Hey sorry Patrick. I missed your earlier question. +Feng who also has done a great deal of work on the VPP side. We allow passing the UIO driver to use in the deploy settings. So if we had a package with the Mellanox driver installed on Overcloud image, we should be able to use it. Is that something you are interested in doing? Or are you going to use a different NIC?
I see a package here: http://www.mellanox.com/downloads/Drivers/MLNX_DPDK_2.2_4.2.tar.gz Tim Rozet Red Hat SDN Team ----- Original Message ----- From: "Daniel Bernier (520165)" <daniel.bern...@bell.ca> To: "Patrick Lemay" <patrick.le...@bell.ca>, "Tim Rozet" <tro...@redhat.com> Cc: opnfv-tech-discuss@lists.opnfv.org, "Jamo Luhrsen" <jluhr...@redhat.com>, "Jocelyn Poulin (6007251)" <jocelyn.pou...@bell.ca>, "Francois Guay (A214312)" <francois.g...@bell.ca>, "Brian Smith (3010640)" <b.sm...@bell.ca> Sent: Friday, December 2, 2016 8:56:18 AM Subject: Re: [opnfv-tech-discuss] [Apex] [Fdio]deployment error FYI, Issue is that the Mellanox PMD driver is not compiled by default with DPDK since it requires proprietary libs. Without it, VPP cannot bind to these NICs. There is a thread regarding this in vpp-...@lists.fd.io -----Original Message----- From: Patrick Lemay <patrick.le...@bell.ca> Date: Friday, November 25, 2016 at 10:43 AM To: Tim Rozet <tro...@redhat.com> Cc: "opnfv-tech-discuss@lists.opnfv.org" <opnfv-tech-discuss@lists.opnfv.org>, Jamo Luhrsen <jluhr...@redhat.com>, "Poulin, Jocelyn (6007251)" <jocelyn.pou...@bell.ca>, "Bernier, Daniel (520165)" <daniel.bern...@bell.ca>, "Guay, Francois (A214312)" <francois.g...@bell.ca>, "Smith, Brian (3010640)" <b.sm...@bell.ca> Subject: RE: [opnfv-tech-discuss] [Apex] [Fdio]deployment error I Also found those messages from the compute node. Does it mean I have to change my network card? If yes do you have a list of supported 10 gig devices? Nov 25 14:43:53 localhost puppet-user[15262]: (/Stage[main]/Fdio::Service/Service[vpp]/ensure) ensure changed 'stopped' to 'running' Nov 25 14:43:53 localhost vpp: vlib_plugin_early_init:213: plugin path /usr/lib/vpp_plugins Nov 25 14:43:53 localhost /usr/bin/vpp[15849]: dpdk_bind_devices_to_uio:893: Unsupported Ethernet PCI device 0x15b3:0x1015 found at PCI address 0000:81:00.0 Nov 25 14:43:53 localhost vpp: /usr/bin/vpp[15849]: dpdk_bind_devices_to_uio:893: Unsupported Ethernet PCI device 0x15b3:0x1015 found at PCI address 0000:81:00.0 -----Original Message----- From: Lemay, Patrick Sent: November-24-16 8:40 PM To: 'Tim Rozet' Cc: 'opnfv-tech-discuss@lists.opnfv.org'; 'Jamo Luhrsen'; Poulin, Jocelyn (6007251); Bernier, Daniel (520165); Guay, Francois (A214312); Smith, Brian (3010640) Subject: RE: [opnfv-tech-discuss] [Apex] [Fdio]deployment error Hi Tim, I found the problem with the network. You can forget this message. It was related to the physical interface on the switch that cause a problem with the dhcp request. I've been able to deploy opnfv basic with ovs. Now, I'm working on os-odl_l2-fdio-noha.yaml. I ran this command with the default settings. opnfv-deploy -n network_settings.yaml -i inventory.yaml -d os-odl_l2-fdio-noha.yaml --interactive --debug I receive some errors at the end of the overcloud deployment.. Cannot find vpp interface matching: 81/0/0 Any Idea? Regards, Pat -----Original Message----- From: Lemay, Patrick Sent: November-22-16 10:40 AM To: 'Tim Rozet' Cc: opnfv-tech-discuss@lists.opnfv.org; Jamo Luhrsen; Poulin, Jocelyn (6007251); Bernier, Daniel (520165); Guay, Francois (A214312); Smith, Brian (3010640) Subject: RE: [opnfv-tech-discuss] [Apex] deployment error Hi Tim, we are nearly having a working deployment setup. I can confirm that pxe work well. I think you are right regarding the post deployment task. There is something wrong on my network_settings.yaml file. When the deployment start the interfaces enp1s0f0 obtain an ip from undercloud dhcp 10.66.20.10. 