| *** jwagner is now known as jwagner_away | 00:07 | |
| *** sdake has joined #openstack-ansible | 00:08 | |
| *** Mudpuppy has quit IRC | 00:08 | |
| *** Mudpuppy has joined #openstack-ansible | 00:08 | |
| *** gardenshed has quit IRC | 00:09 | |
| *** daneyon_ has quit IRC | 00:09 | |
| *** jwagner_away is now known as jwagner | 00:10 | |
| *** sdake_ has quit IRC | 00:11 | |
| *** galstrom_zzz is now known as galstrom | 00:15 | |
| *** sdake has quit IRC | 00:28 | |
| *** galstrom is now known as galstrom_zzz | 00:29 | |
| *** sdake has joined #openstack-ansible | 00:31 | |
| *** jwagner is now known as jwagner_away | 00:38 | |
| *** sdake_ has joined #openstack-ansible | 00:42 | |
| *** alop has quit IRC | 00:43 | |
| *** CheKoLyN has quit IRC | 00:44 | |
| *** sdake has quit IRC | 00:45 | |
| *** sdake_ has quit IRC | 00:53 | |
| *** sdake has joined #openstack-ansible | 00:53 | |
| *** tlian2 has joined #openstack-ansible | 01:00 | |
| *** galstrom_zzz is now known as galstrom | 01:01 | |
| *** markvoelker has joined #openstack-ansible | 01:02 | |
| *** tlian has quit IRC | 01:02 | |
| *** k_stev has joined #openstack-ansible | 01:03 | |
| *** tlian2 has quit IRC | 01:04 | |
| *** sdake_ has joined #openstack-ansible | 01:05 | |
| *** sdake has quit IRC | 01:06 | |
| *** tlian has joined #openstack-ansible | 01:11 | |
| *** galstrom is now known as galstrom_zzz | 01:11 | |
| *** sdake_ is now known as sdake | 01:31 | |
| *** k_stev has quit IRC | 02:25 | |
| *** galstrom_zzz is now known as galstrom | 02:34 | |
| *** shausy has joined #openstack-ansible | 02:40 | |
| *** markvoelker has quit IRC | 02:42 | |
| *** metral is now known as metral_zzz | 03:05 | |
| *** shausy has quit IRC | 03:05 | |
| *** galstrom is now known as galstrom_zzz | 03:08 | |
| *** galstrom_zzz is now known as galstrom | 03:13 | |
| *** gardenshed has joined #openstack-ansible | 03:13 | |
| *** metral_zzz is now known as metral | 03:14 | |
| *** gardenshed has quit IRC | 03:18 | |
| *** dolpher has joined #openstack-ansible | 03:36 | |
| dolpher | container['ansible_ssh_host'] = networks[old_address]['address'] KeyError: 'address' | 03:45 |
|---|---|---|
| dolpher | any idea what this problem is? | 03:45 |
| dolpher | it's a fresh install, after configuration, I got this error | 03:46 |
| *** daneyon has joined #openstack-ansible | 04:04 | |
| *** daneyon_ has joined #openstack-ansible | 04:05 | |
| *** galstrom is now known as galstrom_zzz | 04:06 | |
| *** daneyon has quit IRC | 04:09 | |
| cloudnull | dolpher: hard to tell. maybe the address for the given container is missing a cidr from the cidr nets list ? | 04:39 |
| cloudnull | you mind pasting your openstack_user_config.yml file ? | 04:40 |
| openstackgerrit | Merged openstack/openstack-ansible: Removing extra 'provides' from doc https://review.openstack.org/237737 | 04:44 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Removing extra 'provides' from doc https://review.openstack.org/237869 | 04:45 |
| openstackgerrit | Merged openstack/openstack-ansible-security: AllowUnauthenticated fails when it shouldn't https://review.openstack.org/237007 | 04:47 |
| *** daneyon_ has quit IRC | 04:48 | |
| *** sdake_ has joined #openstack-ansible | 05:00 | |
| *** sdake has quit IRC | 05:01 | |
| *** openstackgerrit has quit IRC | 05:16 | |
| *** javeriak has joined #openstack-ansible | 05:16 | |
| *** gardenshed has joined #openstack-ansible | 05:16 | |
| *** openstackgerrit has joined #openstack-ansible | 05:16 | |
| *** gardenshed has quit IRC | 05:21 | |
| *** shausy has joined #openstack-ansible | 05:29 | |
| *** shausy has quit IRC | 05:42 | |
| *** shausy has joined #openstack-ansible | 05:43 | |
| *** javeriak has quit IRC | 05:50 | |
| dolpher | cloudnull: problem found | 06:03 |
| dolpher | cidr_networks: | 06:05 |
| dolpher | container: 172.29.236.0/22 | 06:05 |
| dolpher | ... | 06:05 |
| dolpher | global_overrides: | 06:05 |
| dolpher | provider_networks: | 06:05 |
| dolpher | - network: | 06:05 |
| dolpher | container_bridge: "br-mgmt" | 06:05 |
| dolpher | container_type: "veth" | 06:05 |
| dolpher | container_interface: "eth1" | 06:05 |
| dolpher | ip_from_q: "management" | 06:05 |
| dolpher | The network name mismatch, so it's my config file problem:) | 06:05 |
| *** javeriak has joined #openstack-ansible | 06:10 | |
| *** gardenshed has joined #openstack-ansible | 06:19 | |
| *** gardenshed has quit IRC | 06:23 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Install Guide Cleanup https://review.openstack.org/237513 | 06:24 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Install Guide Cleanup https://review.openstack.org/237513 | 06:25 |
| *** daneyon has joined #openstack-ansible | 06:26 | |
| *** daneyon_ has joined #openstack-ansible | 06:27 | |
| openstackgerrit | Merged openstack/openstack-ansible: Install Guide Cleanup https://review.openstack.org/237513 | 06:27 |
| *** daneyon has quit IRC | 06:31 | |
| *** tlian has quit IRC | 06:33 | |
| *** k_stev has joined #openstack-ansible | 06:35 | |
| *** javeriak has quit IRC | 06:36 | |
| *** Mudpuppy has quit IRC | 06:36 | |
| odyssey4me | nice patch matt https://review.openstack.org/237513 - care to try and backport to kilo? it'll have some conflicts as there are some differences in content between the branches | 06:41 |
| odyssey4me | mattt ^ | 06:41 |
| mattt | odyssey4me: i can certainly try | 06:42 |
| mattt | this was all prep work to getting the ceph work added :( | 06:43 |
| mattt | kinda got derailed | 06:43 |
| dolpher | "http://rpc-repo.rackspace.com/container_images/rpc-trusty-container.tgz" download failed | 06:44 |
| dolpher | because the download speed is slow, any workaround? | 06:44 |
| odyssey4me | try again dolpher ;) | 06:46 |
| odyssey4me | mattt do you know of any reason why 11.2.4 should not release today? | 06:47 |
| mattt | odyssey4me: nothing is jumping out at me ... the no-venv patches went through, so that should all work now | 06:50 |
| odyssey4me | mattt yeah, that doesn't touch kilo anyway - no no issues there | 06:50 |
| mattt | ah true then def. none that i'm aware of | 06:55 |
| *** daneyon has joined #openstack-ansible | 07:02 | |
| odyssey4me | ok, 11.2.4 has been released | 07:04 |
| *** daneyon_ has quit IRC | 07:06 | |
| *** daneyon has quit IRC | 07:06 | |
| *** karimb has joined #openstack-ansible | 07:19 | |
| *** javeriak has joined #openstack-ansible | 07:19 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Update kilo for new dev work - 21 Oct 2015 https://review.openstack.org/237907 | 07:23 |
| *** javeriak_ has joined #openstack-ansible | 07:23 | |
| *** javeriak has quit IRC | 07:25 | |
| odyssey4me | mattt if you have a gap to review https://review.openstack.org/234926 it'd be appreciated | 07:27 |
| *** jongchoi has joined #openstack-ansible | 07:30 | |
| *** javeriak_ has quit IRC | 07:39 | |
| *** javeriak has joined #openstack-ansible | 07:42 | |
| *** mpavone has joined #openstack-ansible | 07:44 | |
| *** karimb has quit IRC | 07:59 | |
| mattt | odyssey4me: alright cool will have a look | 08:10 |
| *** gardenshed has joined #openstack-ansible | 08:10 | |
| odyssey4me | morning hughsaunders :) | 08:29 |
| hughsaunders | sup odyssey4me | 08:29 |
| odyssey4me | so cloudnull and I had a chat about https://review.openstack.org/230716 and https://review.openstack.org/236183 earlier this morning | 08:29 |
| odyssey4me | while the patches are functional as-is, they put a whole mess of stuff into playbooks and not into roles | 08:31 |
| *** k_stev has quit IRC | 08:31 | |
| *** fawadkhaliq has joined #openstack-ansible | 08:31 | |
| odyssey4me | the question is whether we should let them through as-is, or whether we should insist on a revision and have them in one or more roles | 08:32 |
| odyssey4me | then of course we have to consider whether we shouldn't let them through now and evolve them into roles later | 08:33 |
| odyssey4me | your thoughts? | 08:33 |
| *** gardenshed has quit IRC | 08:44 | |
| *** jongchoi has quit IRC | 08:48 | |
| *** gardenshed has joined #openstack-ansible | 08:49 | |
| *** sdake_ is now known as sdake | 08:51 | |
| *** neilus has quit IRC | 09:00 | |
| *** neilus has joined #openstack-ansible | 09:00 | |
| *** openstackgerrit has quit IRC | 09:01 | |
| *** openstackgerrit has joined #openstack-ansible | 09:01 | |
| *** karimb has joined #openstack-ansible | 09:12 | |
| *** gparaskevas has joined #openstack-ansible | 09:20 | |
| gparaskevas | morning everyone! | 09:21 |
| gparaskevas | ready for the summit? | 09:22 |
| *** javeriak has quit IRC | 09:23 | |
| *** javeriak has joined #openstack-ansible | 09:24 | |
| openstackgerrit | Merged openstack/openstack-ansible: Update Neutron Configuration for Liberty https://review.openstack.org/234926 | 09:24 |
| *** karimb has quit IRC | 09:26 | |
