Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel, 2014-08-25

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

All times shown according to UTC.

Time Nick Message
01:05 lordd_ joined #fuel
01:20 angdraug joined #fuel
03:05 lordd_ joined #fuel
04:08 ArminderS joined #fuel
05:06 lordd_ joined #fuel
05:33 lordd_ joined #fuel
05:55 nimion joined #fuel
06:37 adanin joined #fuel
06:46 eshumkher joined #fuel
06:57 pasquier-s joined #fuel
07:13 dancn joined #fuel
07:16 Longgeek joined #fuel
07:17 HeOS joined #fuel
07:24 warpc_ joined #fuel
07:32 bookwar joined #fuel
07:34 kaliya Havard: glad to know you've solved the issue :)
07:39 hyperbaba joined #fuel
08:03 lordd_ joined #fuel
09:06 adanin joined #fuel
09:08 tuvenen joined #fuel
09:24 t_dmitry joined #fuel
09:29 lordd_ joined #fuel
09:30 lordd_ joined #fuel
09:35 holser left #fuel
09:48 dancn joined #fuel
09:51 sressot joined #fuel
09:51 evg Havard: I'm glad to hear it! Actually you can't run any openstack services on the fuel master except as instal/confifure then by hand, please don't try.
10:05 e0ne joined #fuel
10:12 e0ne joined #fuel
10:20 e0ne joined #fuel
10:26 e0ne joined #fuel
10:32 e0ne joined #fuel
10:43 e0ne joined #fuel
10:48 e0ne joined #fuel
10:56 e0ne joined #fuel
11:01 kaliya joined #fuel
11:07 e0ne joined #fuel
11:12 e0ne joined #fuel
11:15 kaliya joined #fuel
11:19 e0ne joined #fuel
11:22 [HeOS] joined #fuel
11:25 kaliya joined #fuel
11:31 e0ne joined #fuel
11:54 e0ne joined #fuel
11:55 lordd_ joined #fuel
11:58 f13o joined #fuel
12:01 lordd_ joined #fuel
12:02 e0ne joined #fuel
12:13 pasquier-s joined #fuel
12:22 jaypipes joined #fuel
12:24 e0ne joined #fuel
12:29 youellet joined #fuel
12:37 rmoe joined #fuel
13:20 Havard evg: thanks a lot, that means that we might need to run the fuel on a separate VM and than use the cloud server as controller for the system. Is it possible to run HA with one controller? If that controller dies will everything be messed up, or will the nodes still work as normal?
13:20 Havard and kaliya and evg, thanks a lot for the assistant on the issue last week :-) It was great having that solved :-)
13:31 evg Havard: You can deploy HA with the only one controller (temporarily, due to lack of HW for example) but it wouldn't a real HA of course.
13:31 Havard evg: so two or more controllers is the best way to do it?
13:33 evg Havard: Your environment will be useless with out the controller node. See here the roles of the controller http://docs.openstack.org/grizzly/basic-install/apt/content/basic-install_controller.html
13:34 evg Havard: Three controllers nodes are recommended for production env
13:35 Havard evg: ok, so than they work as the same way as the ceph monitors, i have to check if we than can buy some kind of 1U twin nodes etc. :-) Will talk to my CTO after my vacation
13:36 Havard leaving to Crete for a week off on thursday :-P
13:36 evg Havard: here's a better link about roles http://docs.mirantis.com/openstack/fuel/fuel-5.0/pre-install-guide.html#nodes-and-roles
13:37 evg Havard: yes, something like ceph mons
13:38 evg Havard: You can use the only one controller for testing purposes of course
13:38 Havard evg: but as i understand the controller do some more tasks, like the control panel, database and API.
13:40 evg Havard: I like Crete too :)
13:40 evg Havard: yes, you're right
13:40 Havard evg: thanks a lot for your kind help :-) And for accepting me to be a greenhorn ;-)
13:40 Havard looking forward to that week in the heat on Crete ;-)
13:42 f13o Anyone installing fuel with libvirt/kvm ?
13:42 f13o Anything I should care about specifically ?
13:42 evg Havard: Have a nice holiday :)
13:42 Havard evg: thanks, a few days on meeting before we are leaving down there :-))
13:47 Chocobo joined #fuel
13:47 evg Havard: aah, you've still got a time to duck into fuel/openstack.
13:47 Havard evg: yeah, but not really since my L3 tech is away the three first days of this week
13:48 Chocobo Hi all, I am trying to deploy Fuel 5.0.1 on 5 nodes HA + Ceph and I get a "Timeout of Deployment is Exceeded"  Is there anything I can do to increase the timeout, or is it likely that I have something else configured incorrectly?   One of the HA nodes makes it to "Ready".
