[Openstack] [neutron][networking-sfc] Load distribution within a port pair group
If running networking-sfc on ocata, pike or queens what functionality works when it comes to distribution of traffic over the port-pairs in a port-pair-group ? /Lars ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
[Openstack] [nova] Compute node in Pike does not register itself
I can not get my compute nodes to register themelves when using Pike. It works OK in Ocata. Is there some additional config, service or software package required in Pike ? If I monitor the IP traffic on the compute node the following can be seen when nova-compute is started: Ocata: Compute node “registration” message sent from compute node to the controller node, along with calls to the placement api. Pike: No activity at all is seen on the network. The nova-compute log does not show anything after 2018-01-17 11:10:41.044 DEBUG os_brick.initiator.connector [req-1095033c-df63-4123-9d89-e3a5610a4b61 None None] Factory for VERITAS_HYPERSCALE on x86_64 from (pid=35345) factory /usr/lib/python2.7/dist-packages/os_brick/initiator/connector.py:290 Anyone that could provide any hints on this ? /Lars ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
[Openstack] [heat] AutoScalingGroup and dependents
I am using OS::Heat::AutoScalingGroup in the following type of scenario resources: my_vm: type: SOME_TEMPLATE_WITH_A_VM_CONNECTED_TO_MULTIPLE_NETWORKS properties: networks: { get_attr: [my_networks, outputs_list, my_network] } my_networks: type: OS::Heat::AutoScalingGroup properties: resource: type: SOME_TEMPLATE_WITH_A_NETWORK_ATTRIBUTE_my_network_HOLDS_NETWORK_ID When I deploy my stack everything works as expected. When I trigger (using a ScalingPolicy and its alarm_url) the my_networks group scales as expected. Now I want the my_vm resource to adjust to the new state (re-evaluation of { get_attr: [my_networks, outputs_list, my_network] } ). Can this be done ? /Lars ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
[Openstack] [heat]
I am trying to create a deployment scenario with the following characteristics: - Controllable via the Horizon Heat UI - Uses nested resource types The only way I have been able to do this is to create an external server where all nested templates are stored and either reference the nested templates via explicit URLs in the nesting template or in an environment file. Could this be done without the need for an external server ? /Lars ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
[Openstack] [heat] Using value_specs
A number of resource types supports a ”value_specs” property. I have problem finding out valid vales for this property. I have tried various map values but whatever I try I get the message that the value is not allowed. How to find out the possible value sets? /Lars ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
Re: [Openstack] [designate] Synchronize bind9 backend
Hi Graham, We were using the designate version in newton bit is now using ocata, however the behaviour is the same (i.e. no synch) Can I look for traces of the period synchs in some log file? Is there some config parameter that controls the sunch? /Lars On 2017-05-04, 14:23, "Graham Hayes" <gha...@suse.de> wrote: On 03/05/17 10:51, Lars-Erik Helander wrote: > We are using designate with a bind9 backend in a newton based Openstack > system. > > When the designate processes and the bind9 process are restarted they > get out of synch. The zones in designate are no longer in bind9. How can > I get the bind9 backend to get synchronized after a restart? > > > > /Lars Designate should run a periodic sync to check that all the zones are on the bind9 server - what version of Designate are you using? - Graham > > > ___ > Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack > Post to : openstack@lists.openstack.org > Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack > ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
[Openstack] [designate] Synchronize bind9 backend
We are using designate with a bind9 backend in a newton based Openstack system. When the designate processes and the bind9 process are restarted they get out of synch. The zones in designate are no longer in bind9. How can I get the bind9 backend to get synchronized after a restart? /Lars ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
[Openstack] [heat][cinder] How to manage my volumes ?
