Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel, 2014-07-03

| Channels | #fuel index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:11 crandquist joined #fuel
00:18 mattgriffin joined #fuel
00:38 IlyaE joined #fuel
01:19 xarses joined #fuel
01:27 IlyaE joined #fuel
02:56 mattgriffin joined #fuel
03:22 rongze joined #fuel
03:44 dhblaz joined #fuel
04:33 rongze_ joined #fuel
05:13 IlyaE joined #fuel
05:29 jonathan1 ok my next task is to get ssl working on horizon
06:12 IlyaE joined #fuel
06:19 jeremydei joined #fuel
06:36 guillaume_ joined #fuel
06:57 e0ne joined #fuel
07:00 kevein joined #fuel
07:04 artem_panchenko joined #fuel
07:05 e0ne joined #fuel
07:09 al_ex8 joined #fuel
07:11 e0ne joined #fuel
07:19 e0ne joined #fuel
07:24 e0ne joined #fuel
07:31 e0ne joined #fuel
07:39 e0ne joined #fuel
07:41 eshumakher joined #fuel
07:48 e0ne joined #fuel
07:59 e0ne joined #fuel
08:15 e0ne joined #fuel
08:47 monester joined #fuel
08:51 e0ne joined #fuel
08:55 brain461 joined #fuel
09:23 e0ne joined #fuel
09:35 geekinutah joined #fuel
09:37 e0ne joined #fuel
09:46 AndreyDanin_ joined #fuel
10:27 taj joined #fuel
10:36 brain461 joined #fuel
10:45 e0ne joined #fuel
10:54 e0ne joined #fuel
10:55 omelchek joined #fuel
10:59 scroiset joined #fuel
11:06 e0ne joined #fuel
11:23 e0ne joined #fuel
11:29 e0ne joined #fuel
11:37 e0ne joined #fuel
11:56 e0ne joined #fuel
12:15 e0ne joined #fuel
12:22 e0ne joined #fuel
12:29 rongze joined #fuel
12:33 e0ne joined #fuel
12:36 AndreyDanin_ joined #fuel
12:42 sallum joined #fuel
12:47 sallum Hi! I have a question related to fuel 5.0, I installed the master fuel node and when I tried to set the networks and click "verify networks", I get an error saying: "Error occurred while running method 'verify_networks'. Inspect Orchestrator logs for the details."
12:47 sallum It matches this bug: https://bugs.launchpad.net/fuel/+bug/1325347
12:48 sallum does anyone know any work-around to make it work until the 5.0.1 version is released?
13:02 e0ne joined #fuel
13:08 e0ne joined #fuel
13:33 e0ne joined #fuel
13:35 jaypipes joined #fuel
13:46 e0ne joined #fuel
13:46 TVR_ jonathan1  .. let me know how this works for you please....
13:47 TVR_ when I have (to date) tried, the only thing I manage to do is break it....
13:48 TVR_ Every resource I have read states it is all done in the local_settings.py ..
13:48 TVR_ in /usr/share/openstack-dashboard/openstack_dashboard/local/local_settings.py
14:14 AndreyDanin_ joined #fuel
14:17 wrale_ I think the docs are missing context here: http://docs.mirantis.com/openstack/fuel/fuel-5.0/reference-architecture.html#known-limitations  ... It says "To deploy OpenStack using Neutron with GRE segmentation, each node requires at least 2 NICs.".. It seems this is true in the case of Neutron VLAN and untrue in the case of Neutron GRE.  With GRE, can I not place both br-ex and the storage bridge on the same NIC?
14:20 TVR_ on eth0, I have the PXE network, on the second nic, I have everything else.... I tag a vlan for storage, tag a vlan for management, and go untagged for public (Neutron with GRE)
14:21 TVR_ I have this up in more that 5 sites across the US, so this works and is repeatable
14:22 TVR_ I also have instances with two public IP's on them from different subnets if anyone wants to use neutron with GRE and have more than one subnet for public IP space
14:22 mattgriffin joined #fuel
14:23 wrale_ TVR_ sounds good.. I'm planning to rebuild my nova-network install as GRE :  eth0 = PXE, eth0.5 = Management, eth1 = Public, eth2 (10GbE) = private, eth2.30 (10GbE) = storage
14:23 wrale_ and crossing my fingers that this will work
14:23 wrale_ because I'd like to use murano, and I have only the one physical 10GbE link per box
14:23 TVR_ I found that putting EVERYTHING except the PXE on the 10G switch makes it much faster
14:24 wrale_ glad to know.. hmm..
14:24 TVR_ I have ceiliometer, murano and savanna installed as well from the fuel setup
14:24 wrale_ cool.. i would like to keep public 1gbe to discourage use of public IPs
14:25 wrale_ it's a private cloud :)
14:25 TVR_ the issue is... the management network ~actually~ host the dashboard VIP... so that is why I keep it on the 10G nic
14:25 wrale_ ah.. didn't know that.. haproxy must allow access from public
14:25 wrale_ (guess)
14:26 wrale_ becuase pub dash is working for me and they are on separate nics
14:26 TVR_ you connect to the first public as the (public) x.x.x.x/dashboard... but in reality, that redirects you to (management IP) x.x.x.2/dashboard
14:27 wrale_ hmm.. i think i read that GRE tunnels go over the mgmt bridge without a separate private bridge
14:27 wrale_ so, maybe i need mgmt on 10gbe after all
14:28 wrale_ back to docs :)
14:29 TVR_ you can verify this by ifconfig | grep for management .2 address.... then ssh to that bos, then edit /usr/share/openstack-dashboard/openstack_dashboard/local/local_settings.py and change it to SITE_BRANDING = 'Mirantis OpenStack Dashboard --test' and then service httpd restart and then go to the dashboard....
14:29 TVR_ you will see the site header change...
14:30 TVR_ getting an instance to have two real IP's made me learn this shit to a whole new level.... and now that the onion has been peeled back, I like it\
14:31 wrale_ do you suppose neutron with GRE would be possible with discrete 1GbE ports for each network plus layer two unmanaged switches for each?
14:31 TVR_ let me know when you want to deal with heat templates as well.... that shit get's ....... interesting
14:31 wrale_ for cloud on cheap?
14:31 TVR_ heh
14:31 wrale_ TVR_ i appreciate the help.. i'll keep this in mind for sure
14:32 TVR_ you can do one nic per service if you want to isolate them all... if you have say, 4 nics
14:33 wrale_ cool.. i guess unmanaged switches are trunk mode by default, perhaps?
14:33 wrale_ on all ports
14:33 TVR_ the part to remember is, the storage network NEEDS the 10G network if you want the instances to have any good backend storage access speed
14:34 TVR_ yea.. like a trunk mode with one vlan.. vlan1
14:34 wrale_ agreed.. it would be more for a desktop-based frankencloud
14:34 wrale_ recycling
14:34 wrale_ as in slow = expected
14:35 TVR_ you doing HA?
14:35 TVR_ the way pacemaker works is interesting
14:35 TVR_ in a good way... but you have to learn how to play it's game
14:35 wrale_ well, separate from frankencloud ideas, i have 72 nodes of supermicro with the 1x10GbE in which i'm doing H/A... i have a small proxmox ve + ceph cluster running in h/a too
14:36 wrale_ which uses some of that pacemaker stuff
14:36 wrale_ fun stuff for sure lol
14:36 TVR_ ok.. so 3 controllers+ceph and all others are compute and compute+ceph?
14:36 wrale_ for openstack.. 3 controllers + celiometer + ceph mon stuff and the rest as compute ceph
14:37 wrale_ for proxmox i'm doing 5-node h/a
14:37 TVR_ ok.. cool...
14:37 TVR_ neutron agent-list and crm status are your best friend
14:37 wrale_ different tools for different uses i guess.. i don't know that openstack does automated h/a evacuation
14:38 wrale_ good to know.. not used those at all..lol
14:38 wrale_ yet
14:38 TVR_ ok... well let me know when you have a ha cluster up.. and I will give you a quick tutorial on those and what they mean
14:38 TVR_ brb
14:40 wrale_ hoping to build a new 5.0.1 iso today with the hopes that the "fix" from my bug is present within: https://bugs.launchpad.net/mos/5.0.x/+bug/1335628
14:41 wrale_ thanks.. i wonder if there is any fencing.. haven't seen it in my purusing the docs
14:44 TVR_ back
14:45 wrale_ here's a procedure i wrote to build the latest 5.0.1: https://gist.github.com/josh-wrale/72ac65127cfdf4f319f5
14:45 wrale_ supposing the repos themselves aren't broken
14:46 TVR_ awe shit... there be all vagrant up in the house...
14:46 TVR_ heh
14:46 wrale_ you have to read between the lines a little.. after vagrant ssh all is inside vm
14:47 wrale_ beats clean room ubuntu install all the time..lol
14:47 TVR_ I hear ya
14:49 wrale_ are your novnc consoles working in horizon?
14:49 wrale_ in 5.x?