13:56 evg f13o: I use qemu/kvm, but without libvirt, don't think there is something specific there.
14:05 evg Chocobo: I shouldn't happened. I'd check the network the first.
14:08 miroslav_ joined #fuel
14:08 lordd_ joined #fuel
14:09 Chocobo evg: network verication passed.  Is there anything else I can do to help verify that everything is setup correctly?
14:12 Chocobo Is a working public gateway required?   Does the public network require access to the internet?  That is the only thing I can think of that may be off.   We are basically using the default configuration, except we changed the Admin (PXE) network during the initial install.
14:15 evg Chocobo: no, it's not the case
14:15 evg Chocobo: you don't have to have a working public gw nor internet access for public network
14:16 evg Chocobo: I've seen a bug https://bugs.launchpad.net/fuel/+bug/1360437
14:17 evg Chocobo: Isn't it your case?
14:20 Chocobo Well it doesn't hang on installing OpenStack on the first node.  The first node makes it to "Ready"  I will check to see if that iptables rule is missing.
14:22 xarses joined #fuel
14:24 evg Chocobo: aah, really.
14:25 evg Chocobo: it seems not your case
14:26 holser joined #fuel
14:26 evg Chocobo: May be you see something wrong in your logs?
14:29 mutex_ joined #fuel
14:29 warpc__ joined #fuel
14:31 Chocobo We are looking.  Not exactly sure which logs are most likely to contain the error.
14:31 Chocobo We are trying a redeployment right now.
14:32 aglarendil joined #fuel
14:32 bdudko joined #fuel
14:32 evg Chocobo: i think you should increase the timeout for deployment. Wait...
14:33 Chocobo we just got "neutron can not pre-fetch net list.  Neutron or keystone API not available"  hmmm.
14:33 lordd_ joined #fuel
14:33 accela-dev joined #fuel
14:34 monester_ joined #fuel
14:36 xarses Chocobo: as a notification? or an error?
14:37 Chocobo Error in puppet log.
14:38 xarses on the first controller?
14:41 evg Chocobo: here is the insruction for increasing timeouts: http://pastebin.com/DH7ZqQDp
14:45 Chocobo thanks evg!
14:48 jobewan joined #fuel
14:50 bdudko joined #fuel
14:50 aglarendil joined #fuel
14:53 lordd_ joined #fuel
14:56 lordd_ joined #fuel
15:03 blahRus joined #fuel
15:11 lordd_ joined #fuel
15:22 ArminderS joined #fuel
15:44 adanin joined #fuel
16:12 xarses joined #fuel
16:19 ArminderS- joined #fuel
16:39 MiroslavAnashkin Chocobo: BTW, Astute is inside the astute container in 5.x. To find the astute settings you need to get this container's shell with `dockerctl shell astute`
17:13 blahRus1 joined #fuel
17:24 Chocobo MiroslavAnashkin: Ok, thanks.
17:27 Chocobo So we re-started the install to see if we could catch the problem.  Right now two of the three controller nodes have "Finished Catalog" in the puppet logs and one of them seems to have been at "... ceilometer_config/..../ensure completed" for the last 20 minutes without anything else in the puppet logs.   There seems to be a lot of network activity on private/admin network though.
17:29 MiroslavAnashkin Ceilometer generates a lot of traffic. Have you configured a separate MongoDB node for ceilometer?
17:33 Chocobo We just got an error with ceilometer.   I will paste it shortly.
17:34 Chocobo http://i.imgur.com/lH7s2lA.png
17:34 Chocobo Is this the timeout I need to increase?
17:39 e0ne joined #fuel
17:39 Chocobo By "master node" you mean the fuel master correct?   I do not seem to have any directories in /opt (even when I do dockerctl shell astute)
17:40 MiroslavAnashkin No, your timeout error rather indicates that storage network is not connected properly.
17:42 Chocobo MiroslavAnashkin: We have the Ceph OSD's set up to be on the compute nodes.   Those haven't even begun to initialize yet.  Hrm.  It passed the network check.
17:42 MiroslavAnashkin Did your ran Verify Network before clicking Deploy Changes?
17:44 Chocobo MiroslavAnashkin: Yes, and I can ping the other three nodes on the storage network (192.168.2.0/24)
17:44 angdraug joined #fuel
17:49 rmoe joined #fuel
18:01 t_dmitry joined #fuel
18:01 ykotko joined #fuel
18:05 holser left #fuel
18:09 kupo24z joined #fuel
18:11 Chocobo We are trying again with Ceilometer.  :(
18:20 aleksandr_null joined #fuel
18:20 e0ne joined #fuel
18:23 fandi joined #fuel
18:26 e0ne joined #fuel
18:29 lordd_ joined #fuel
18:31 lordd_ joined #fuel
18:57 e0ne joined #fuel
19:03 Chocobo can you install ceilometer on multiple nodes?
