Fishbone's profile - activity

2019-12-29 16:25:13 -0500 marked best answer Openstack on Proxmox

Hello I am running openstack in a development environment on my PC. My PC hardware is very capable of supporting openstack; however I have run into a bit of a dilemma.

I have the Proxmox hypervisor running on my host machine and I have installed openstack icehouse under the 3-node architecture with the neutron networking feature under guest kvms. All components of my setup work 100% except one simple flaw. I cannot get the compute node working for various reasons and I have tried the following:

Originally I setup the compute node as a guest kvm as I did with the other nodes. This setup produced the error: libvirtError: internal error: no supported architecture for os type 'hvm' even after adjusting the nova.conf file with the libvirt kvm directive as instructed by the documentation. I also receive an error: An error occurred while trying to define a domain with xml: <domain type="kvm">

I soon attempted to force hardware virtualization on the kvm by disabling it in the hypervisor web GUI as instructed by the proxmox wiki to enable hardware virtualization from within a guest kvm. This produced an unreliable guest OS in regards to interface, as expected, but did allow hardware accelleration within the guest, however; then the nova-compute service would not be able to start properly. Any time I attempted to restart the service it would claim it refers to an unknown instance and echo that it started but the results wouldn't change.

I abolished the idea of using a kvm and attempted to modify an openvz container. The container would never be able to install the nova-compute package, echoing invoke-rc.d: initscript nova-compute, action "start" failed. All other features work properly, when the libvirtError: internal error: no supported architecture for os type 'hvm' message is the issue, the compute node is detected by the controller and able to communicate with the neutron server. I am running the icehouse deployment on 3 ubuntu 14.04 nodes as kvms and are provisioned as follows:

Controller = 1 core, 3GB RAM, 25GB HDD Neutron = 1 core, 2 GB RAM, 20 GB HDD KVM/Openvz - Nova = 6 core, 6 GB RAM, 750 GB HDD

All of these errors lead me to believe that a true nova compute node may be unable to run under a proxmox hypervisor host. Which leads to my question, is it possible to run a compute node under proxmox and if so, what would be the best approach.

Thank you for your time.

2019-08-06 02:43:26 -0500 received badge  Famous Question (source)
2019-02-04 14:44:23 -0500 received badge  Notable Question (source)
2018-10-18 04:05:23 -0500 received badge  Popular Question (source)
2018-10-16 20:58:14 -0500 asked a question Rexray reflect panic

Hello all
I have magnum api version 2.8.0 and I am creating a swarm-mode coe with 2 nodes and 1 master, using fedora-atomic as the image. I am able to create clusters perfectly fine, however; I am running into an issue with the rexray driver. I am running cinder block storage through an iscsi SAN via the cinder synology driver and I am not having any issues with any other services that use cinder.
I have attempted to manually structure the rexray service and it comes out with the same result, logs below:

time="2018-10-04T01:01:53Z" level=info msg="created pid file" pid=19376 time=1538614913689
time="2018-10-04T01:01:53Z" level=info msg="libStorage auto service mode enabled" driver=openstack time=1538614913706
time="2018-10-04T01:01:53Z" level=info msg="configured logging" libstorage.logging.httpRequests=false
libstorage.logging.httpResponses=false libstorage.logging.level=info server=fate-venom-sj time=1538614913749
time="2018-10-04T01:01:53Z" level=info msg="initializing server" server=fate-venom-sj time=1538614913750
time="2018-10-04T01:01:53Z" level=info msg="initializing default endpoint" autoEndpointMode=unix server=fate-venom-sj time=1538614913762
time="2018-10-04T01:01:53Z" level=info msg="initializing auto unix endpoint" endpoint=libstorage.server.endpoints.localhost server=fate-venom-sj time=1538614913771
time="2018-10-04T01:01:53Z" level=info msg="configured endpoint" address="unix:///var/run/libstorage/232641277.sock" endpoint=localhost server=fate-venom-sj time=1538614913787
time="2018-10-04T01:01:53Z" level=info msg="server created" server=fate-venom-sj time=1538614913788
time="2018-10-04T01:01:53Z" level=info msg="initialized endpoints" server=fate-venom-sj time=1538614913788
time="2018-10-04T01:01:53Z" level=info msg="initializing server services" server=fate-venom-sj time=1538614913789
time="2018-10-04T01:01:54Z" level=panic msg="reflect: call of reflect.Value.Elem on struct Value"

2018-10-16 20:28:08 -0500 received badge  Supporter (source)
2018-10-16 20:26:20 -0500 commented answer NoneType has no attribute 'get_user_id'

