Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel, 2014-08-29

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

All times shown according to UTC.

Time Nick Message
00:20 rmoe joined #fuel
01:11 mpetason joined #fuel
01:12 kupo24z On the fuel master, what container is dnsmasq hosted in? Im getting a ton of requests I need to look into on the server and its causing DNS timeouts
01:12 kupo24z [root@fuel ~]# wc -l /var/log/docker-logs/dnsmasq.log 12336812 /var/log/docker-logs/dnsmasq.log
01:23 mpetason joined #fuel
01:36 mattgriffin joined #fuel
02:02 angdraug joined #fuel
02:11 xarses joined #fuel
02:31 Longgeek joined #fuel
02:32 Longgeek joined #fuel
03:02 mpetason joined #fuel
03:12 Longgeek_ joined #fuel
04:06 Longgeek joined #fuel
04:21 ArminderS joined #fuel
05:29 Longgeek joined #fuel
05:57 Longgeek joined #fuel
06:07 HeOS joined #fuel
06:24 sbartel joined #fuel
06:25 e0ne joined #fuel
06:29 sbartel hello
06:29 sbartel Have a question about deployment
06:29 sbartel I have tried to modify the hostname of my nodes
06:29 sbartel to have something more friendly thant 'node-xx'
06:30 sbartel i made me modification un the yaml file of the deployment
06:30 sbartel but i should have miss something as I have an error during deployment
06:30 sbartel Node node-18 is not defined in the hash structure at /etc/puppet/manifests/site.pp:39 on node node-18.domain.tld
06:31 sbartel it seems it keep somewhere a reference to the old name
06:31 sbartel do you know what i have missed
06:41 dancn joined #fuel
06:55 Longgeek joined #fuel
06:56 monester_ joined #fuel
06:56 sbartel it seems do have been possible in fuel 3
06:58 strictlyb joined #fuel
07:00 pasquier-s joined #fuel
07:17 sc-rm_ evg: Yep, to me it looks like it’s only in the tests.
07:17 sc-rm_ evg: Because I can ssh into the instance through the assigned floating IP, so I would say it’s only the test.
07:20 HeOS joined #fuel
07:31 e0ne joined #fuel
07:41 pasquier-s joined #fuel
07:45 sbartel have tried to rename it also through the UI
07:45 sbartel but don't seem to be take in count
07:48 hyperbaba joined #fuel
07:49 e0ne joined #fuel
08:00 adanin joined #fuel
08:14 eshumkher joined #fuel
08:29 Skadefro joined #fuel
08:34 eshumkher joined #fuel
08:42 rmoe joined #fuel
08:42 kaliya sbartel, which file did you modify and with which modification?
08:43 sbartel kaliya: finally sucess to do it
08:43 sbartel by modifying the deployment yaml files
08:44 sbartel and the provisioning yaml files retrieves with the provisioning default commend of FUEL cli
08:48 kaliya sbartel, thanks I will ask the team if the modification can be documented also
08:48 sc-rm_ evg: Our setup right now is pre-production so I can test whatever patches there might be to see if it changes the outcome of the tests.
08:54 eshumkher joined #fuel
09:08 harybahh joined #fuel
09:09 harybahh Hello
09:10 harybahh I'm using Fuel to test several openstack deployments, and now I'm pretty sure about the production configuration I would like to proper deploy with reseted node numbers
09:10 harybahh Where could I reset node table and numbers without reinstall Fuel ?
09:17 strictlyb joined #fuel
09:17 Longgeek joined #fuel
09:17 ArminderS joined #fuel
09:17 boris-42 joined #fuel
09:17 fhond joined #fuel
09:17 Bomfunk__ joined #fuel
09:20 evg sc-rm_: hi, if the bug I refered to yesterday is your case there are no any patch yet. Please try run it another time.
09:21 evg sc-rm_: is it "Check network connectivity from instance via floating IP" that was falled?