14:49 TVR_ don't have 5.x in my production... only 4.1
14:50 TVR_ once the upgrade path is complete, we will update all the sites
14:50 wrale_ cool..also good to know.. i tried building 4.1 but no luck when slave install happened
14:50 wrale_ i heard there wouldn't be an upgrade path until 5.0.1
14:50 wrale_ glad to be wrong, though
14:50 TVR_ yes, so it is why I am waiting
14:50 wrale_ makes sense
14:58 IlyaE joined #fuel
15:08 jobewan joined #fuel
15:22 evg_ left #fuel
15:23 evg_ joined #fuel
15:37 e0ne joined #fuel
15:48 jaypipes_ joined #fuel
15:57 taj joined #fuel
16:01 angdraug joined #fuel
16:05 xarses joined #fuel
16:22 xarses_ joined #fuel
16:25 miroslav_ joined #fuel
16:27 dhblaz joined #fuel
16:34 taj joined #fuel
16:37 whywefight joined #fuel
16:37 whywefight left #fuel
16:38 AndreyDanin joined #fuel
17:20 odyssey4me joined #fuel
17:30 Kupo24z1 joined #fuel
17:33 monester joined #fuel
17:34 Kupo24z joined #fuel
17:49 wrale_ Here's the reference I was speaking about: "Open vSwitch (OVS) GRE tunnels are provided through Management Network." pg18 http://docs.mirantis.com/fuel/fuel-3.2.1/pdf/Mirantis-OpenStack-3.2.1-Pre-InstallationGuide.pdf
17:56 TVR_ yea... version 3.2.1 was long enough ago to where there is much info that may well be outdated there
17:57 thehybridtech joined #fuel
18:30 fhond Repeating my question
18:30 fhond Can someone help with with a basic question on Fuel?  I am first time user and wanted to find out how to set hostnames of the nodes discovered by fuel.  By default I get the names node-1, node-2 and so on but I would like them to have hostnames like, controller-01, controller-02, compute-01 and so on.  Is this possible and if so,  how would I do this?
18:31 TVR_ cobbler handles all that, so you would have to modify the behavior of cobbler to achieve your desired result.
18:32 fhond do I need to make the changes in the docker container running the cobbler for this to take effect?
18:33 fhond Also,  is there a way to run post configuration scripts on each of the nodes after they are configured with Openstack roles,  I want tosetup some static routes on each of the nodes
18:34 TVR_ why not just edit the /etc/sysconfig/network file after the fact... wouldn't that be much easier?
18:34 TVR_ add /etc/hosts entries to fuel server as well... after
18:36 fhond Thanks TVR_,  good idea.  I am tasked with creating an env with 50+ nodes.  I am going start off with a small scale setup to get my hostnames straights and then add the post config static routes to each of the nodes.
18:37 fhond Using Ubuntu for the compute hosts,  so I guess the location would be /etc/network/interfaces or /etc/network/interfaces.d/
18:38 TVR_ I think you may be looking at the "purpose" of fuel incorrectly.... fuel is designed to provision the systems, so an individual doesn't have to deal with all the quirks of all the moving parts.... if you need to modify many things, for esoteric or other reasons, grab all the puppet modules off of a fuel server, and then call them as puppet modules as needed. This will allow you to pass to cobbler the names you want, add routes, and / or d
18:39 TVR_ you may want to edit the templates and modules before you add nodes etc
18:40 TVR_ I find it easiest to find what I need to edit, and then grep for those files in the fuel servers modules and most likely, it is an .erb somewhere that controls it.
18:41 fhond got it.  With the docker containers,  finding it difficult to find the right file to edit,  guess it is unfamiliarity on my part witht the fuel and I am very new to docker stuff.  Gonna play around with it over the weekend
18:41 TVR_ I needed to have twin external networks, so I ran a default install, and then scripted the bridge mods, port adds, and after it all stabilized ran neutron commands to add what I needed for networks...
18:42 TVR_ the advantage there is, when I go to fuel 6.0 or whatever, the upgrade should be OK because my mods are after the fact
18:44 fhond thanks TVR_
18:48 e0ne joined #fuel
18:49 wrale_ i read in one of the guides that i should enable STP fast switching on my network switch.  if i disabled STP altogether, am I alright, or should i turn it on with fast switching? rhetorical question, i guess.. i read somewhere else that disabling stp was better
18:49 IlyaE joined #fuel
18:54 wrale_ pardon me, it says "stp edge port or a spanning tree port fast trunk"
18:55 TVR_ I believe the .erb to control the behavior of cobbler is /etc/puppet/modules/cobbler/templates/settings.erb
19:51 IlyaE joined #fuel
19:59 vidalinux joined #fuel
20:05 IlyaE joined #fuel
20:31 taj joined #fuel
20:39 dburmistrov joined #fuel
20:45 mattgriffin joined #fuel
20:48 IlyaE joined #fuel
21:22 CTWill joined #fuel
21:34 dhblaz joined #fuel
22:14 mattgriffin joined #fuel
22:14 dhblaz joined #fuel
22:25 IlyaE joined #fuel
22:35 dhblaz joined #fuel
23:14 eshumakher joined #fuel
23:17 bookwar joined #fuel

| Channels | #fuel index | Today | | Search | Google Search | Plain-Text | summary