These options resolved My issue thank you very much!!! @lukaslemes

2018-09-24 23:35:56 -0500 received badge  Notable Question (source)
2018-09-24 21:29:03 -0500 received badge  Popular Question (source)
2018-09-14 11:37:23 -0500 asked a question NoneType has no attribute 'get_user_id'

Hello all
I have murano-api and murano-engine setup and upgraded from ocata to queens. Since the upgrade I have not been able to launch any environments. There were many issues with the keystone config that forced me to not follow documentation from openstack docs and I instead needed to use the password auth_type in the config.
Regardless of this I maintain the error when launching an environment:
"NoneType has no attribute 'get_user_id'"
I have tried recreating the murano database, reinstalling murano from apt repository and reinstalling murano from the git repository and the problem persists.
Please help. This has been like this for a while.

2018-06-26 02:28:26 -0500 received badge  Famous Question (source)
2018-06-26 02:28:26 -0500 received badge  Notable Question (source)
2018-06-26 02:28:26 -0500 received badge  Popular Question (source)
2018-04-12 06:54:53 -0500 received badge  Taxonomist
2018-02-06 09:05:34 -0500 received badge  Famous Question (source)
2017-10-03 02:26:50 -0500 received badge  Famous Question (source)
2017-03-21 05:14:35 -0500 received badge  Notable Question (source)
2017-02-02 15:10:24 -0500 asked a question ClusterTypeNotSupported

I have setup magnum in openstack newton. I can create cluster templates with no issue, however when I try to create a cluster from the template I receive the error in the magnum conductor logs, Error ClusterTypeNotSupported: Cluster type (vm, fedora_atomic, swarm) not supported.

I have tried other network drivers for the template and I have followed the online documentation for ubuntu 16. The error happens in both the GUI and in the CLI.

Thank you.

2016-11-28 01:59:16 -0500 received badge  Famous Question (source)
2016-11-22 19:29:13 -0500 received badge  Popular Question (source)
2016-10-28 16:58:25 -0500 asked a question manila share ssh error

Hello all, I am running Openstack Mitaka in a 3 Node architecture complete with Cinder, Manila, and Neutron.

Whenever I try to create a manila share I receive an error "ServiceInstanceException: SSH connection has not been established to in 300s. Giving up."

I am using the linuxbridge driver for neutron and the generic driver for manila. I have tested that the cinder volime service is working.
I receive this error on the manila share server(also the block storage node), however; I do not receive any other errors from any other servers. I have checked the tap and bridge adapter connections as requested, I can connect between the Vnetworks gracefully and all instances can connect to the internet just fine. Also I have tested connectivity to the share server instance created via the console and ssh from another instance on a routed network. I cannot ping any instances or Vnetworks from any nodes except through floating IPs and the public interface network, however; this reality has not affected network connectivity in any other way.
Please advise

2016-10-11 10:25:23 -0500 received badge  Notable Question (source)
2016-09-30 05:17:51 -0500 received badge  Popular Question (source)
2016-09-13 17:52:03 -0500 asked a question lbaasv2 dashboard issues

Hello all
I am running openstack Mitaka in a 2-node architecture, utilizing the linuxbridge agent, and following the openstack mitaka documentation for ubuntu. I have setup the load balancing V2 agent in mitaka using the haproxy agent, also following the documentation. I have installed the lbaasv2 horizon panel and I have verified the procedure as I am also using the panels for trove and manila.

When I try to create any load balancer from the web panel I get an error "error: source_type". I am able to create load balancers in the cli, however if I use the cli to create anything the lbaasv2 panel does not work at all with error: "cannot retrieve pools list". It will work again if I remove all load balancer devices.

I have the haproxy installed and I see none of these errors in the log, however; If I have a load balancer created I see "WARNING neutron_lbaas.drivers.haproxy.namespace_driver [-] Stats socket not found for loadbalancer" also the command ip netns does not show any load balancer devices and I do not have the "q-svc.log" file for checking api requests.

All other neutron services are running with no interuptions.
I believe the issue is either due to the haproxy driver or horizon api calls, however; I do not know what I can troubleshoot next, please assist.

2016-09-06 12:18:02 -0500 answered a question IOError: [Errno 13] Permission denied: '/tmp/.secret_key_store

If you followed the documentation from the Openstack official website ensure that you copy the proper directives from the original conf file. Notable, The documentation mentions to use the www-data or apache user to setup https on the server. Your configured user may be different than the user that is authorized for the file.