09:21 wayneeseguin joined #fuel
09:23 geekinutah joined #fuel
09:23 sc-rm_ evg: yep, that is the failing one
09:24 sc-rm_ evg: but as i read the bug, it describes it as users are not able to connect to instances after asignment of a floating IP?
09:25 evg harybahh: please go to UI->your env->Action. There are "reset" and "delete" buttons there.
09:25 sc-rm_ evg: I also tried to run it multiple times and all witht the same result
09:27 ArminderS brb
09:27 harybahh evg: that's what I do, but in anew environement, node number starts from the last deployed node number, in a reseted and deleted env
09:27 harybahh there's a blueprin https://blueprints.launchpad.net/fuel/+spec/reset-count-from-one
09:28 harybahh but no solution provided
09:30 ykotko joined #fuel
09:30 brain461 joined #fuel
09:30 geekinutah joined #fuel
09:31 evg sc-rm_: but you _can_ connect to an instance?
09:31 sc-rm_ evg: Yep I can :-)
09:32 evg harybahh: yep, i've got it
09:38 sc-rm_ evg: also the tests for ceilometer fails even thought ceilometer is installed according to the settings
09:46 evg harybahh: may be you could clean the db tables and then reset the autoincrement counter. i'm not sure and i would actualy not try any hacks on a pre-production configuration
09:47 obcecado_ joined #fuel
09:48 eshumkher joined #fuel
09:48 IlluminatiJudas joined #fuel
09:51 bogdando joined #fuel
09:52 sressot joined #fuel
09:54 warpc__ joined #fuel
09:56 artem_panchenko_ joined #fuel
09:57 artem_panchenko_ left #fuel
09:57 MorAle_ joined #fuel
10:00 jseutter_ joined #fuel
10:01 evgeniyl_ joined #fuel
10:01 Samos123 joined #fuel
10:01 harybahh evg: you sugest me to reinstall Fuel master node ? :)
10:02 harybahh not a big deal but that's sad
10:03 harybahh another question, my nodes have a crappy software RAID chip, and Fuel sees the physical disks and not RAID volumes
10:04 harybahh How can I, either work on RAID volumes or configure through Fuel, software RAID volume ?
10:09 holser joined #fuel
10:10 izinovik joined #fuel
10:10 meow-nofer joined #fuel
10:10 xenolog joined #fuel
10:10 ddmitriev joined #fuel
10:11 monester joined #fuel
10:11 kaliya joined #fuel
10:11 bookwar joined #fuel
10:11 book` joined #fuel
10:12 sbog joined #fuel
10:13 vkramskikh joined #fuel
10:14 MiroslavAnashkin joined #fuel
10:15 evg joined #fuel
10:16 mihgen joined #fuel
10:17 apalkina joined #fuel
10:18 akislitsky joined #fuel
10:22 eshumkher joined #fuel
10:23 [HeOS] joined #fuel
10:25 aedocw_ joined #fuel
10:26 dburmistrov_ joined #fuel
10:34 mrasskazov joined #fuel
10:36 bookwar left #fuel
10:46 bookwar joined #fuel
10:51 KodiakF2 joined #fuel
10:56 kaliya joined #fuel
10:57 Llew joined #fuel
10:58 dshulyak joined #fuel
11:17 eshumkher joined #fuel
11:18 tuvenen__ joined #fuel
11:22 DaveJ__ Hi Guys - is it still possible to install RHOS using Fuel with the latest release ?  I seem to remember you used to be able to add your redhat subscription to do it
11:24 bogdando joined #fuel
11:29 gleam joined #fuel
11:29 sanek_ joined #fuel
11:33 t_dmitry joined #fuel
11:34 christopheraedo joined #fuel
11:41 sc-rm joined #fuel
11:41 kaliya DaveJ__, the master runs CentOS. Do you mean you want to add a RH subscription on it?
11:43 DaveJ__ kaliya: Yes - so I want to install RH on my Controller and Compute nodes as the base OS
11:49 getup- joined #fuel
11:56 kaliya DaveJ__, we supported RHEL for a while in node deployments, but now we don't anymore.