| *** openstackgerrit has quit IRC | 09:31 | |
| *** openstackgerrit has joined #openstack-ansible | 09:31 | |
| mattt | gparaskevas: you going? | 09:46 |
| gparaskevas | mattt: no unfortunatly i cant, but my collegue will be there | 09:46 |
| gparaskevas | i will watching you from youtube | 09:47 |
| gparaskevas | will be* | 09:47 |
| *** ashishjain has joined #openstack-ansible | 09:53 | |
| ashishjain | Hello | 09:53 |
| ashishjain | Need some advice. | 09:53 |
| *** fawadkhaliq has quit IRC | 09:53 | |
| ashishjain | How do you configure the provider extenal network in openstack_user_config.yml? | 09:54 |
| ashishjain | *external | 09:54 |
| odyssey4me | ashishjain although I'm not the best to advise - typically if you have a provider network it's either a vlan or a 'flat' network | 09:54 |
| odyssey4me | in both cases you have bridges for those | 09:55 |
| odyssey4me | a 'flat' network is untagged from a server standpoint, whereas a vlan network is tagged - ie the tag must be configured in the network conf for the server | 09:55 |
| odyssey4me | but if you're asking from the point of view of inside a tenant network - then you'd use neutron and set the network accordingly - its type and (if applicable) tag | 09:56 |
| odyssey4me | hughsaunders mattt can I borrow your attention for 10 mins? | 09:56 |
| hughsaunders | yep | 09:56 |
| mattt | odyssey4me: doing training atm | 09:56 |
| ashishjain | odyssey4me: Yes you are correct, it is a vlan or flat. In my case I have got br-vlan mapped to vlan net type as well as flat, this is as per the example provided | 09:58 |
| ashishjain | Now when I see the linux bridge configuration which is again as per osad I see physical_interface_mapping as flat:eth12 | 10:00 |
| ashishjain | This is again as per osad where flat is always mapped to eth12 in the linux agent container | 10:00 |
| ashishjain | Now afaik any external communication has to go through neutron agent node | 10:01 |
| *** dolpher has quit IRC | 10:02 | |
| ashishjain | Now to have connectivity to external world I need to basically have some interface in my agent talking to the external world, because I want to use flat for external does it mean eth12 has to be mapped to some bridge which is connected to external world? | 10:03 |
| ashishjain | I think that is the only way external connectivity would work | 10:03 |
| hughsaunders | ashishjain: it depends, if you attach instances directly to a flat provider network, there traffic doesn't have to go via a neutron agent node | 10:03 |
| hughsaunders | but if you are using a tenant network with floats on the provider network, then you need a router which sits on one of the agent nodes | 10:04 |
| ashishjain | But when I see the openstack_user_config eth12 is having container bridge as br-vlan which is definitely not a bridge which provided an external connectivity | 10:04 |
| hughsaunders | br-vlan should have an interface on a network that has external access | 10:04 |
| ashishjain | hughsaunders: You are correct I am using a tenant network and than using a floating ip address. as you said I am using a router which sits on agent node | 10:05 |
| ashishjain | hughsaunders: I will give you some more details of my setup | 10:06 |
| *** javeriak has quit IRC | 10:06 | |
| *** javeriak has joined #openstack-ansible | 10:07 | |
| hughsaunders | ashishjain: http://docs.rackspace.com/rpc/api/v11/bk-rpc-installation/content/ch-overview.html#fig_overview_neutron-agents | 10:07 |
| gparaskevas | ashishjain: Hello, let me tell you my example , i use vlan typte for external connectivity, i create an external network, or a siple network and i select provider vlan, and segment id let say 10(vlan id) on my router i have already created a vlan and interface for that tag | 10:08 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-specs: Add Liberty Release spec https://review.openstack.org/221189 | 10:08 |
| ashishjain | hughsaunders: My setup is hostservermachine->VM->osad containers. Hostmachine has got 3 virtual bridges and one network interface which connects to the external network | 10:09 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-specs: Add Liberty Release spec https://review.openstack.org/221189 | 10:09 |
| ashishjain | My VM on which the containers are running uses vlans as suggested by the rackspace or osad docs. | 10:10 |
| ashishjain | gparaskevas, hughsaunders I will create a paste to present the actual configs | 10:11 |
| gparaskevas | sure | 10:11 |
| ashishjain | This is my host machine configuration. This machine has got 64 GB RAM and 16 cores. We are dividing this machine into multiple VM's to have a multi node openstack environment using osad | 10:13 |
| *** fawadkhaliq has joined #openstack-ansible | 10:13 | |
| ashishjain | Here is the network configuration of this host machine http://paste.openstack.org/show/476972/ | 10:13 |
| hughsaunders | ashishjain: I think its confusing to enable vlan and flat if you aren't using both | 10:13 |
| hughsaunders | ashishjain: so your OSAD VMs have interfaces that attach to bridge0-2 ? | 10:15 |
| ashishjain | hughsaunders: Yes you are correct . Here is the network configuration of VM whcih is the target node http://paste.openstack.org/show/476973/ | 10:15 |
| ashishjain | My eth0 interface has go the network access | 10:16 |
| ashishjain | Network configuration has been done as per the osad or rackspace guide | 10:17 |
| hughsaunders | so is em1 a sub interface of bridge0 on the host? | 10:17 |
| ashishjain | No em1 is not a subinterface of bridge0 , it is an actual physical interface | 10:18 |
| ashishjain | This is the document which has been refered for network configuration, we are not using bonding https://osad.readthedocs.org/en/latest/install-guide/targethosts-networkexample.html | 10:19 |
| ashishjain | Now the problems I am facing and looks like I have figured out the issue but dont know how to solve it | 10:19 |
| ashishjain | I am creating a tenant network using vxlan and external network using flat | 10:21 |
| hughsaunders | ashishjain: I'm not sure how your VMs have external access if your hyp's physical interface isnt' attached a bridge? | 10:21 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Update heat keystone_authtoken config https://review.openstack.org/235978 | 10:21 |
| ashishjain | hughsaunders: You are correct but I do not think the intention was to have all the bridges connected to internet or physical interface | 10:22 |
| hughsaunders | ashishjain: not all, but one if you need external access | 10:22 |
| ashishjain | each of these bridges serve different purpose where bridge0 is my management network, bridge1 is my instance tunnel network and bridge2 is my storage network | 10:23 |
| odyssey4me | ashishjain you're using the wrong docs | 10:23 |
| odyssey4me | ashishjain readthedocs is old, this is the updated documentation: http://docs.openstack.org/developer/openstack-ansible/kilo/ | 10:24 |
| ashishjain | odyssey4me: I have seen that as well their is hardly an difference. | 10:24 |
| odyssey4me | ashishjain there are differences - please use that as your reference | 10:24 |
| ashishjain | hughsaunders: Any of these bridges cannot have external connectivity. My point is where in osad openstack_user_config I can define an external network. All the provider network definitions are internal to openstack | 10:26 |
| ashishjain | odyssey4me: sure I will use that as a reference | 10:26 |
| *** jaypipes has joined #openstack-ansible | 10:26 | |
| ashishjain | In the neutron agents container /etc/neutron/plugins/ml2/ml2_conf.ini has an entry physical_interface_mappings = vlan:eth11,flat:eth12 | 10:27 |
| ashishjain | So this means If I have to use flat for external connectivity my eth12 interface should be mapped to an external bridge but the openstack_user-config maps it to br-vlan | 10:28 |
| hughsaunders | br-vlan is the external bridge | 10:29 |
| hughsaunders | Or should be... | 10:29 |
| ashishjain | Okay if br-vlan is external than what if I want to have a vlan tenant nework? | 10:31 |
| hughsaunders | ashishjain: i haven't tried that but I think it would go via br-vxlan | 10:33 |
| ashishjain | hughsaunders: I get your point and that is what even I am thinking | 10:33 |
| odyssey4me | nothing stops your vlan bridge from being used as the external network - you just need to make sure that the external network is tagged | 10:33 |
| odyssey4me | I could be talking rubbish though :p | 10:33 |
| hughsaunders | yeah, I guess you could have a provider network and vlan tentnat networks on the same bridge separated by vlans | 10:34 |
| odyssey4me | but as I recall at my previous workplace, we had to use a vlan bridge for all provider networks - whether tenant/project or for /external/floats | 10:34 |
| *** fawadkhaliq has quit IRC | 10:35 | |