2016-07-07 08:10:21 -0500 received badge  Popular Question (source)
2016-06-15 16:48:09 -0500 asked a question load balancer rdp

Hello all, I have an openstack Liberty setup in a 2-node architecture with 2 compute controllers. Utilizing the linuxbridge agent and vxlan interfaces. I have been able to configure the stack in a virtualized environment running on a debian proxmox backend. The virtualized nodes themselves are running ubuntu 14.04 server. I have lbaas v1 setup and was wondering if it is possible to setup a load balancer for an RDP connection.
The load balancer is setup to the proper subnet with a pingable VIP and a floating IP attached. The loadbalancer has the proper servers as members and has the appropriate port assigned; the protocol implemented was TCP. The VIP is not pingable through the floating IP however; it is pingable from within the subnet. Is there anything I may be doing wrong?

---OK I am adjusting the question as I have verified that RDP works in the load balancer. The problem I noticed I am having is that the load balancer is not being routed to via the floating IP.

2016-04-12 18:12:42 -0500 commented question liberty add another compute node

Hi Thanks for the reply. They each have 2 interfaces. The primary interfaces are appended as -1 and -2 respectively so their primary/management interfaces are different. Their secondary interfaces, which are to be configured manually, with no ip, are the same, however.

2016-04-12 18:10:22 -0500 received badge  Notable Question (source)
2016-04-04 18:07:29 -0500 received badge  Famous Question (source)
2016-04-04 18:07:29 -0500 received badge  Popular Question (source)
2016-04-04 18:07:29 -0500 received badge  Notable Question (source)
2016-03-03 02:47:09 -0500 received badge  Popular Question (source)
2016-03-02 14:04:19 -0500 asked a question liberty add another compute node

Hello all
I am running openstack liberty in, currently, a 3-node architecture utilizing the linuxbridge agent on the controller and compute nodes and isolating the glance service to its own node.

I am able to create instances just fine and most features are working properly. One error I have is when attaching a secondary compute node all instances lose network access. I have the second compute node with the same neutron configuration settings as the initial compute node, I am able to create instances on both nodes and logs show no errors on any nodes.

I am almost positive I am required to tweak the neutron configuration on the second compute node in some way, as implicitly specified in the openstack liberty documentation, however; I am not sure what those tweaks should be and I have found no leads in reference to the setup guidelines of the documentation.
Any information is highly appreciated.

2016-03-02 13:42:21 -0500 asked a question Windows updates stuck

Hello all,
I am running openstack liberty in, currently, a 3-node architecture utilizing the linuxbridge agent on the controller and compute nodes, and isolating the glance service to its own node.

I recently had issues downloading web files in windows instances that I was able to resolve by disabling the windows TCP autotuninglevel setting. I was unable to download windows updates prior to that change and I am still unable to. That was the only change that seemed to resolve the http download timeout errors and I am sure it is not the resolution for the windows update issue I am having.

Currently the windows updates stay in a "checking for updates" state and this will not change for days at a time.

I have only had one case where the updates were able to go through however I had the updates checking while I was not monitoring the instance and I was experimenting at the time; therefore I do not know if there were any specific settings the OS may have had that could have affected the outcome.

If there is any information that I may be missing please provide me with assistance. I have tried many windows update fixes and none seem to work.

2016-03-02 13:28:32 -0500 received badge  Notable Question (source)
2016-03-02 13:28:32 -0500 received badge  Popular Question (source)
2016-03-02 13:28:32 -0500 received badge  Famous Question (source)
2016-03-01 18:00:59 -0500 answered a question Instance HTTP issues

Issue resolved. The instances needed to adjust the tcp window size using the netsh int tcp set global autotuninglevel

2016-01-20 10:03:40 -0500 asked a question Reset Packet sent from instance

Hello all, I am running openstack Liberty, currently in a 2-node architecture, a controller and a compute controller. My openstack system is running on top of a proxmox 3-node cluster. The Nodes are running ubuntu server 14, currently only with essential services up, such as keystone, glance, nova, etc.

I am able to create instances and connect to the internet perfectly fine, however; when I try to download any item from a web browser the connection is reset. There are no errors showing in any of the logs, and as mentioned any other internet communications perform uninterrupted.

I had this exact issue when running openstack icehouse and juno and I am aware that icehouse used the gre implementation, while juno initiated flat for external and liberty now uses vxlan for internal; this makes me believe the issue is not with the setup. When I attempted to trace the packets by running wireshark inside the vm I noticed that there is an RST flag being sent from the custom port, eg: 49462, to the destination port; in this case 443.
I am using the default security group and even tried a custom security group that allowed all anywhere to and from(which I believe the default does already), but that did not change anything

I have been plagued with this issue since my pursuance of openstack and I am too enticed by its possibilities to give up. Are there any ideas that anybody might have?

Thank you for your time.

2015-09-10 13:51:50 -0500 received badge  Famous Question (source)
2015-09-09 09:59:38 -0500 received badge  Popular Question (source)