I am struggling to find a solution given the following: - A cluster of two or more servers - Each server has a storage volume attached - If a server goes down (disappears) it should attach to the same (identical content) volume when it is re-created - Minimal manual intervention for all life-cycle use cases - Fully managed vi Heat If it would be possible to define the following logic, it would be great, but I have not found any way to do this: - For each server, pre-declare the name of the volume to attach - When server is created and a volume with the pre-declared name exists, that volume is attached - When a server is created and a volume with the pre-declared name does not exist, such a volume is automatically created - When a server is deleted, its attached volume is deattached but not deleted (could do this using “deletion_policy: ‘Retain’” ) I would be happy to receive feedback on any idea that will take me closer to a solution. I have tried to use “external_id:” in conjunction with OS::Cinder::Volume in order to get a pre-declared name for each volume, but that did not work as I expected it to. I ma currently on “Newton”. Thanks ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
Re: [Openstack] [openstack] Tenant/Project resource name spaces does not seem to work
Akihiro and James, Thanks for your input. I did suspect that “admin” role could be the issue, so I had tested that but not in all ways. What I have found is this: - Having a user that has had the admin role assignment but then the assignment is removed, then using Horizon you will only see its own (and shared) networks. But if you use the CLI (“openstack network list”) the list of presented networks are still ALL. - Creating a user from scratch without any admin role assignment, it will only show its own (and shared networks) when using “openstack network list”. So there seems to be some “caching” of the “role assignment”. /Lars On 2017-04-06, 22:48, "Akihiro Motoki" <amot...@gmail.com> wrote: Do you use a user with admin role? If so this is the expected behavior because Neutron returns all networks from all projects when admin role is used. 2017-04-06 22:30 GMT+09:00 Lars-Erik Helander <lars-erik.helan...@proceranetworks.com>: > If I create networks in a project and define the networks to be non-shared, > I still can use these networks from other projects. Not via Horizon but via > the Openstack CLI commands (openstack, nova, neutron etc) and via Heat (heat > templates may refer to networks in other projects). > > > > Is this how it is supposed to be? > > Might I have misconfigured my Openstack in order to get this behaviour, if > so any hints on where to look for the settings that causes the current > behaviour? > > > > /Lars > > > ___ > Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack > Post to : openstack@lists.openstack.org > Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack > ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
[Openstack] [openstack] Tenant/Project resource name spaces does not seem to work
If I create networks in a project and define the networks to be non-shared, I still can use these networks from other projects. Not via Horizon but via the Openstack CLI commands (openstack, nova, neutron etc) and via Heat (heat templates may refer to networks in other projects). Is this how it is supposed to be? Might I have misconfigured my Openstack in order to get this behaviour, if so any hints on where to look for the settings that causes the current behaviour? /Lars ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
Re: [Openstack] [Heat] Conditions on resource properties
Thanks, found the example in the documentation but I could not make my case to work. What if the values are more complex than strings, e.g. a list of maps, do anyone have an example of that? On 2017-03-22, 20:11, "Zane Bitter" <zbit...@redhat.com> wrote: On 22/03/17 07:45, Lars-Erik Helander wrote: > The Heat documentation states that conditions could be applied to > resources, resource properties and output values. I have failed to set > conditions on resource properties and I have found no examples. Anyone > that has a working example? You'll want to use the 'if' intrinsic function to specify different property values based on conditions. cheers, Zane. ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
[Openstack] [Heat] Conditions on resource properties
The Heat documentation states that conditions could be applied to resources, resource properties and output values. I have failed to set conditions on resource properties and I have found no examples. Anyone that has a working example? ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
Re: [Openstack] HOT ResourceGroup index referencing with get_attr