| ashishjain | hughsaunders: So what is the change you think I should make to my config in http://paste.openstack.org/show/476973/. I think I will have to create one more bridge on my host machine which probably can talk to em1 on my host machine. | 10:36 |
| ashishjain | and than create one more network interface on my target VM and change the bridge_ports eth2 for auto br-vlan to this new network interface. Does it sound logical? | 10:37 |
| hughsaunders | ashishjain: yep | 10:39 |
| ashishjain | alright thanks I will try that out | 10:39 |
| openstackgerrit | Merged openstack/openstack-ansible: Removing extra 'provides' from doc https://review.openstack.org/237869 | 10:48 |
| openstackgerrit | Merged openstack/openstack-ansible-security: V-38643: World writable files https://review.openstack.org/233216 | 10:49 |
| ashishjain | hughsaunders: One more thing eth0 interface on my VM is directly connected to internet as per this paste http://paste.openstack.org/show/476973/ | 10:49 |
| openstackgerrit | Merged openstack/openstack-ansible-security: Some checks stop the playbook run https://review.openstack.org/237014 | 10:49 |
| ashishjain | hughsaunders: If I use bridge_ports eth0 can that help, I tried it but than my eth0 stops having any ip addr and I am totally disconnected from internet | 10:51 |
| hughsaunders | ashishjain: you can put an ip on the bridge | 10:51 |
| ashishjain | hughsaunders: I did not get you | 10:52 |
| hughsaunders | ashishjain: if you add eth0 to a bridge, you can assign the ip that eth0 did have to the bridge | 10:55 |
| ashishjain | hughsaunders: but than will dhcp work when i create an ext-net? | 10:58 |
| hughsaunders | ashishjain: dhcp should be ok across a bridge. I don't recommend mixing br-mgmt and br-vlan though | 10:59 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Set Keystone endpoints to be v3 by default https://review.openstack.org/205192 | 11:07 |
| *** mgoddard has quit IRC | 11:09 | |
| *** fawadkhaliq has joined #openstack-ansible | 11:10 | |
| *** shausy has quit IRC | 11:14 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Update heat keystone_authtoken config https://review.openstack.org/235978 | 11:25 |
| openstackgerrit | Matt Thompson proposed openstack/openstack-ansible: Fix Install Guide's TOC https://review.openstack.org/237996 | 11:31 |
| *** dolpher has joined #openstack-ansible | 11:35 | |
| *** Mudpuppy has joined #openstack-ansible | 11:40 | |
| *** fawadkhaliq has quit IRC | 11:40 | |
| *** Mudpuppy has quit IRC | 11:45 | |
| *** daneyon has joined #openstack-ansible | 11:45 | |
| *** fawadkhaliq has joined #openstack-ansible | 11:50 | |
| *** daneyon_ has joined #openstack-ansible | 11:52 | |
| *** daneyon has quit IRC | 11:54 | |
| *** fawadkhaliq has quit IRC | 11:55 | |
| *** fawadkhaliq has joined #openstack-ansible | 11:56 | |
| *** manas has joined #openstack-ansible | 12:03 | |
| *** fawadkhaliq has quit IRC | 12:04 | |
| *** karimb has joined #openstack-ansible | 12:04 | |
| *** KLevenstein has joined #openstack-ansible | 12:15 | |
| *** openstackgerrit has quit IRC | 12:16 | |
| *** openstackgerrit has joined #openstack-ansible | 12:17 | |
| *** daneyon_ has quit IRC | 12:19 | |
| *** daneyon has joined #openstack-ansible | 12:26 | |
| ashishjain | hughsaunders: I created a new bridge on the host and routed it to em1. Next I bring up a new ethernet interface on the VM and modify the bridge ports on br-vlan to this new ethernet interface. | 12:29 |
| ashishjain | While my router is able to get a new ip address and I am able to ping it | 12:29 |
| ashishjain | this ping is working from an external host. Even my vm's are able to ping this router. | 12:30 |
| ashishjain | However when I create a floating ip address and assign it to the VM i am not able to ping or ssh through my VM | 12:30 |
| odyssey4me | ashishjain did you open the icmp ports in your security groups for the project? | 12:32 |
| ashishjain | odyssey4me: Yes it is open for icmp and tcp 22 | 12:32 |
| *** dolpher has quit IRC | 12:33 | |
| hughsaunders | ashishjain: can you ping instances if you boot them with an interface on your provider network? | 12:34 |
| *** dolpher has joined #openstack-ansible | 12:34 | |
| *** markvoelker has joined #openstack-ansible | 12:35 | |
| ashishjain | hughsaunders: I am able to ping the VM's when I use the net namespaces using qdhcp and qrouter and even ssh into those | 12:35 |
| *** mgoddard has joined #openstack-ansible | 12:36 | |
| ashishjain | hughsaunders: If I run this command I am able to ssh into my instance. | 12:38 |
| ashishjain | ip netns exec qdhcp-68a718d2-3bd4-48ce-9a5c-58d8903019d7 ssh cirros@10.200.208.226 | 12:38 |
| ashishjain | this is being done from my network agents node | 12:38 |
| ashishjain | hughsaunders: "can you ping instances if you boot them with an interface on your provider network?" So you mean I should create a subnet with an existing provider network and create a floating ip address, assign it to VM and see if it works? | 12:39 |
| openstackgerrit | Matt Thompson proposed openstack/openstack-ansible: Install Guide Cleanup https://review.openstack.org/238034 | 12:40 |
| hughsaunders | ashishjain: provide the id of the provider network when booting an instance (--nic net-id=...) | 12:40 |
| ashishjain | hughsaunders: but these network are not part of neutron, when I run neutron net-list I just get the one's which I have created ext-net and demo-net. How do I get the id's of these provider networks | 12:42 |
| hughsaunders | ashishjain: you should have a neutron network that represents your provider network (has router-external=true and an external gateway) | 12:43 |
| openstackgerrit | Matt Thompson proposed openstack/openstack-ansible: Install Guide Cleanup https://review.openstack.org/238034 | 12:43 |
| hughsaunders | sounds like ext-net may be that in your case? | 12:43 |
| ashishjain | yes you are correct aahh so you mean directly use the external network instead of going through the floating ip route | 12:43 |
| hughsaunders | yeah | 12:44 |
| ashishjain | I think I have tried this sometime back | 12:44 |
| hughsaunders | its a good test of connectivity | 12:44 |
| ashishjain | Okay I will try this out 1 sec | 12:44 |
| *** Bjoern_ has joined #openstack-ansible | 12:44 | |
| ashishjain | hughsaunders: instance is up | 12:45 |
| ashishjain | but I am unable to ping this instance the ip is 10.200.208.227 | 12:45 |
| ashishjain | instance is still booting up | 12:47 |
| hughsaunders | ashishjain: use nova console-log to check the instance acquired the ip, then tcpdump in various places to work out where the problem is | 12:48 |
| *** tlian has joined #openstack-ansible | 12:51 | |
| ashishjain | hughsaunders: No it did not get the IP as per the console-log, here are some of the errors which are present in the log | 12:52 |
| ashishjain | Sending discover... Usage: /sbin/cirros-dhcpc <up|down> No lease, failing WARN: /etc/rc3.d/S40-network failed | 12:52 |
| *** Bjoern_ has quit IRC | 12:52 | |
| ashishjain | == pinging gateway failed, debugging connection === | 12:52 |
| ashishjain | cat: can't open '/etc/resolv.conf': No such file or directory | 12:53 |
| ashishjain | route -n is also empty | 12:53 |
| ashishjain | hughsaunders: Just wanted to tell you this floating ip address belongs to office network | 12:54 |
| ashishjain | So do I need to specify nameservers etc when I am booting the instance? | 12:54 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Add theme fix for browsable source code https://review.openstack.org/238042 | 12:54 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-security: Add theme fix for browsable source code https://review.openstack.org/238045 | 12:56 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-security: Add theme fix for browsable source code https://review.openstack.org/238045 | 12:57 |
| *** galstrom_zzz is now known as galstrom | 13:00 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-specs: Add theme fix for browsable source code https://review.openstack.org/238061 | 13:12 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-security: Add theme fix for browsable source code https://review.openstack.org/238045 | 13:13 |
| *** manas has quit IRC | 13:14 | |
| hughsaunders | ashishjain: I've had a few issues with unreliable DHCP, I'd try another instance and see if it gets an address, otherwise poking in the dhcp namespace and using tcpdump to figure out how far the dhcp requests are getting | 13:20 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-security: Add theme fix for browsable source code https://review.openstack.org/238045 | 13:24 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Add theme fix for browsable source code https://review.openstack.org/238042 | 13:26 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-specs: Add theme fix for browsable source code https://review.openstack.org/238061 | 13:27 |