11:57 artem_panchenko_ joined #fuel
11:57 DaveJ__ kaliya: Thanks.  Maybe you could help me with another query?  We can kickstart machines in our lab with our own CentOS image, but when we use Fuel it installs - seems to be missing network drivers.  Is it possible to replace the copy of CentOS shipped with Fuel, with our own version?  Or has it been modified?
12:11 kaliya DaveJ__, the Fuel CentOS image includes some custom packages. If you try to substitute with your own OS img, you should rebuild and add those packages by yourself
12:13 ganso Hi guys. I am facing a weird problem here. I have 3 slaves which I have redeployed environments several times already. But now I noticed that I cannot reboot them, if I do, it cannot boot the kernel, it stops in busybox. For some reason, GRUB allows me to boot a previous kernel version (ubuntu 3.2) and it boots. I have a saved snapshot before rebooting and I
12:13 ganso can see it is running ubuntu 3.11. Has anyone faced this problem before? I believe fuel-master installs ubuntu 3.11 on my slaves when I deploy, since I did not upgrade them myself, but it does not make sense being able to boot 3.2 if I never had it.
12:13 fhond joined #fuel
12:14 Bomfunk joined #fuel
12:17 harybahh joined #fuel
12:19 jseutter joined #fuel
12:20 gleam joined #fuel
12:23 strictlyb joined #fuel
12:23 e0ne joined #fuel
12:34 sc-rm Is it possible to have sysstat and iptraff added to the repository, so it’s possible to install on nodes to debug performance problems?
12:37 eshumkher joined #fuel
12:58 Longgeek joined #fuel
13:24 evg sc-rm: You can add an external repository on your nodes. If you wish to see these packages in MOS you can launch a bug/blueprint on https://launchpad.net/fuel.
13:26 dancn left #fuel
13:31 MiroslavAnashkin DaveJ__: Which network driver is missing and what is your Fuel version?
13:35 eshumkher joined #fuel
13:37 DaveJ__ Its the latest version of Fuel 5.1.  I'll need to go check the console again.
13:37 DaveJ__ We were able to install with Fuel 4.0
13:37 DaveJ__ so it seems something changed inbetween
13:38 sc-rm evg: I’ll look into that :-)
13:39 MiroslavAnashkin One change we know is about TG3 driver. 5.x requires addition of longer timeout to wait until this NIC start on boot.
13:44 jaypipes joined #fuel
13:56 alexz joined #fuel
13:57 ianunruh hmmm... one of my controller nodes is stuck on "Running preseed" but all my compute nodes finished just fine
13:58 ianunruh I've even tried restarting that node and it's stuck on "Running preseed" again
14:01 mattgriffin joined #fuel
14:04 Longgeek joined #fuel
14:09 eshumkher joined #fuel
14:24 MiroslavAnashkin ianunruh: please check if it actually stalled on preseed stage. Please try to connect to this node via ssh or IPMI
14:25 ianunruh I spoke too soon, it finished, just took quite a while
14:36 mpetason joined #fuel
14:44 Llew joined #fuel
14:45 geekinutah joined #fuel
14:53 sanek joined #fuel
15:16 homegrown joined #fuel
15:21 harybahh_ joined #fuel
15:30 MiroslavAnashkin ianunruh: preseed makes all the big filesystems. It may take some time if you have machine with dozens of hard drives
15:40 eshumkher joined #fuel
15:41 kaliya joined #fuel
15:52 fandi joined #fuel
15:53 pasquier-s joined #fuel
15:53 mattgriffin joined #fuel
15:54 ArminderS joined #fuel
15:58 homegrown left #fuel
16:02 jobewan joined #fuel
16:02 rmoe joined #fuel
16:12 getup- joined #fuel
16:12 getup- joined #fuel
16:21 eshumkher joined #fuel
16:33 mutex joined #fuel
17:29 e0ne joined #fuel
17:36 boris-42 joined #fuel
17:45 e0ne joined #fuel
17:51 emagana joined #fuel
17:58 xarses joined #fuel