30 seconds after, it wipes the config and no communication is possible after. You'll see it from the screenshot. Could you look at my network_settings.yaml file please? We can also do a session to show you. I'm pretty sure that I miss configure something in that file that cause some problem in my post deployment task. Interface enp1s0f0 should be on 10.66.20.xx/24 admin_network Interface ens2f0 should be on 10.66.25.XX/24 public_network Both interfaces are in access mode and gateway is .1. Deployment server Undercloud) is 10.66.20.10. [stack@undercloud ~]$ nova list +--------------------------------------+-------------------------+--------+------------+-------------+----------------------+ | ID | Name | Status | Task State | Power State | Networks | +--------------------------------------+-------------------------+--------+------------+-------------+----------------------+ | 494e3987-057a-4ba3-87a1-8169e8f9c126 | overcloud-controller-0 | BUILD | spawning | NOSTATE | ctlplane=10.66.20.26 | | d70c26ff-4d34-4787-bec6-9b596fd03019 | overcloud-novacompute-0 | ACTIVE | - | Running | ctlplane=10.66.20.24 | | 99b4b166-985a-41ff-a425-4ae95cc28db1 | overcloud-novacompute-1 | ACTIVE | - | Running | ctlplane=10.66.20.25 | +--------------------------------------+-------------------------+--------+------------+-------------+----------------------+ Regards, Patrick Lemay Consultant Managed Services Engineering Bell Canada 671 de la Gauchetière O. Bur. 610, Montreal, Quebec H3B 2M8 Tel: (514) 870-1540 -----Original Message----- From: Tim Rozet [mailto:tro...@redhat.com] Sent: November-16-16 3:10 PM To: Lemay, Patrick Cc: opnfv-tech-discuss@lists.opnfv.org; Jamo Luhrsen; Poulin, Jocelyn (6007251); Bernier, Daniel (520165); Guay, Francois (A214312); Smith, Brian (3010640) Subject: Re: [opnfv-tech-discuss] [Apex] deployment error Hi Patrick, In Apex, configuring bonds is not supported yet. It is possible in TripleO and with some advanced config you could use a workaround, but for now let's just focus on getting the deployment to work. When you execute the deployment (use --debug arg to opnfv-deploy), can you console into one of the servers and see if it actually PXE boots into linux? If it does then it means your network connectivity on your admin network is correct and take note of which NIC MAC is used to PXE. Now go to your host and do: opnfv-util undercloud . stackrc nova list ping <each node's ip> If one or more pings fail, it means the network configuration of the box post-pxe boot is wrong. At this point you need to console into the overcloud node. Since you used --debug, a default root password of 'opnfvapex' was applied to all the overcloud nodes. So you can login to the node and look at /var/log/messages for: Nov 15 09:58:26 localhost os-collect-config: [2016/11/15 09:58:26 AM] [INFO] nic1 mapped to: eth0 Nov 15 09:58:26 localhost os-collect-config: [2016/11/15 09:58:26 AM] [INFO] nic2 mapped to: eth1 Nov 15 09:58:26 localhost os-collect-config: [2016/11/15 09:58:26 AM] [INFO] nic3 mapped to: eth2 Nov 15 09:58:26 localhost os-collect-config: [2016/11/15 09:58:26 AM] [INFO] nic4 mapped to: eth3 In your network settings file if you are using logical nic mapping (meaning nic<#>), then you will need to double check that the admin (ctrlplane) network is correctly wired from your undercloud VM (host admin network NIC) to the logical nic mapping in your network settings file. This logical nic name resolves to a physical NIC name on the host using the mapping above. If it is wrong, then you can either fix that by providing the correct logical nic name in your network settings file or use the real physical nic name (in this case eth0) for your settings. You can compare the MAC of the PXE boot interface from earlier to determine which physical NIC is your admin network nic. The nic settings are per each network and per profile (compute/control). Tim Rozet Red Hat SDN Team ----- Original Message ----- From: "Patrick Lemay" <patrick.le...@bell.ca> To: "Tim Rozet" <tro...@redhat.com>, opnfv-tech-discuss@lists.opnfv.org Cc: "Jamo Luhrsen" <jluhr...@redhat.com>, "Jocelyn Poulin (6007251)" <jocelyn.pou...@bell.ca>, "Daniel Bernier (520165)" <daniel.bern...@bell.ca>, "Francois Guay (A214312)" <francois.g...@bell.ca>, "Brian Smith (3010640)" <b.sm...