| ashishjain | hughsaunders: Can you please tell me how shall I about using tcpdump, As soon as the instance is fired where shall I run the tcpdump on has it be network agent container because that is where the dhcp agent sevices are running? | 13:29 |
| ashishjain | correction: where shall I run the tcpdump, does it has to be network agent container because that is where the dhcp agent sevices are running? | 13:29 |
| *** javeriak has quit IRC | 13:30 | |
| hughsaunders | ashishjain: I would start on your compute node targeting the bridge that is connected to your provider network, if you see requests there, then check the same bridge on the neutron agents node | 13:31 |
| ashishjain | hughsaunders: Here we are talking about directly using the external network which afaik does not reside on compute node, it is all on the networking node | 13:33 |
| hughsaunders | ashishjain: for direct attached instances on a flat/vlan network, the computes must also be attached to the provider network | 13:36 |
| ashishjain | hughsaunders: You mean the ml2_conf.ini in compute node should have same definition as well as config as the target host for network services for example this should be same physical_interface_mappings = vlan:br-vlan,flat:eth0 | 13:38 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Add theme fix for browsable source code https://review.openstack.org/238042 | 13:39 |
| hughsaunders | ashishjain: yeah, osad will drop the same config on network nodes and compute nodes as both have the linuxbridge agent | 13:40 |
| ashishjain | hughsaunders: I think I will have to re-run the playbooks for making some changes on compute node than :( | 13:42 |
| ashishjain | on an another note I am seeing the following on the target node, not sure if you see anything like this while running tcpdump | 13:42 |
| ashishjain | 19:07:50.200298 IP 10.200.208.202.34370 > XXXXXX.com.domain: 51345+ AAAA? openstack002_neutron_agents_container-12be2114. (64) 19:07:50.201241 IP XXXXX.com.domain > 10.200.208.202.34370: 51345 ServFail 0/0/0 (64) | 13:42 |
| *** subscope has joined #openstack-ansible | 13:43 | |
| hughsaunders | ipv6 resolution failure | 13:43 |
| *** subscope has quit IRC | 13:44 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-security: Add theme fix for browsable source code https://review.openstack.org/238045 | 13:45 |
| ashishjain | hughsaunders: okay so this should not be considered | 13:45 |
| *** KLevenstein has quit IRC | 13:45 | |
| tiagogomes__ | Hi, shouldn't keystone use :q | 13:46 |
| ashishjain | hughsaunders: actually there was an issue with osad because os using the same eth12 interface for container as well as compute and hence has to use host_ovveride option. I think I will have to make some changes to the playbook and rerun the neutron playbook. | 13:46 |
| tiagogomes__ | let's try again | 13:46 |
| ashishjain | I will ping you later in the night if you are available | 13:47 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-specs: Add theme fix for browsable source code https://review.openstack.org/238061 | 13:47 |
| ashishjain | I mean my night time | 13:47 |
| tiagogomes__ | Hi, shouldn't keystone use policy.v3cloudsample.json for the policy file? Right now, I can't do some operations as admin | 13:47 |
| tiagogomes__ | So I guess I need to be the cloud admin | 13:47 |
| *** sdake_ has joined #openstack-ansible | 13:47 | |
| *** dolpher has quit IRC | 13:48 | |
| ashishjain | hughsaunders: Thanks for spending so much of time, I think I have moved one more step but I feel still a long way to go | 13:48 |
| *** dolpher has joined #openstack-ansible | 13:49 | |
| hughsaunders | ashishjain: yeah, networking takes a while to figure.. #openstack may be able to help if you have networking questions that aren't osad specific | 13:51 |
| *** sdake has quit IRC | 13:51 | |
| *** ashishjain has quit IRC | 13:52 | |
| hughsaunders | tiagogomes__: which operation are you denied? | 13:54 |
| tiagogomes__ | hughsaunders Adding a _member_ role to an user | 13:54 |
| *** sdake_ has quit IRC | 13:57 | |
| *** sdake has joined #openstack-ansible | 13:58 | |
| hughsaunders | tiagogomes__: wfm with default ks policy | 13:58 |
| hughsaunders | also default openrc from utility container | 13:59 |
| *** cbits has left #openstack-ansible | 13:59 | |
| tiagogomes__ | hughsaunders are you testing with multiple domains? | 13:59 |
| *** mgoddard_ has joined #openstack-ansible | 14:00 | |
| *** persia has quit IRC | 14:02 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-security: Do not check doc folder in pep8 check https://review.openstack.org/238104 | 14:03 |
| *** mgoddard has quit IRC | 14:03 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-security: Add theme fix for browsable source code https://review.openstack.org/238045 | 14:04 |
| hughsaunders | tiagogomes__: just retested with a new domain, project and user within that, then assigned the _member_ role to the new user. | 14:04 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-security: Do not check doc folder in pep8 check https://review.openstack.org/238104 | 14:04 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-security: Add theme fix for browsable source code https://review.openstack.org/238045 | 14:05 |
| *** mnestheu1 has joined #openstack-ansible | 14:05 | |
| *** mnestheu1 is now known as scarlisle | 14:05 | |
| tiagogomes__ | hughsaunders thanks for testing. Something looks to off on my setup, I'll look into that | 14:05 |
| *** sigmavirus24_awa is now known as sigmavirus24 | 14:06 | |
| *** tristanC has quit IRC | 14:07 | |
| openstackgerrit | Major Hayden proposed openstack/openstack-ansible-security: Updating getting started docs https://review.openstack.org/236066 | 14:07 |
| mhayden | odyssey4me: ^^ that should shore up the docs | 14:08 |
| *** jongchoi has joined #openstack-ansible | 14:08 | |
| *** jwagner_away is now known as jwagner | 14:09 | |
| *** persia has joined #openstack-ansible | 14:10 | |
| *** persia has quit IRC | 14:10 | |
| *** persia has joined #openstack-ansible | 14:10 | |
| *** sdake has quit IRC | 14:12 | |
| *** daneyon has quit IRC | 14:12 | |
| *** daneyon has joined #openstack-ansible | 14:14 | |
| *** galstrom is now known as galstrom_zzz | 14:17 | |
| tiagogomes__ | is there a playbook to clear all the databases used in OpenStack? | 14:18 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-security: Do not check doc folder in pep8 check https://review.openstack.org/238104 | 14:19 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Add theme fix for browsable source code https://review.openstack.org/238042 | 14:20 |
| hughsaunders | tiagogomes__: not that I'm aware of. You could delete all the DBs manually or delete the db containers and rebuild them via ansible | 14:20 |
| *** k_stev has joined #openstack-ansible | 14:20 | |
| tiagogomes__ | ok, but a playbook to that could be handy | 14:23 |
| tiagogomes__ | I saw a patch to add multidomain support to Horizon, but AFAIK horizon doesn't support it yet https://wiki.openstack.org/wiki/Horizon/DomainWorkFlow | 14:24 |
| *** galstrom_zzz is now known as galstrom | 14:24 | |
| *** Mudpuppy has joined #openstack-ansible | 14:27 | |
| *** Mudpuppy has quit IRC | 14:27 | |
| *** Mudpuppy has joined #openstack-ansible | 14:28 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-security: Add theme fix for browsable source code https://review.openstack.org/238045 | 14:28 |
| *** jongchoi has quit IRC | 14:28 | |
| odyssey4me | tiagogomes__ the wiki is the worst place for up to date information | 14:30 |
| tiagogomes__ | odyssey4me maybe, but that wiki has links to patches... that aren't merged yet :) | 14:31 |
| odyssey4me | tiagogomes__ well, the multi-domain support is only enabled explicitly anyway - so it's available for experimentation to see if it supports everything you want it to | 14:32 |
| tiagogomes__ | odyssey4me I see. So shouldn't used on production | 14:34 |
| *** galstrom is now known as galstrom_zzz | 14:37 | |
| cloudnull | morning | 14:38 |
| cloudnull | odyssey4me: so whats the verdict ? | 14:40 |
| cloudnull | to role or not to role , that is the question ?:) | 14:40 |
| odyssey4me | cloudnull haven't had a chance to chat with anyone yet - it appears that everyone is busy | 14:41 |
| odyssey4me | palendae ping? | 14:41 |
| odyssey4me | hughsaunders mattt ping? | 14:41 |
| hughsaunders | sup odyssey4me | 14:41 |
| *** phalmos has joined #openstack-ansible | 14:41 | |
| odyssey4me | with regards to https://review.openstack.org/230716 and https://review.openstack.org/236183 - would you feel more comfortable voting these through if they were put into a role? | 14:42 |