Here are two templates foo.yaml and bar.yaml foo.yaml heat_template_version: newton description: > Create a set of channel ports on a specific network. parameters: network: type: string description: Network label: Network port_count: type: number description: Number of channel ports to create label: Port count default: 1 channel_ports: type: comma_delimited_list description: List of channel ports label: Channel ports resources: ports: type: OS::Heat::ResourceGroup properties: count: { get_param: port_count } resource_def: type: bar.yaml properties: network: { get_param: network} ports: { get_param: channel_ports } index: "%index%" bar.yaml heat_template_version: newton description: > A port created based on a list of names parameters: ports: type: comma_delimited_list description: List of channel ports label: Channel ports index: type: number description: Index into channel ports list label: index network: type: string description: Network to allocate port from label: Network resources: port: type: OS::Neutron::Port properties: network_id: { get_param: network} name: { get_param: [ports, { get_param: index } ] } port_security_enabled: False On 2017-03-12, 17:59, "Turbo Fredriksson" <tu...@bayour.com> wrote: On 12 Mar 2017, at 15:59, Lars-Erik Helander <lars-erik.helan...@proceranetworks.com> wrote: > I think you need to pass %index% in a string, e.g. Could you give me some examples? I’m really struggling to get this working! ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
Re: [Openstack] HOT ResourceGroup index referencing with get_attr
I think you need to pass %index% in a string, e.g. type: Turbo::LoadBalancerMember properties: index: “%index%” instance: { get_attr: [instance, instance] } pool: { get_attr: [listener_tcp_8301, pool] } /Lasse On 2017-03-12, 16:09, "Turbo Fredriksson"wrote: On 12 Mar 2017, at 14:50, Turbo Fredriksson wrote: > So how do you pass on ‘%index%’ to that child? > > Because I still need to “retrieve” the IP with get_attr so I can send it to the child.. Well, I could just send the whole ‘instance’ ID, and in the child retrieve it. But I still can’t figure out how to “send” it to the child: —— s n i p —— member_tcp_8301: type: OS::Heat::ResourceGroup properties: count: { get_param: capacity } resource_def: type: Turbo::LoadBalancerMember properties: index: %index% instance: { get_attr: [instance, instance] } pool: { get_attr: [listener_tcp_8301, pool] } port: 8301 —— s n i p —— gives: —— s n i p —— found character '%' that cannot start any token in "", line 71, column 18: index: %index% —— s n i p —— This with a child like: —— s n i p —— heat_template_version: 2015-04-30 description: Add an instance to a Loadbalancer pool listener. parameters: instance: type: string description: Instance ID of the instance to add to the listener. pool: type: string description: Pool ID of the pool to add the instance to. port: type: string description: Port number to forward from the listner to the instance. index: type: number description: Index in address list to use resources: member: type: OS::Neutron::LBaaS::PoolMember properties: address: { get_attr: [{ get_param: instance }, address, index] } subnet: { get_attr: [{ get_param: subnet }, port_subnet, index] } pool: { get_param: pool } protocol_port: { get_param: port } outputs: member: value: { get_resource: member } description: The member ID. —— s n i p —— ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
Re: [Openstack] HOT ResourceGroup index referencing with get_attr
I have also had similar problems with %index% and have resolved these using an additional “child” resource where I use “%index%” as a parameter and within the child resource use { get_attr: [instance, address, index] } /Lasse On 2017-03-12, 13:28, "Turbo Fredriksson"wrote: I'm trying to figure out a way to setup a [number of] 'OS::Neutron::LBaaS::PoolMember’ resource, from a list of IPs… I’m setting up my initial resources like this: —— s n i p —— resources: lbaas: type: Turbo::LoadBalancer properties: name: { get_param: 'OS::stack_name' } network: { get_param: network } protocol: TCP port_nr: 8300 dnsname: consul instance: type: OS::Heat::ResourceGroup properties: count: { get_param: capacity } resource_def: type: Turbo::InstancePoolMember properties: name: { list_join: ['-', [ { get_param: "OS::stack_name"} ] ] } image: { get_param: image } pool_id: { get_attr: [lbaas, pool] } port_nr: { get_attr: [lbaas, port] } [etc] —— s n i p —— My “Turbo::LoadBalancer” is a [seriously] nested resource to setup the LBaaS, Pool, Monitor, FloatingIP, DNS forward and reverse resource and _one_ listener (the “initial” one). My "Turbo::InstancePoolMember” creates the instance(s) _and_ adds it/them to the listener created in the ‘lbaas’ resource. This work just fine, took quite a while to get all that working in a generic way. But it works. However, this specific service (Hashicorps Consul) I’m trying to create now needs additional listeners. Over all, I don’t “know” how many listeners a service need, so the nested resources only create one (at least one is required). My plan/idea was then to create additional listeners and add the ‘instance’s created from/in the ‘instance’ resource above to that listener in the HOT. I can create the listener just fine (this creates a listener and attaches a pool to that listener, then attaches the listener to the LBaaS created above): —— s n i p —— listener_tcp_8301: type: Turbo::LoadBalancerListener properties: name: { list_join: ['-', [ { get_param: "OS::stack_name"} ] ] } lbaas: { get_attr: [lbaas, lbaas] } protocol: TCP port_nr: 8301 —— s n i p —— However, creating the members have shown to be a lot more complicated: —— s n i p —— member_tcp_8301: type: OS::Heat::ResourceGroup properties: count: { get_param: capacity } resource_def: type: OS::Neutron::LBaaS::PoolMember properties: protocol_port: 8301 pool: { get_attr: [listener_tcp_8301, pool] } address: { get_attr: [instance, address, %index%] } subnet: { get_attr: [instance, port_subnet, %index%] } —— s n i p —— (All resources in this mail is in _one_ HOT, just referencing others, not shown). The problem is that get_attr: [instance, address, %index%] doesn't seem to work :( This is what I’ve tried so far: —— s n i p —— address: { get_attr: [instance, address, "%index%"] } subnet: { get_attr: [instance, port_subnet, "%index%"] } 2017-03-11 21:22:42Z [member_tcp_8301]: CREATE_FAILED resources.member_tcp_8301: Property error: resources[1].properties.subnet: Value must be a string 2017-03-11 21:22:42Z [admin-consul]: CREATE_FAILED Resource CREATE failed: resources.member_tcp_8301: Property error: resources[1].properties.subnet: Value must be a string —— s n i p —— —— s n i p —— address: { get_attr: [instance, address.%index%] } subnet: { get_attr: [instance, port_subnet.%index%] } 2017-03-11 23:31:35Z [admin-consul]: CREATE_FAILED Resource CREATE failed: The Referenced Attribute (0 port_subnet.%index%) is incorrect. —— s n i p —— —— s n i p —— address: { get_attr: [instance, %index%, address] } subnet: { get_attr: [instance, %index%, port_subnet] } found character '%' that cannot start any token in "", line 71, column 43: ... address: { get_attr: [instance, %index%, address] } —— s n i p —— —— s n i p —— address: { get_attr: [instance, address, %index%] } subnet: { get_attr: [instance, port_subnet, %index%] } found character '%' that cannot start any token in "", line 71, column 52: ... { get_attr: [instance, address, %index%] } —— s n i p —— —— s n i p —— address: { get_attr: [instance, address, { list_join: ['', [ '', "%index%" ] ] } ] } subnet: { get_attr: [instance, port_subnet, { list_join: ['', [ '', "%index%" ] ] } ] }
[Openstack] [Heat] Dynamically create OS::Nova:Server "networks" property
I have tried in numerous ways to compose a proper value for the “networks” property of OS::Nova::Server but with no success. Basically I have two “inputs” that together shall result in the complete networks lists, e.g. List1 : “[{port: foo1}, {port: foo2}]” List2 : “[{network: foo3}, {port: foo4}, network: foo5}]” Should be combined into a networks specification that is similar to this: type: OS::Nova::Server properties: networks: - port: foo1 - port: foo2 - network: foo3 - port: foo4 - network: foo5 I have tried the repeat intrinsic function but have not found any working syntax. List input should be both from an input parameter and/or from an output attribute of some resource. Any help is highly appreciated. ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
Re: [Openstack] [openstack][tacker] Support for resource outputs
1) Is it possible to use the output value (e.g. the IP address since that is what exists today) as input parameter (or similar) in some other VNFD ? 2) Is it possible to use the assigned IP address from one VDU as input property to some other VDU (e.g. in cloud config) within the same VNFD ? Thanks Lars From: Sripriya Seetharam <ssee...@brocade.com> Date: Tuesday, 22 November 2016 at 01:09 To: Lars-Erik Helander <lars-erik.helan...@proceranetworks.com>, "openstack@lists.openstack.org" <openstack@lists.openstack.org> Subject: Re: [Openstack] [openstack][tacker] Support for resource outputs What exactly are you looking for in terms of defining output attributes for VNF? Right now, VNF ip address is the only thing returned as an output attribute from Heat to Tacker. Cheers, Sripriya From: Lars-Erik Helander <lars-erik.helan...@proceranetworks.com> Date: Wednesday, November 16, 2016 at 7:05 AM To: "openstack@lists.openstack.org" <openstack@lists.openstack.org> Subject: [Openstack] [openstack][tacker] Support for resource outputs Do tacker support, in some way, something similar to what Heat supports in terms of defining output attributes to a resource and referencing these output attributes in other resources? ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
[Openstack] [openstack][tacker] Support for resource outputs
Do tacker support, in some way, something similar to what Heat supports in terms of defining output attributes to a resource and referencing these output attributes in other resources? ___ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack