mode

whats the result of openstack port list --router ROUTER_ID and openstack port show PORT_ID

2018-08-08 07:31:03 -0600 answered a question Unknown Router Interface in DVR Mode

Hi,

the extra interface is the FIP (floating IP namespace), this interface makes snat/dnat in the router namespace north-south in compute hosts, the allow the floating IP traffic goes outside directly from compute nodes and not need to flow through neutron nodes.

The interface in qrouter (rfp-*) is the patch link to the FIP namespace (patch fpr).

This is the reason of needing a external bridge in compute hosts, and the usage of a free floating IP per router/node.

Check this link for more information about how DVR works. https://docs.openstack.org/liberty/ne...

Regards

2018-08-06 03:01:12 -0600 answered a question tacker (devstack Queens) , generating Tosca outputs

Hi,

It would look something like:

outputs:
  floating_ip:
    value:
      get_attr: [FIP1, floating_ip, 0, floating_ip_address]

As for example the MGMT IP output:

outputs:
  mgmt_ip-VDU1:
    value:
      get_attr: [CP1, fixed_ips, 0, ip_address]

Regards

2018-07-09 01:27:59 -0600 commented question OSAD - vmware integration

Just a quick note, there is no official tool to deploy openstack, there are many options and alternatives under OpenStack umbrella

2018-07-03 07:57:40 -0600 commented question There is no Heat orchestration option in horizon dashboard Queens

Guess, you mean there is not dashboard, before queens heat was inside horizon code, in queens and later, heat dashboard is in a different package and need to be installed, package name: openstack-heat-ui

2018-07-03 05:19:31 -0600 answered a question issue in building neutron-server image kolla(stable/queens)

Hi, issue was just fixed. Could you try pulling stable/queens and build again?

https://review.openstack.org/#/c/579500/

Regards

2018-06-28 10:55:26 -0600 received badge  Nice Answer (source)
2018-06-27 05:38:45 -0600 edited question How do I revive/boot up an openstack Linux VM that has been shutdown (shutdown -h now) and is therefore in the SHUTOFF status

I have a Linux VM that I have shutdown with shutdown -h now, and it goes into the SHUTOFF status, which in my case I can see clearly in the GUI of Cisco's VM Maestro product that runs on top of VIRL (Virtual Internet Routing Lab). But anyway none of those layers should matter ... I cannot figure out for the life of me how and if you can revive a VM that is in this status - I am really starting to think that you cannot - I do NOT want to lose all my work, and STOPPING the VM would result in that. I have shutdown the Linux VM and so everything should be intact on its hard drive, it just needs to be rebooted / TURNED ON!

What is interesting is that the following documentation talks about VM states but then also mentions that SHUTOFF is a state but really it is a "status":

https://docs.openstack.org/nova/lates...

This SHUTOFF status can be seen in the following output.

virl@virl:~$ openstack server show 3799583b-cf87-4e82-80b6-dc8baa371056 
+--------------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| Field                                | Value                                                                                                                                                                                 |
+--------------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| OS-DCF:diskConfig                    | MANUAL                                                                                                                                                                                |
| OS-EXT-AZ:availability_zone          | nova                                                                                                                                                                                  |
| OS-EXT-SRV-ATTR:host                 | compute4                                                                                                                                                                              |
| OS-EXT-SRV-ATTR:hypervisor_hostname  | compute4                                                                                                                                                                              |
| OS-EXT-SRV-ATTR:instance_name        | instance-00002629                                                                                                                                                                     |
| OS-EXT-STS:power_state               | 4                                                                                                                                                                                     |
| OS-EXT-STS:task_state                | None                                                                                                                                                                                  |
| OS-EXT-STS:vm_state                  | stopped                                                                                                                                                                               |
| OS-SRV-USG:launched_at               | 2018-06-14T20:41:36.000000                                                                                                                                                            |
| OS-SRV-USG:terminated_at             | None                                                                                                                                                                                  |
| accessIPv4                           |                                                                                                                                                                                       |
| accessIPv6                           |                                                                                                                                                                                       |
| addresses                            | </ddamerji/endpoint>-<PGW_CPS_Remote_Topology_Stub_Config-liE7E1>-<Router1-to-lab>=10.255.255.2; </ddamerji/endpoint>-<PGW_CPS_Remote_Topology_Stub_Config-liE7E1>-<mgmt>=10.255.0.39 |
| config_drive                         | True                                                                                                                                                                                  |
| created                              | 2018-06-14T20:41:31Z                                                                                                                                                                  |
| flavor                               | CPS_16GbRAM_8CPU_100Gb (6c8148e4-b555-4038-953f-58d164073ac8)                                                                                                                         |
| hostId                               | 75acc970ac906aad964d02b87dfa7c56f670c69be8c82e14424d7b62                                                                                                                              |
| id                                   | 3799583b-cf87-4e82-80b6-dc8baa371056                                                                                                                                                  |
| image                                | ddamerji-server-lab-shutoff (59c5e758-117f-4fa5-9c24-17f408fe4d7c)                                                                                                                    |
| key_name                             | None                                                                                                                                                                                  |
| name                                 | </ddamerji/endpoint>-<PGW_CPS_Remote_Topology_Stub_Config-liE7E1>-<lab>                                                                                                               |
| os-extended-volumes:volumes_attached | []                                                                                                                                                                                    |
| project_id                           | 18a1a4dd4ef7454cae1ca9ab1a97219a                                                                                                                                                      |
| properties                           | node_id='lab', simulation_id='PGW_CPS_Remote_Topology_Stub_Config-liE7E1', user_id='ddamerji'                                                                                         |
| security_groups                      | [{u'name': u'default'}, {u'name': u'default'}]                                                                                                                                        |
| status                               | SHUTOFF                                                                                                                                                                               |
| updated                              | 2018-06-15T21:11:38Z                                                                                                                                                                  |
| user_id                              | 5cbe5c346a9e47f2b5e27dbd431e8f04                                                                                                                                                      |
+--------------------------------------+------------------------------