| cloudnull | i can move the wheel build process into the repo-server role or create a new one if needed | 14:43 |
| cloudnull | ditto for shippable venv | 14:43 |
| *** galstrom_zzz is now known as galstrom | 14:43 | |
| palendae | openstackgerrit: pong, I wasn't feeling the best when I got up | 14:44 |
| openstackgerrit | Merged openstack/openstack-ansible: Fix Install Guide's TOC https://review.openstack.org/237996 | 14:44 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Implement L3HA support https://review.openstack.org/233389 | 14:45 |
| hughsaunders | That would mean the template snippets could be split out | 14:46 |
| *** mgoddard has joined #openstack-ansible | 14:47 | |
| odyssey4me | hughsaunders yeah, and the tasks in the play could mostly be split into meaningful task files in the repo-build role - or another role | 14:47 |
| *** mgoddard_ has quit IRC | 14:47 | |
| palendae | cloudnull, odyssey4me My vote would be for a role, personally. Having a bunch of templates and scripts, even if they're small, in the same file, is kind of unwieldy in my mind | 14:47 |
| palendae | Whether a new role or existing isn't as important | 14:48 |
| cloudnull | palendae: a new role ? or moved into the repo-sever role ? | 14:48 |
| cloudnull | whats your preference ? | 14:48 |
| palendae | ^^ :) | 14:48 |
| hughsaunders | my hesitancy with shippable venvs is I'm not sure which direction we're going in - will we end up with image based deploys and not need to ship venvs? should we get replace existing pip installs with shipped venvs rather than adding another path? | 14:48 |
| palendae | cloudnull: Probably new, since the repo-server is really about setting up nginx and lsync, right? | 14:48 |
| hughsaunders | s/get// | 14:48 |
| cloudnull | palendae: yes | 14:48 |
| cloudnull | hughsaunders: i dont think image based deployments will replace shippable venvs especially considering we have several services run on hosts | 14:49 |
| *** k_stev has quit IRC | 14:49 | |
| palendae | hughsaunders: I have a feeling image-based deployments will be an optional thing | 14:49 |
| cloudnull | eg swift, cinder, nova, neutron (lxb agent) | 14:49 |
| odyssey4me | hughsaunders I think that shippable venvs will be around for at least the liberty lifetime. | 14:49 |
| palendae | For example, if you run 100% on metal, image-based deploys don't help | 14:49 |
| cloudnull | ^ that | 14:50 |
| odyssey4me | realistically we'll only be able to build an aimage-based deployment mechanism into Mitaka, and then *may* consider backporting it. | 14:50 |
| odyssey4me | then there is also the factor is mixed styles, as palendae and cloudnull have already intimated | 14:50 |
| palendae | That's if we agree it's a good idea, too; afaik, we're not quite decided yet | 14:50 |
| cloudnull | and even with the images we create, we can seed them with a shippable venv | 14:50 |
| hughsaunders | so should we remove the direct pip install eventually? | 14:51 |
| cloudnull | my 2cents | 14:51 |
| odyssey4me | I still see the shippable venv as useful, even for images - primarily due to conflicts in apt-python vs wheel packages | 14:51 |
| cloudnull | hughsaunders: id say no. having direct pip install is a "fall back option" if shippable venv goes in. | 14:51 |
| cloudnull | however i do think we'll end up refactoring some of our approach here. odyssey4me mentioned before that maybe we need a "crud" role for rabbit, db, venv deploy, etc... per OS_* service | 14:53 |
| cloudnull | and that may be where we head with all of this. but idk at this point. | 14:54 |
| hughsaunders | I think a role would be useful in the 'client' side of shippable venvs - so we don't have to duplicate the block of 5 tasks into all the os roles | 14:54 |
| odyssey4me | hughsaunders agreed - although it's a little late to get that done for liberty - I think we should explore that for mitaka | 14:55 |
| mattt | odyssey4me: wasn't the review only put in 3 days ago? | 14:56 |
| odyssey4me | given that the shippable venv and updated repo build is done - the option we have for now is to refactor now, in the hope that we can merge it before fri - or we let it through with the agreement that it'll be refactored afterwards | 14:56 |
| odyssey4me | the trouble is that I'd like to have had a few days to validate stabilisation and for final bugfixing before release | 14:57 |
| cloudnull | mattt: yes and no. when i submitted that orginal review for shippable venvs i was trying to make it depend on all of the other venv reviews but that caused a mess | 14:58 |
| cloudnull | https://review.openstack.org/#/c/233792/ < this is the abandoned review for shippable venv | 14:58 |
| *** fawadkhaliq has joined #openstack-ansible | 14:58 | |
| mattt | cloudnull: ah ok | 14:59 |
| mattt | still seems like big features going in right at the end tho | 14:59 |
| cloudnull | obviously still not all that long ago . however its longer than 3 days. | 14:59 |
| mattt | just playing devil's advocate here :) | 14:59 |
| cloudnull | for sure. | 14:59 |
| cloudnull | advocate away sir. :) | 14:59 |
| odyssey4me | mattt sure, but the repo build process is an optimisation - not really a new feature | 14:59 |
| *** k_stev has joined #openstack-ansible | 15:00 | |
| palendae | s/feature/change/ | 15:00 |
| cloudnull | ^ it makes us not need yaprt any longer | 15:00 |
| odyssey4me | shippable venvs is a logical finalisation of the venv work | 15:00 |
| *** Mudpuppy_ has joined #openstack-ansible | 15:00 | |
| cloudnull | yaprt is a big piece of software which i'd like to deprecate in liberty if at all possible. | 15:00 |
| *** mpavone has quit IRC | 15:00 | |
| *** k_stev has quit IRC | 15:01 | |
| cloudnull | we can do just about everything that yaprt does in ansible using straight up pip | 15:01 |
| *** k_stev has joined #openstack-ansible | 15:01 | |
| odyssey4me | note that the updated repo build cuts the build timing down by around 30% - even more if you build on a host | 15:01 |
| *** Mudpuppy_ has quit IRC | 15:01 | |
| *** alejandrito has joined #openstack-ansible | 15:01 | |
| hughsaunders | Yep, and its nice not to have to dig through another repo to work out how the build process weorks | 15:01 |
| cloudnull | ^ main motivator | 15:02 |
| *** Mudpuppy has quit IRC | 15:02 | |
| *** alejandrito has quit IRC | 15:02 | |
| *** Mudpuppy has joined #openstack-ansible | 15:02 | |
| *** alejandrito has joined #openstack-ansible | 15:02 | |
| palendae | Yeah, I don't think anyone disagrees with those points from what I'm seeing. Mostly that it's an important change super late | 15:02 |
| *** jwagner is now known as jwagner_away | 15:03 | |
| odyssey4me | palendae yep, which is why I'm inclined to rather allow it through as-is with the agreement for a refactor early in mitaka | 15:03 |
| odyssey4me | the repo-build has been successfully gating almost every time from its initial submission early this month | 15:04 |
| *** phalmos has quit IRC | 15:04 | |
| *** fawadkhaliq has quit IRC | 15:04 | |
| *** greg_a has joined #openstack-ansible | 15:05 | |
| logan2 | am I understanding this failure correctly? http://paste.gentoolinux.info/gavoqayiya.md | 15:06 |
| hughsaunders | logan2: container_networks should not be {} | 15:07 |
| odyssey4me | mattt hughsaunders palendae what's the preference? block it, let it through as-is, or refactor now? | 15:11 |
| logan2 | well on this particular host there is no containers being deployed, and ansible does not need to configure any networking (it is a compute node that will use calico networking so no bridges or overlay network) | 15:11 |
| logan2 | it seems like the next task "- name: Set nova management bridge (is_metal no container network)" accounts for this but it is unreachable due to how "- name: Set nova management address (is_metal)" is set up | 15:11 |
| logan2 | so I think it is a bug because how would "Set nova management bridge (is_metal no container network)" ever be reached in any configuration? | 15:13 |
| *** phalmos has joined #openstack-ansible | 15:14 | |
| palendae | odyssey4me: My inclination is block or refactor now; it's a big change late, and letting it through then adds to technical debt. If it goes in, I'd rather it go in cleanly | 15:14 |
| cloudnull | palendae: refactor into a role ? | 15:15 |
| palendae | cloudnull: Yeah | 15:15 |
| odyssey4me | I'm inclined to agree. | 15:15 |
| cloudnull | im game , hughsaunders mattt ? | 15:15 |
| cloudnull | logan2: is the br-mgmt device on your compute nodes ? | 15:15 |
| odyssey4me | Neither are essential to the liberty release. If we can do it right and agree that it'll be a backported feature if it merges after liberty, then I don't think we've lost anything. | 15:15 |
| cloudnull | logan2: it should fall back to the ansible_ssh_address | 15:17 |
| cloudnull | https://github.com/openstack/openstack-ansible/blob/master/playbooks/os-nova-install.yml#L130-L137 | 15:17 |
| logan2 | no, the ip is on bond0, which is the IP configured in compute_hosts, so ansible_ssh_address is perfect | 15:17 |
| *** fawadkhaliq has joined #openstack-ansible | 15:17 | |
| *** fawadkhaliq has quit IRC | 15:17 | |
| cloudnull | is it dying on the conditional "hostvars[inventory_hostname]['container_networks']['container_address']['bridge'] is undefined" ? | 15:17 |
| logan2 | but it fails before it ever reaches that because "Set nova management address (is_metal)" kills the run before it reaches the fallback | 15:17 |
| palendae | odyssey4me: Agreed | 15:18 |
| logan2 | yep that is where it is dying | 15:18 |
| hughsaunders | +1 for refactor, possibly multiple roles. One for generating /serving the venvs and one for using them | 15:18 |
| cloudnull | palendae hughsaunders, ill get it done | 15:18 |
| hughsaunders | cloudnull: w00t | 15:18 |
| logan2 | i just added - hostvars[inventory_hostname]['container_networks']['container_address']['bridge'] is defined to "Set nova management address (is_metal)" and it worked as expected | 15:18 |
| palendae | (I was just talking about the repo stuff; I haven't looked at shippable venvs) | 15:19 |
| cloudnull | logan2: you mind making a PR for the change :) | 15:19 |
| cloudnull | palendae: the shippable venvs build process follows much of the same form as the new wheel build process | 15:20 |
| logan2 | yep will do. wanted to raise it here first since I am probably sort of an edge use case :) | 15:20 |
| cloudnull | smaller play , but same concept | 15:20 |
| cloudnull | it sounds like a logic bug | 15:20 |
| palendae | Ah. My general feeling is the same - if it's not ready in a clean form and it's not essential, I don't see a reason to rush it in | 15:20 |
| cloudnull | we could also add a conditional around, "management_address is defined" which would give you access to override it to something else skipping all of the logic steps all together. | 15:21 |
| odyssey4me | palendae mattt hughsaunders thanks for weighing in, I appreciate the points of view | 15:21 |
| logan2 | gotcha | 15:21 |
| odyssey4me | cloudnull thank you for being patient with me :) | 15:21 |
| *** galstrom is now known as galstrom_zzz | 15:24 | |
| cloudnull | odyssey4me: https://review.openstack.org/#/c/233389/ on that i've replied inline i can add the case statements to the various config files however its cleaner to set a fact | 15:25 |
| *** gardensh_ has joined #openstack-ansible | 15:28 | |
| *** gardensh_ has quit IRC | 15:28 | |
| *** gardensh_ has joined #openstack-ansible | 15:28 | |
| *** gardenshed has quit IRC | 15:30 | |
| *** dolpher has quit IRC | 15:32 | |
| odyssey4me | cloudnull I figured that was the reason. It works and I'm down with that, now that we have the reason. :) We should figure out a good pattern for when to do logic in variables vs tasks vs templates, but now is not the time for that. | 15:33 |
| *** dolpher has joined #openstack-ansible | 15:33 | |
| *** shausy has joined #openstack-ansible | 15:35 | |
| *** daneyon has quit IRC | 15:36 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Add theme fix for browsable source code https://review.openstack.org/238042 | 15:37 |
| odyssey4me | cloudnull it still needs an edit of the commit message | 15:38 |
| cloudnull | for what ? | 15:38 |
| cloudnull | ok | 15:38 |
| cloudnull | one sec | 15:39 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Implement L3HA support https://review.openstack.org/233389 | 15:39 |
| cloudnull | done | 15:39 |
| odyssey4me | cloudnull I'm wanting to put together an infra job that builds a changelog automatically and publishes it with the docs. This will be the best way, I think, to maintain this without it being too much of a burden. | 15:39 |
| cloudnull | sounds good to me | 15:39 |
| cloudnull | we should also see about having docs for the various versions | 15:40 |
| palendae | Yeah, a chngelog would help | 15:40 |
| odyssey4me | bascially I'm thinking that the change log should be the list of commits and their messages | 15:40 |
| palendae | Yeah, that sounds like the simplest approach | 15:40 |
| *** greg_a has quit IRC | 15:40 | |
| odyssey4me | cloudnull there are docs per branch already: http://docs.openstack.org/developer/openstack-ansible/kilo/ | 15:40 |
| cloudnull | ah fair enout | 15:40 |
| cloudnull | *enough | 15:40 |
| odyssey4me | so basically I've been thinking that we need to link those from the README per branch after cutting the branch | 15:40 |
| cloudnull | a change log using something like: git log --abbrev-commit --pretty=oneline --no-merges would be great | 15:41 |
| cloudnull | in this case git log --abbrev-commit --pretty=oneline --no-merges kilo...liberty | 15:41 |
| cloudnull | etc... | 15:42 |
| odyssey4me | yep, but then the sha should link to the actual commit so that someone can inspect it if they want to | 15:42 |
| odyssey4me | so we need to be a lot better at our commit titles and messages | 15:43 |
| *** mgoddard_ has joined #openstack-ansible | 15:46 | |
| *** gardensh_ has quit IRC | 15:46 | |
| *** gardenshed has joined #openstack-ansible | 15:47 | |
| *** gardenshed has quit IRC | 15:48 | |
| *** gardenshed has joined #openstack-ansible | 15:48 | |
| *** mgoddard has quit IRC | 15:49 | |
| d34dh0r53 | odyssey4me: are you still targeting Friday for 10.1.16? | 15:50 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Implement L3HA support https://review.openstack.org/233389 | 15:52 |
| odyssey4me | cloudnull ^ updated the commit message | 15:53 |
| *** fawadkhaliq has joined #openstack-ansible | 15:53 | |
| b3rnard0 | odyssey4me: is there a target date for 10.2.0? | 15:53 |
| cloudnull | is 10.2 a thing ? | 15:53 |
| odyssey4me | d34dh0r53 I'm happy to release 10.1.16 now, unless there are still patches in flight that you need in there. | 15:53 |
| b3rnard0 | cloudnull: there's a milestone page for it | 15:53 |
| odyssey4me | b3rnard0 10.2.0 is a placeholder for new bugs, which is why it doesn't have a date on it | 15:53 |
| cloudnull | the question still stnads | 15:53 |
| cloudnull | :) | 15:53 |
| odyssey4me | b3rnard0 when bugs are actually in review, then get moved to the next milestone | 15:54 |
| b3rnard0 | okay | 15:54 |
| odyssey4me | b3rnard0 this is in the absence of a better tool for not losing bugs | 15:54 |
| b3rnard0 | so we'll likely have a 10.1.17 then? | 15:54 |
| odyssey4me | b3rnard0 yes, and a 10.1.18 after that :) | 15:55 |
| b3rnard0 | okie dokie, that clarifies it | 15:55 |
| *** fawadk has joined #openstack-ansible | 15:55 | |
| *** fawadkhaliq has quit IRC | 15:56 | |
| *** shausy has quit IRC | 15:57 | |
| *** mgoddard_ has quit IRC | 15:57 | |
| *** mgoddard has joined #openstack-ansible | 15:57 | |
| *** greg_a has joined #openstack-ansible | 16:03 | |
| d34dh0r53 | odyssey4me: so the neutron sha needs to be bumped in order to fix a race condition when hard rebooting bug and Charles was wondering if that would make it in 10.1.16 | 16:04 |
| d34dh0r53 | odyssey4me: neutron sha needs to include https://review.openstack.org/#/c/209708/ | 16:05 |
| d34dh0r53 | scratch that | 16:06 |
| *** dmsimard is now known as dmsimard|tokyo | 16:09 | |
| *** gparaskevas has quit IRC | 16:12 | |
| *** gardenshed has quit IRC | 16:14 | |
| *** jwagner_away is now known as jwagner | 16:14 | |
| *** dolpher has quit IRC | 16:19 | |
| *** logan2 has quit IRC | 16:27 | |
| *** galstrom_zzz is now known as galstrom | 16:33 | |
| odyssey4me | lol d34dh0r53 would you like to start again? ;) | 16:36 |
| odyssey4me | the arp protection was included in the last sha bump | 16:36 |
| d34dh0r53 | yeah, confuserated by support | 16:37 |
| d34dh0r53 | never mind anything I've said this morning | 16:37 |
| *** CheKoLyN has joined #openstack-ansible | 16:37 | |
| odyssey4me | and this makes it configurable: https://review.openstack.org/227963 | 16:37 |
| *** elo has quit IRC | 16:37 | |
| odyssey4me | also, all those fixes picked up by cloudnull are merged already | 16:38 |
| cloudnull | odyssey4me: https://review.openstack.org/#/c/236151/ this one is still pending | 16:39 |
| odyssey4me | d34dh0r53 the only patches currently in flight that could go in - if someone tests them - are https://review.openstack.org/233806 and https://review.openstack.org/226621 | 16:39 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Updated the repo-build process https://review.openstack.org/230716 | 16:39 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Backport ulimit setting to Juno https://review.openstack.org/236151 | 16:39 |
| cloudnull | hughsaunders mattt palendae odyssey4me https://review.openstack.org/230716 is now using a repo build role | 16:40 |
| odyssey4me | yes, and that one | 16:40 |