@bell.ca> Sent: Tuesday, November 15, 2016 11:58:01 AM Subject: RE: [opnfv-tech-discuss] [Apex] deployment error Hi Tim, adding configuration on enp1s0f0 and f1 worked well it use this config to create br-admin and br-public. Thanks for that workaround. Now, undercloud deploy successfully. IPMI with sol work fine also. I'm able to poweroff and poweron servers with ironic. The only thing, the deployment script failed . It's probably related to the network between the node and the undercloud that is not well configure. We are supposed to create a bond interface on 10g but this part has not been configure because there is no example. How can I confirm that? Is there a journal with more details? Also, my ctlplane interface is well configure but unreachable from the undercloud when source by stackrc. [root@undercloud ~]# ironic node-list +--------------------------------------+------+--------------------------------------+-------------+--------------------+-------------+ | UUID | Name | Instance UUID | Power State | Provisioning State | Maintenance | +--------------------------------------+------+--------------------------------------+-------------+--------------------+-------------+ | 9f069885-b9cb-4dec-85ab-945fb08a8753 | None | b1efb2e1-6e78-4fc0-9d5a-5b323088bebb | power on | active | False | | f273975c-06ba-4c36-b7e6-25872d7efce0 | None | a4c77e61-6d18-4e1d-94ca-cd007282856a | power on | active | False | +--------------------------------------+------+--------------------------------------+-------------+--------------------+-------------+ [root@undercloud ~]# nova list +--------------------------------------+-------------------------+--------+------------+-------------+----------------------+ | ID | Name | Status | Task State | Power State | Networks | +--------------------------------------+-------------------------+--------+------------+-------------+----------------------+ | b1efb2e1-6e78-4fc0-9d5a-5b323088bebb | overcloud-controller-0 | ACTIVE | - | Running | ctlplane=10.66.20.24 | | a4c77e61-6d18-4e1d-94ca-cd007282856a | overcloud-novacompute-0 | ACTIVE | - | Running | ctlplane=10.66.20.23 | +--------------------------------------+-------------------------+--------+------------+-------------+----------------------+ When I try to source overcloudrc the file is not present but stackrc is present. [stack@undercloud ~]$ source overcloudrc -bash: overcloudrc: No such file or directory [stack@undercloud ~]$ source stackrc openstack server list [stack@undercloud ~]$ openstack server list +--------------------------------------+-------------------------+--------+----------------------+ | ID | Name | Status | Networks | +--------------------------------------+-------------------------+--------+----------------------+ | b1efb2e1-6e78-4fc0-9d5a-5b323088bebb | overcloud-controller-0 | | ACTIVE | ctlplane=10.66.20.24 | | a4c77e61-6d18-4e1d-94ca-cd007282856a | overcloud-novacompute-0 | | ACTIVE | ctlplane=10.66.20.23 | +--------------------------------------+-------------------------+--------+----------------------+ [stack@undercloud ~]$ ssh heat-admin@10.66.20.24 ssh: connect to host 10.66.20.24 port 22: No route to host [stack@undercloud ~]$ ssh heat-admin@10.66.20.23 ssh: connect to host 10.66.20.23 port 22: No route to host [stack@undercloud ~]$ ping 10.66.20.24 PING 10.66.20.24 (10.66.20.24) 56(84) bytes of data. ^C --- 10.66.20.24 ping statistics --- 2 packets transmitted, 0 received, 100% packet loss, time 1000ms Ping the undercloud works: [stack@undercloud ~]$ ping 10.66.20.101 PING 10.66.20.101 (10.66.20.101) 56(84) bytes of data. 64 bytes from 10.66.20.101: icmp_seq=1 ttl=64 time=0.055 ms 64 bytes from 10.66.20.101: icmp_seq=2 ttl=64 time=0.049 ms -----Original Message----- From: Tim Rozet [mailto:tro...@redhat.com] Sent: November-09-16 11:14 AM To: Guay, Francois (A214312) Cc: Jamo Luhrsen; Lemay, Patrick; opnfv-tech-discuss@lists.opnfv.org; Poulin, Jocelyn (6007251) Subject: Re: [opnfv-tech-discuss] [Apex] deployment error Hi Patrick, For the Undercloud VM to bridge to your host, it needs to be able to get the IP information off of the host interfaces. It does this by checking your ifcfg files under /etc/sysconfig/network-scripts for the interface you specify for each network. Can you check that enp1s0f1 ifcfg file is not set to dhcp and has IP and NETMASK/PREFIX settings in the file? Thanks, Tim Rozet Red Hat SDN Team ----- Original Message ----- From: "Francois Guay (A214312)" <francois.g...@bell.ca> To: "Jamo Luhrsen" <jluhr...@redhat.com>, "Patrick Lemay" <patrick.le...@bell.ca>, opnfv-tech-discuss@lists.opnfv.org Cc: "Jocelyn Poulin (6007251)" <jocelyn.pou...@bell.ca> Sent: Friday, November 4, 2016 2:59:14 PM Subject: Re: [opnfv-tech-discuss] [Apex] deployment error Patrick, Jamo, I did the libvirt-python install and it solve the problem. (see attached) Now, we need to add the two missing nodes to get the five required nodes. I will give it a try and let you know. Thanks Jamo. Patrick, I'll let you share with Jamo what you did with the admin_network and public_network stuff. François -----Original Message----- From: Jamo Luhrsen [mailto:jluhr...