18:03 adanin joined #fuel
18:22 angdraug joined #fuel
18:25 e0ne joined #fuel
18:25 getup- joined #fuel
18:25 mattgriffin joined #fuel
18:38 kupo24z xarses: On the fuel master, what container is dnsmasq hosted in? Im getting a ton of requests I need to look into on the server and its causing DNS timeouts. [root@fuel ~]# wc -l /var/log/docker-logs/dnsmasq.log 12336812 /var/log/docker-logs/dnsmasq.log
18:39 xarses there is either a dnsmasq container, or its being run by cobbler
18:39 kupo24z Definetly not in docker, at least in 'docker ps'
18:39 xarses in 5.1 we should have a dockerctl list
18:40 xarses also you can ask supervisorctl for a list of what it's managing, all the containers are listed there
18:40 kupo24z Not in that list either
18:40 xarses so, then it should be in cobbler then
18:42 kupo24z yeh, so far i've gotten 7,538,659 requests from 172.17.42.1
18:52 emagana joined #fuel
18:59 emagana joined #fuel
19:02 e0ne joined #fuel
19:02 emagana joined #fuel
19:13 e0ne joined #fuel
19:20 e0ne joined #fuel
19:27 e0ne joined #fuel
19:30 kupo24z xarses: do you know if its possible to bind an additional floating ip to an instance aka ip aliasing? Or does it need a new network created on the instance with a matching private
19:30 xarses kupo24z: I would think its fine, but I don't know if i tried it
19:32 kupo24z Looks like it doesnt work via the traditional methods at least; Error: 409-{u'NeutronError': {u'message': u'Cannot associate floating IP 23.105.69.140 (1cff780d-52fa-4233-b3e3-3ac892da34ad) with port daf5d9ef-7444-42fd-8d13-50e8ffa98813 using fixed IP 192.168.111.41, as that fixed IP already has a floating IP on external network 84fa65c1-794d-4f10-b23c-c94ec9ba8a73.', u'type': u'FloatingIPPortAlreadyAssociated', u'detail': u''}}
19:33 xarses hmm, maybe it needs two fixed nic's then
19:43 kupo24z looks like this might do it https://ask.openstack.org/en/question/30690/add-multiple-specific-ips-to-instance/
19:43 e0ne joined #fuel
19:48 xarses fun =)
19:49 xarses looks about what you are looking for
19:51 kupo24z btw the dude that original coded the iops limits is working on a patch for rbd support https://review.openstack.org/#/c/117463/
19:51 kupo24z should be done soon
19:53 xarses cool
20:11 artem_panchenko joined #fuel
20:11 artem_panchenko left #fuel
20:25 mattgriffin joined #fuel
20:39 kupo24z xarses: when adding additional floating public ip ranges do you know the steps required? Can it be done post-deployment in fuel?
20:40 xarses if its within the same network as is the br-ex (public) yes,
20:40 xarses if its another network its more complicated, you have to create a new router, config file, and add another router service
20:40 kupo24z its a different range 23.105.69.0/24 vs 23.106.0.0/27 but its on the same vlan on the switch
20:41 xarses in Juno it will become less complicated (no seperate agent)
20:41 kupo24z and they use different gateways (.1 for each)
20:41 kupo24z would that be considered another network?
20:41 xarses yes
20:42 kupo24z If it could use the same gateway could it be avoided?
20:42 xarses yes
20:42 xarses ie a supernet
20:43 xarses someone in here set one up a month or two ago
20:58 kupo24z There any documentation on how to do it the hard way?
21:14 xarses yes, there is a slightly vauge docs.openstack.org mention of it
21:23 xarses kupo24z: http://docs.openstack.org/admin-guide-cloud/content/adv_cfg_l3_agent_multi_extnet.html
21:33 getup- joined #fuel
21:47 BillTheKat joined #fuel
21:47 Bomfunk_ joined #fuel
21:48 gleam_ joined #fuel
21:51 boris-42_ joined #fuel
21:54 kupo24z xarses: what is changing in juno?