| palendae | cloudnull: Great, will look at it in a bit | 16:40 |
| d34dh0r53 | has https://review.openstack.org/236151 been testing upgrading from 10.1.15 to 10.1.16? | 16:41 |
| odyssey4me | d34dh0r53 not by me | 16:41 |
| d34dh0r53 | odyssey4me: ok | 16:42 |
| d34dh0r53 | odyssey4me: I'll look at that today | 16:42 |
| *** jwagner is now known as jwagner_away | 16:45 | |
| *** gardenshed has joined #openstack-ansible | 16:46 | |
| *** logan2 has joined #openstack-ansible | 16:47 | |
| *** elo has joined #openstack-ansible | 16:50 | |
| *** gardenshed has quit IRC | 16:54 | |
| *** gardenshed has joined #openstack-ansible | 17:02 | |
| * tiagogomes__ is confused by the plenitude of nova passwords, although if I am interpreting correctly the code only one is being used | 17:10 | |
| cloudnull | d34dh0r53: likely no. | 17:16 |
| cloudnull | that said, that change is one of the simpler changes to be backported to juno | 17:17 |
| cloudnull | tiagogomes__: which passwords are you looking into ? | 17:18 |
| tiagogomes__ | nova_ec2_service_password nova_v21_service_password nova_v3_service_password nova_service_password | 17:18 |
| cloudnull | in master? | 17:18 |
| tiagogomes__ | Aren't the passwords associated with an user? | 17:18 |
| tiagogomes__ | In the Kilo branch | 17:19 |
| *** gardenshed has quit IRC | 17:21 | |
| cloudnull | looks like they're used scripts/run-upgrade-old.sh:283:nova_v21_service_password: etc/openstack_deploy/user_secrets.yml:77:nova_v3_service_password: etc/openstack_deploy/user_secrets.yml:75:nova_ec2_service_password: but could likely be removed . all of those services are deprecated in kilo and removed in liberty . | 17:22 |
| tiagogomes__ | yes, but I still need to define a nova_v21_service_password for example, as nova_v21_enabled is enabled by default | 17:23 |
| cloudnull | additionally they all target the same service_user_name: "{{ nova_service_user_name }}" so they may just be remnence from days long past :) | 17:23 |
| odyssey4me | yeah, that sounds like a bad copy/paste | 17:24 |
| odyssey4me | cloudnull reviewed https://review.openstack.org/230716 - some comments there | 17:24 |
| odyssey4me | miguelgrinberg are you ready to rescue https://review.openstack.org/235978 ? | 17:29 |
| miguelgrinberg | odyssey4me: yeah, I'm going to try to have it done before I leave for Tokyo | 17:30 |
| odyssey4me | thanks miguelgrinberg ideally I'd like to have it done for the liberty release... which would be on friday :/ | 17:31 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Updated the repo-build process https://review.openstack.org/230716 | 17:31 |
| miguelgrinberg | odyssey4me: I leave Friday morning, so yes, should be done by then | 17:31 |
| *** ggillies has quit IRC | 17:31 | |
| palendae | Yeah, have a feeling anything not resolved by Friday will be hanging | 17:32 |
| odyssey4me | miguelgrinberg awesome, I really appreciate the help - we can do it! | 17:32 |
| *** ggillies has joined #openstack-ansible | 17:32 | |
| palendae | I know the laptop I'm taking to Tokyo will be pretty restricted | 17:32 |
| palendae | Purposefully not putting many SSH keys or passwords or anything on it | 17:32 |
| cloudnull | odyssey4me: https://review.openstack.org/230716 updated based on your comments | 17:32 |
| odyssey4me | I do find the results for https://review.openstack.org/205192 rather curious, but I think that heat is a contributor. | 17:32 |
| openstackgerrit | Merged openstack/openstack-ansible-specs: Add Liberty Release spec https://review.openstack.org/221189 | 17:33 |
| *** persia has quit IRC | 17:35 | |
| *** persia has joined #openstack-ansible | 17:36 | |
| *** persia has quit IRC | 17:36 | |
| *** persia has joined #openstack-ansible | 17:36 | |
| *** KLevenstein has joined #openstack-ansible | 17:38 | |
| *** metral is now known as metral_zzz | 17:40 | |
| *** metral_zzz is now known as metral | 17:48 | |
| *** g3rms_ has joined #openstack-ansible | 17:49 | |
| *** karimb has quit IRC | 17:51 | |
| *** harlowja has quit IRC | 17:57 | |
| odyssey4me | cloudnull back at you | 17:57 |
| *** daneyon has joined #openstack-ansible | 17:57 | |
| *** harlowja has joined #openstack-ansible | 18:02 | |
| *** jwagner_away is now known as jwagner | 18:03 | |
| *** k_stev has quit IRC | 18:14 | |
| *** pabelanger has quit IRC | 18:19 | |
| *** then3rd has quit IRC | 18:19 | |
| *** pabelanger has joined #openstack-ansible | 18:20 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Implement shippable venvs https://review.openstack.org/236183 | 18:29 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Updated the repo-build process https://review.openstack.org/230716 | 18:29 |
| *** KLevenstein has quit IRC | 18:30 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Updated the repo-build process https://review.openstack.org/230716 | 18:31 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Implement shippable venvs https://review.openstack.org/236183 | 18:42 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Updated the repo-build process https://review.openstack.org/230716 | 18:42 |
| *** k_stev has joined #openstack-ansible | 18:44 | |
| *** k_stev has quit IRC | 18:44 | |
| *** k_stev has joined #openstack-ansible | 18:44 | |
| cloudnull | mattt hughsaunders palendae odyssey4me https://review.openstack.org/236183 and https://review.openstack.org/230716 updated to run within the build role | 18:51 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Implement shippable venvs https://review.openstack.org/236183 | 18:55 |
| palendae | cloudnull: Just so you're not blind sided, I'll probably reference that venv idea in the image-based deployment discussion | 19:02 |
| cloudnull | sweet! | 19:03 |
| palendae | Cause it's along similar lines | 19:03 |
| palendae | I think images would snag both python and OS packages, though | 19:04 |
| odyssey4me | palendae so, interestingly, apt packages sometimes conflict with wheels installed - for instance the requests/urllib thing | 19:05 |
| odyssey4me | so installing wheels into a venv, even in a packaged container, is good | 19:05 |
| palendae | odyssey4me: Sure | 19:05 |
| palendae | That's a really good point - these aren't necessarily exclusive discussions | 19:05 |
| palendae | But certainly related | 19:06 |
| odyssey4me | exactly :) | 19:06 |
| odyssey4me | it's a packaging discussion of sorts, to start with | 19:07 |
| palendae | It looks like the current work is mostly to 'contain' the OpenStack services themselves so our installs and upgrades are less painful | 19:07 |
| odyssey4me | after that it comes down to other complexities like how an image is transferred across the wire, laid down and how live traffic is transitioned from one to the other | 19:08 |
| palendae | Not necessarily underlying infra services like mariadb or rabbit, which are mostly OS packaged | 19:08 |
| palendae | I also have questions about where data would live for those infrastructure services | 19:08 |
| odyssey4me | palendae yes, our source-based deployment is specifically openstack - we provide the infra to do others, but the focus is openstac | 19:08 |
| palendae | I haven't looked at how Kolla does it, but my general understanding of Docker containers that have vital data is to have layered containers | 19:09 |
| palendae | Granted, we're doing LXC, but approaches shouldn't be too wildly different | 19:09 |
| odyssey4me | the layering has nothing to do with docker, and everything to do with file syslems like overlayfs | 19:09 |
| palendae | Because, for example, MariaDB and RabbitMQ are not microservices, and can't just come and go | 19:09 |
| palendae | odyssey4me: Docker implements some intelligence around when to automatically delete bindmounts | 19:10 |
| odyssey4me | you overlay a service install on top of a base OS | 19:10 |
| palendae | THat's what I'm referring to | 19:10 |
| palendae | That's a separate discussion | 19:10 |
| odyssey4me | ok, that's a whole different thing | 19:10 |
| palendae | I'm talking about how MariaDB or Glance or Cinder don't lose their data when updating the container | 19:10 |
| palendae | Right | 19:10 |
| palendae | Docker uses a layered approach to keep the bindmounts alive | 19:10 |
| odyssey4me | ok, maybe that'd be possible with lxc too - we just haven't looked into it | 19:11 |
| palendae | So you end up with 2 containers - a glance container that runs the service, and a glance container that just sits around to hold the bindmount | 19:11 |
| palendae | Right | 19:11 |
| odyssey4me | ah, that's how - of course | 19:11 |
| palendae | Which is what the discussion is intended to ferret out. | 19:11 |
| odyssey4me | three milion containers for a service | 19:11 |
| palendae | https://docs.docker.com/userguide/dockervolumes/ | 19:11 |
| palendae | Yeah | 19:11 |