I'd appreciate any assistance here, internet searches have not turned up anything definitive, I did find one hit from 2014 where the person said they could just do an OpenStack server start, but I tried that in this case and it did NOTHING, the status remained shutoff:

image description

2018-06-15 03:45:30 -0600 commented answer High level question, How to deploy openstack pike version using kolla-ansible including OpenDaylight ?

yes, add the dlux features in the opendaylight_features: "odl-mdsal-apidocs,odl-netvirt-openstack" variable

2018-06-15 03:35:39 -0600 commented question DevStack Tacker installation fail

maybe is lack of hardware resources? for tacker many services are required to be running

2018-06-14 09:24:12 -0600 commented question DevStack Tacker installation fail

Could you share logs from tacker-conductor servcice, also logs from tacker services. Paste to paste.openstack.org instead of an image

2018-06-11 08:14:59 -0600 commented answer openstack kolla - HA databases unable to start

Sorry, is undocumented, will add a patch later to docs. https://github.com/openstack/kolla-ansible/blob/master/tools/kolla-ansible#L66

See kolla-ansible --help

2018-06-11 03:49:36 -0600 answered a question openstack kolla - HA databases unable to start

Hi, stopping galera needs to be done in proper order manually. kolla-ansible stop is not enough for it.

Find the master node in your cluster, stop slaves, and finally stop master. To start first master, then slaves.

If this is not done, the cluster is broken due slaves doenst have the last writte.

If now is broken. stop your mariadb containers and execute kolla-ansible -i inventory mariadb_recovery.

This command will find the last master, and start the database recovery process.

Regards

2018-05-08 04:33:45 -0600 commented question api-services only bind to INTERNAL api address when not using haproxy with kolla-ansible

To allow different IPs be bind without HAproxy or something, will need to bind services to 0.0.0.0 which is highly insecure. In kolla we dont do that, if need different addresses will need to enable haproxy, even for aio works. HAproxy is required to redirect public IP to internal IPs

2018-05-08 04:30:39 -0600 commented question api-services only bind to INTERNAL api address when not using haproxy with kolla-ansible

public and internal endoint are mostly the same in code, admin only allows admin tasks (iirc is being removed that difference too).

2018-05-08 02:54:51 -0600 commented question api-services only bind to INTERNAL api address when not using haproxy with kolla-ansible

Binding to an specific IP is for security reasons, wont work access to any address in the host because users can jump through networks. You can play with network_interface and api_interface to select on which interface IP address want the services to be bind

2018-05-07 02:50:28 -0600 received badge  Good Answer (source)
2018-04-30 08:57:02 -0600 answered a question kolla deploy controllers with different interface names fails

Hi,

if using different interfaces, should add the variable in the inventory file and comment or remove it from globals.yml.

Is a variable preference thing, as globals.yml is inserted as --extra-vars

Regards

2018-04-27 06:11:06 -0600 answered a question vitrage ERROR: You are not authorized to use global_index.

Hi, the issue is in heat's policies.

By default is set to:

stacks:global_index: rule:deny_everybody

which block any global_index to any user. Vitrage requires this permission.

Will need to change the policie for something suitable for your environment like.

"stacks:global_index": "rule:deny_stack_user"

Regards

2018-04-20 05:28:12 -0600 received badge  Nice Answer (source)
2018-04-19 05:27:23 -0600 commented answer kolla-ansible pip install - Use pike or queens release??

fixed in the docs

2018-04-18 08:23:30 -0600 answered a question kolla-ansible pip install - Use pike or queens release??

Currently latest pip package is queens,

so if you do pip install kolla-ansible will install 6.0.0 (current version).

To install a older version use the version whithin pip command.

pip install kolla-ansible==5.0.x

Kolla/kolla-ansible versions are like:

  • 6.0.x queens
  • 5.0.x pike
  • 4.0.x ocata
  • etc
2018-04-16 04:48:24 -0600 commented question Tacker openwrt driver not working in kolla pike

Does your net_mgmt network allows ssh and icmp to instances, is public or private? If private tacker-server doesnt have connectivity to instances unless you apply some network rules/packet flows to the controller nodes

2018-04-13 05:13:01 -0600 commented question kolla-ansible config adds l3-agent qG-interface in br-int (==> No inet access for VMs)

"The l3-agents qg-interface should have been added to the br-ex bridge": qg should be a nbr-int with a int-br-ex peer poot to br-ex. Reason is routers ports are in br-int, so cannot attach a interface directly from br-ex to a router in br-int

2018-04-12 04:32:27 -0600 commented answer kolla-ansible how to manage/modify configuration files

node_custom_config is /etc/kolla/config.

That means the first file found the that list will be used, first scpecific to a hosts, second your global horizon custom configs, if no custom config will use kolla defaults file.

2018-04-11 09:22:00 -0600 commented answer kolla-ansible how to manage/modify configuration files

Regarding the change into the hosts volume, if you do a reconfigure/upgrade/deploy/genconfig. The config options will be overriden, proper way is using /etc/kolla/config in deploy hosts

2018-04-11 09:21:03 -0600 commented answer kolla-ansible how to manage/modify configuration files
  • "{{ node_custom_config }}/horizon/{{ inventory_hostname }}/local_settings"
    • "{{ node_custom_config }}/horizon/local_settings"
  • "local_settings.j2"