21:54 blahRus joined #fuel
21:55 xarses I'd have to talk to the neutron guys again about it speficically, but it sounds like http://docs.openstack.org/admin-guide-cloud/content/adv_cfg_l3_agent_multi_floatip.html will be reduced or even removed, so you can have a single agent with multiple interfaces
21:55 xarses instead of the singular hard coded interface (external_network_bridge)
21:56 kupo24z Hmm okay
21:56 kupo24z I suppose we could just use a large enough block to not need to add anymore, but that isnt very efficent
22:07 xarses it seems that you can add on-link routes with out creating another l3-agent instance
22:09 leakypipes joined #fuel
22:10 mpetason joined #fuel
22:11 brain461 joined #fuel
22:11 ykotko joined #fuel
22:13 emagana joined #fuel
22:18 brain461 joined #fuel
22:21 BillTheKat joined #fuel
22:44 emagana joined #fuel
22:59 BillTheKat joined #fuel
23:09 BillTheKat joined #fuel
23:10 BillTheKat joined #fuel
23:11 kupo24z xarses: Is it possible to setup neutron to send outgoing traffic on each nova-compute node instead of going to the controller?
23:12 kupo24z Or is that only possible with flatDHCP?
23:12 mpetason http://docs.openstack.org/havana/install-guide/install/apt/content/section_neutron-single-flat.html
23:12 mpetason yeah, flat DHCP
23:12 mpetason unless you hacked it around and ran a L3 agent + others on each compute, then it could be possible with customization.
23:13 mpetason That’s not a clean way though.
23:13 mpetason we will await the expertise of xarses.
23:14 xarses you would be better off using nova networks with flat VLAN manager
23:14 xarses I think neutron can support this, but fuel won't do it for you.
23:17 emagana joined #fuel
23:17 kupo24z Whats the difference between nova-network flatDHCP and vlan manager?
23:17 kupo24z xarses: ^
23:19 xarses flatDHCP is a single dhcp network
23:20 xarses that all tenants share
23:20 xarses I think, but would have to check that vlan mode allows you to seperate tenant traffic, but I may be confusing this with neutron
23:23 kupo24z And you would be limited to 2048 tenants per env
23:23 kupo24z and would need to configure each vlan manually on the router when a tenant is made correct?
23:28 emagana joined #fuel
23:32 rmoe joined #fuel
23:34 kupo24z Ah looks like you need to use neutron for lbaas so nova-network wont work for us
23:35 kupo24z There a way to setup flat neutron at very easily?
23:36 BillTheKat joined #fuel
23:36 emagana joined #fuel
23:36 mpetason Neutron adds too much complexity if you are just looking for Flat DHCP. Nova-network for FlatDHCP removes that complexity and also removes more points of failure for networking.
23:36 xarses kupo24z: haven't tried, It shoudn't be too bad to change the config over after deploying with fuel
23:37 kupo24z xarses: which deployment would you recommend changing from?
23:37 xarses vlan
23:37 xarses probably
23:37 * xarses looks over the deployment guide
23:39 kupo24z Also these hosts only have 1NIC at 10gb
23:39 kupo24z afaik only GRE is compatible with that?
23:39 xarses nah, you could use vlan
23:39 kupo24z wouldnt that still have a 2096 tenant restriction?
23:42 xarses hmm, I can't tell easily I'd have to ask one of the neutron guys
23:43 xarses I'd guess you could do multiple flat networks
23:43 xarses but this outside what I normally work on
23:44 xarses it seems like you would loose per-tenant isolation
23:44 kupo24z Our main concern is a ddos or other large traffic overloading the controllers single point of failure
23:45 xarses well hopfully it would be negated further up the network than the openstack router
23:46 kupo24z So in the ideal neutron setup the controllers should have the same out of throughput as the ISP uplinks they are connected to?
23:46 emagana joined #fuel
23:48 kupo24z Is there a potential problem with overloading packets/s on the controllers with enough compute nodes?
23:50 emagana_ joined #fuel

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