| odyssey4me | that's useful :p | 19:11 |
| palendae | I'm not sure I like it | 19:12 |
| palendae | But that's how they do it | 19:12 |
| odyssey4me | so how about those venvs? | 19:12 |
| palendae | I have the review open :) | 19:12 |
| cloudnull | yup its something akin to having to restart several containers to ensure an app update for a single service which will ba an operational nightmare | 19:12 |
| odyssey4me | if the container is essentially a service, then what is the point of the container | 19:13 |
| odyssey4me | why not just run the service on the OS and use the venv for package isolation? | 19:13 |
| *** daneyon has quit IRC | 19:14 | |
| palendae | odyssey4me: ¯\_(ツ)_/¯ | 19:14 |
| cloudnull | ^ | 19:14 |
| odyssey4me | hahahaha | 19:14 |
| odyssey4me | so, image-based deployment | 19:14 |
| palendae | The thing that's always made me curious about this and stuff like the 12 factor app was - where the hell does the data stay? | 19:15 |
| palendae | Data is not short lived | 19:15 |
| palendae | All well and good to say your web app can just spin up containers, but we're not doing a simple web app here | 19:15 |
| odyssey4me | in a bind mount, which needs another container, which talks through a pipe, which needs another socket | 19:15 |
| palendae | Yeah, all seems very Rube Goldberg to me | 19:16 |
| odyssey4me | fuckit - just store the damned data on shared/distributed storage if you really care about it | 19:16 |
| palendae | Well, yes | 19:16 |
| palendae | Ideally :) | 19:16 |
| odyssey4me | if you don't, trash it | 19:16 |
| palendae | But again, we care about our OpenStack service DBs | 19:16 |
| odyssey4me | which already bind mount to a host | 19:16 |
| odyssey4me | we can already trash the container and replace it | 19:17 |
| palendae | Ah, yeah, I see what you're saying | 19:18 |
| palendae | Yep | 19:18 |
| mgariepy | hey, do you have a good way to replace the resolv.conf from the lxc cache file ? | 19:20 |
| mgariepy | this https://review.openstack.org/#/c/224304/ gives me some trouble.. | 19:20 |
| mgariepy | since 69.20.0.164 dns is not working for me. | 19:21 |
| mgariepy | and this : http://paste.ubuntu.com/12887848/ is quite ugly ;) haha | 19:23 |
| palendae | cloudnull: made a comment on the venv diff - latest patch set has git markers and failed linting | 19:24 |
| odyssey4me | mgariepy so that should only run against a fresh container, ie a fresh build | 19:27 |
| odyssey4me | mgariepy also, perhaps the dns servers for the containers should be configurable | 19:27 |
| mgariepy | yeah i know, but it fail on a fresh build, and re-runing it get it through, but my containers are not updated.. | 19:28 |
| cloudnull | mgariepy: you could update the cache and in container resolve.conf using a straight ansible command . but yes perhaps we should do a better job making that more configurable. | 19:28 |
| mgariepy | or copy the hosts resolv.conf ? | 19:28 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Implement shippable venvs https://review.openstack.org/236183 | 19:30 |
| *** fawadkhaliq has joined #openstack-ansible | 19:31 | |
| *** fawadk has quit IRC | 19:32 | |
| mgariepy | i'll test a quick patch and upload a review. | 19:32 |
| *** fawadk has joined #openstack-ansible | 19:40 | |
| *** fawadkhaliq has quit IRC | 19:43 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Make the container cache resolvers configurable https://review.openstack.org/238223 | 19:43 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Make the container cache resolvers configurable https://review.openstack.org/238223 | 19:44 |
| cloudnull | mgariepy: you mind testing those ^ | 19:44 |
| mgariepy | i will ;) | 19:49 |
| mgariepy | you guys are too fast for me. haha | 19:49 |
| cloudnull | sorry , i turned away from the irc window | 19:49 |
| cloudnull | or i would have seen your orginial message that you we're going to put up a review | 19:50 |
| *** harlowja has quit IRC | 19:50 | |
| mgariepy | yeah but my ansible knowledge is not that great ;) | 19:50 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Make the container cache resolvers configurable https://review.openstack.org/238223 | 19:55 |
| cloudnull | mgariepy: so that one covers the resolvconf files as well as resolv.conf | 19:56 |
| cloudnull | which will make sure that the resolvers are correct even when a container restarts | 19:56 |
| mgariepy | well, once the container is started it's not a proble | 19:56 |
| mgariepy | problem** | 19:56 |
| cloudnull | for sure, because it routes back to the host . | 19:57 |
| cloudnull | but now it will be consistent | 19:57 |
| mgariepy | yeah ok, didn't saw that haha | 19:57 |
| mgariepy | i'll test your patch tomorrow morning. | 19:58 |
| cloudnull | thats great | 19:59 |
| openstackgerrit | Bjoern Teipel proposed openstack/openstack-ansible: Removing package patch versions from APT pinning https://review.openstack.org/238230 | 20:02 |
| *** openstackstatus has joined #openstack-ansible | 20:06 | |
| *** ChanServ sets mode: +v openstackstatus | 20:06 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Updated the repo-build process https://review.openstack.org/230716 | 20:09 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Implement shippable venvs https://review.openstack.org/236183 | 20:16 |
| *** jaypipes has quit IRC | 20:18 | |
| *** fawadk has quit IRC | 20:20 | |
| *** CheKoLyN has quit IRC | 20:25 | |
| *** KLevenstein has joined #openstack-ansible | 20:29 | |
| *** sdake has joined #openstack-ansible | 20:35 | |
| *** sdake_ has joined #openstack-ansible | 20:36 | |
| openstackgerrit | Nolan Brubaker proposed openstack/openstack-ansible: [WIP] Use full command when reporting upgrade failure https://review.openstack.org/237689 | 20:36 |
| *** harlowja has joined #openstack-ansible | 20:39 | |
| *** harlowja_ has joined #openstack-ansible | 20:40 | |
| *** sdake has quit IRC | 20:40 | |
| *** KLevenstein has quit IRC | 20:41 | |
| *** alejandrito has quit IRC | 20:42 | |
| *** harlowja has quit IRC | 20:44 | |
| *** sdake_ is now known as sdake | 20:44 | |
| *** sdake has quit IRC | 20:58 | |
| *** sdake has joined #openstack-ansible | 20:59 | |
| *** jongchoi_ has joined #openstack-ansible | 21:10 | |
| *** darrenc is now known as darrenc_afk | 21:17 | |
| openstackgerrit | Bjoern Teipel proposed openstack/openstack-ansible: Removing package patch versions from APT pinning https://review.openstack.org/238230 | 21:24 |
| *** galstrom is now known as galstrom_zzz | 21:25 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Implement shippable venvs https://review.openstack.org/236183 | 21:32 |
| cloudnull | im out for the day talk to you all later. | 21:33 |
| palendae | Later | 21:33 |
| cloudnull | oh palendae i reviewed https://review.openstack.org/237689 looks good , i'd say remove teh wip seems like a sensible change and looks good so far | 21:34 |
| palendae | cloudnull: I'm running one more test, so still WIP :) | 21:35 |
| cloudnull | if you want i can load that into my test runer | 21:35 |
| cloudnull | *runner | 21:35 |
| cloudnull | that can be the ref test for a few days | 21:35 |
| palendae | Sure | 21:36 |
| *** Mudpuppy_ has joined #openstack-ansible | 21:37 | |
| cloudnull | done | 21:38 |
| cloudnull | check email , next 5 tests will be done using that ref | 21:38 |
| cloudnull | refs/changes/37/237689/5 | 21:38 |
| *** Mudpuppy_ has quit IRC | 21:38 | |
| cloudnull | you can watch progress here http://upgrade-test.cloudnull.io/ | 21:39 |
| cloudnull | top 5 logs | 21:39 |
| *** darrenc_afk is now known as darrenc | 21:39 | |
| palendae | Thanks | 21:39 |
| *** Mudpuppy has quit IRC | 21:40 | |
| *** abitha has joined #openstack-ansible | 21:45 | |
| *** jongchoi_ has quit IRC | 22:02 | |
| *** jongchoi_ has joined #openstack-ansible | 22:04 | |
| openstackgerrit | Merged openstack/openstack-ansible: Allow configration of config_drive_format inside nova.conf https://review.openstack.org/233806 | 22:07 |
| openstackgerrit | Merged openstack/openstack-ansible: Install Guide Cleanup https://review.openstack.org/238034 | 22:08 |
| *** jongchoi_ has quit IRC | 22:10 | |
| *** sdake_ has joined #openstack-ansible | 22:22 | |
| *** dolpher has joined #openstack-ansible | 22:23 | |
| *** sdake has quit IRC | 22:24 | |
| *** phalmos has quit IRC | 22:30 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 22:35 | |
| *** markvoelker has quit IRC | 23:08 | |
| *** k_stev has quit IRC | 23:14 | |
| *** jwagner is now known as jwagner_away | 23:29 | |
| *** sdake_ has quit IRC | 23:33 | |
| *** harlowja_ has quit IRC | 23:33 | |
| *** harlowja has joined #openstack-ansible | 23:37 | |
| *** harlowja_ has joined #openstack-ansible | 23:45 | |
| *** harlowja has quit IRC | 23:49 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!