@redhat.com] Sent: 4 novembre 2016 12:04 To: Lemay, Patrick; opnfv-tech-discuss@lists.opnfv.org Cc: Poulin, Jocelyn (6007251); Guay, Francois (A214312) Subject: Re: [opnfv-tech-discuss] [Apex] deployment error Patrick, I've had similar problems trying to get apex baremetal to work on the jumphost I'm working with. I hit the libvirt issue the other day. I think I just did a "yum install libvirt-python" try that. I'm still stuck on what I need to do with the admin_network, public_network stuff that I think you have overcome recently. Can you recap the steps you took? JamO On 11/04/2016 08:47 AM, Lemay, Patrick wrote: > Hi I finally create the br-public by hand instead of script. It work > but now I’m stock to another. «ImportError: No module named libvirt». > I’m using your deployment cd. All the dependencies should be install. I start to tshoot the problem but I’m sure that you’ve already see that bug. > > > > All python version installed: > > [root@jumphost opnfv-apex]# python > > python python2 python2.7 python3 python3.4 python3.4m > > > > Output from the opnfv-deploy script: > > INFO: virsh networks set: > > Name State Autostart Persistent > > ---------------------------------------------------------- > > admin_network active yes yes > > default active yes yes > > public_network active yes yes > > > > All dependencies installed and running > > 4 Nov 10:58:33 ntpdate[27293]: adjust time server 206.108.0.133 offset > -0.002592 sec > > Volume undercloud exists. Deleting Existing Volume > /var/lib/libvirt/images/undercloud.qcow2 > > Vol undercloud.qcow2 deleted > > > > Vol undercloud.qcow2 created > > > > Traceback (most recent call last): > > File "/usr/libexec/openstack-tripleo/configure-vm", line 8, in > <module> > > import libvirt > > ImportError: No module named libvirt > > > > > > Thanks, > > > > > > > > > > *From:*Lemay, Patrick > *Sent:* November-01-16 11:42 AM > *To:* 'opnfv-tech-discuss@lists.opnfv.org' > *Cc:* Bernier, Daniel (520165); Guay, Francois (A214312); Poulin, > Jocelyn (6007251) > *Subject:* Bell deployment config > > > > Hi guys, I have some issues regarding opnfv baremetal deployment. I > install a jumphost from opnfv cd. I configure the inventory with IPMI ip mac and users. > > > > I’m not sure that I configure the network_settings correctly for public_network. > > > > > > For the deployment I use pod 2 and pod 3 from the drawing. The server > Catherine is use for the jumphost. Interface enp1s0f0 is for pxe and > enp1s0f1 is for public_network . The other 5 servers are for > deployment IPMI ready. All interfaces in vlan > 1020 are pxe ready and disk are configured raid 1. I have a problem deploying undercloud. > > > > > > I have this error related to the undercloud deployment: > > INFO: Creating Virsh Network: admin_network & OVS Bridge: br-admin > > INFO: Creating Virsh Network: public_network & OVS Bridge: br-public > > INFO: Bridges set: > > br-admin > > br-public > > enp1s0f0 > > INFO: Interface enp1s0f0 bridged to bridge br-admin for enabled > network: admin_network > > /var/opt/opnfv/lib/common-functions.sh: line 18: 5 - ( / 8) : syntax > error: operand expected (error token is "/ 8) ") > > ERROR: IPADDR or NETMASK/PREFIX missing for enp1s0f1 > > ERROR: Unable to bridge interface enp1s0f1 to bridge br-public for > enabled network: public_network > > > > > > Could you help please? There is no vmware at all in the setup only baremetal.. > > > > Regards, > > > > > > > > Patrick Lemay > > Consultant Managed Services Engineering Bell Canada > > 671 de la Gauchetière O. Bur. 610, Montreal, Quebec H3B 2M8 > > Tel: (514) 870-1540 > > > > > > _______________________________________________ > opnfv-tech-discuss mailing list > opnfv-tech-discuss@lists.opnfv.org > https://lists.opnfv.org/mailman/listinfo/opnfv-tech-discuss > _______________________________________________ opnfv-tech-discuss mailing list opnfv-tech-discuss@lists.opnfv.org https://lists.opnfv.org/mailman/listinfo/opnfv-tech-discuss _______________________________________________ opnfv-tech-discuss mailing list opnfv-tech-discuss@lists.opnfv.org https://lists.opnfv.org/mailman/listinfo/opnfv-tech-discuss