19:06 kupo24z xarses: Getting 502 Bad Gateway on a stable/5.0 install, anything I should check in the logs?
19:07 xarses kupo24z: 502 from the fuel master?
19:07 kupo24z Yes
19:07 xarses check the ngnix and nailgun containers
19:09 kupo24z Chocobo: With fuel you can install MongoDB on multiple nodes which is the Ceilometer DB
19:11 Chocobo kupo24z: Ok, thanks.  We are having a problem with ceilometer/rsync timeout.  We are trying a new install without ceilometer.
19:11 kupo24z xarses: Looks like a nailgun problem connecting: http://pastebin.mozilla.org/6148716
19:12 xarses kupo24z: thats complaining about postgres
19:13 kupo24z Yeah looks like issues with the container http://pastebin.mozilla.org/6148731
19:14 kupo24z ISO is fubar?
19:16 xarses hmm yep, looks a little jacked, I wonder if the container created correctly
19:16 xarses you built the ISO?
19:17 kupo24z Yeha
19:17 kupo24z Yeah*
19:17 kupo24z Well, not like you guys host latest stable/5.0 images :)
19:17 xarses ya, I'm still poking people about it
19:18 xarses I don't think it will happen for stable/5.0  at this point though
19:18 kupo24z my ISO machine is running ubuntu 12.04.5 on 3.16-1
19:18 xarses hopefully stable/5.1
19:20 kupo24z Can I change MASTER_* in config.mk without hurting anything?
19:20 kupo24z (Network Configuration)
19:25 Chocobo damn!  I got another timeout.  this one was with Swift Ringsync (rsync).   Grrrr
19:26 Chocobo it was copying /etc/swift/object.ring.gz
19:36 Chocobo joined #fuel
19:36 Chocobo joined #fuel
19:37 nerdonmountain joined #fuel
19:37 angdraug Chocobo: https://review.openstack.org/116247
19:37 kupo24z xarses: I deleted my git dir and recloned now remaking ISO. will see if puppet fails on the iso creation
19:39 kupo24z This normal? ** Found 2 pre-existing rpmdb problem(s), 'yum check' output follows: udev-147-2.51.el6.x86_64 has missing requires of /sbin/service udev-147-2.51.el6.x86_64 has missing requires of MAKEDEV >= ('0', '3.11', None)
19:39 kupo24z Could it be because of my newer kernel?
19:58 xarses no, not unless we are building packages, and probably not then
19:58 xarses also, its yum packages
19:59 xarses seems like the package is missing
19:59 kupo24z Did see a few iptables puppet errors as well, i think those are intended though
20:00 xarses can you paste bin all of the output and create a bug? I'll poke some of the build guys ( I will note that the are in eastern europe so they wont respond until late tonight)
20:02 kupo24z Will do when its done
20:02 kupo24z this one looked kinda scary Notice: /Stage[main]/Nailgun::Cobbler/Exec[cobbler_system_add_default]/returns: invalid profile name: bootstrap; Error: cobbler system add --name=default     --profile=bootstrap --netboot-enabled=True returned 1 instead of one of [0]
20:23 angdraug joined #fuel
20:26 Chocobo angdraug: that looks like what we might be running into.
20:27 geekinutah joined #fuel
20:29 angdraug try to apply that patch to /etc/puppet/modules and redeploy
20:33 nullme joined #fuel
20:35 IlluminatiJudas joined #fuel
20:40 nullme hello, quick question, do you know when VPNaas and LBaaS will be introduced into Mirantis Fuel?
20:44 kupo24z nullme: lbaas packages should be avail but not installed on 5.0.1+
20:45 kupo24z They will be on the ISO on 5.1, the repos on 5.0.x
20:46 nullme what is the repo link?
20:46 kupo24z http://fuel-repository.mirantis.com/fwm/
20:46 kupo24z http://fuel-repository.mirantis.com/fwm/5.0.2/ubuntu/pool/main/neutron-lbaas-agent_2:2014.1.1-fuel5.0.2~mira3_all.deb
20:47 kupo24z 4.1.2 in progress it looks like still https://bugs.launchpad.net/fuel/+bug/1330610
20:49 kupo24z xarses: you guys have any plans for installing one of the ceph dashboards in later releases? https://github.com/krakendash/krakendash https://github.com/Crapworks/ceph-dash
20:50 xarses I'd like to, but afaict its not going to be a priority.
20:51 kupo24z Are you moving forward with making compute and ceph-osd nodes totally off publicly accessible IPs? I'd imagine that would cause quite a few issues for users with external monitoring systems
20:51 nullme kupo24z: thank you for the link how about the openstack-neutron-vpn plugin driver?
20:53 kupo24z If they arnt on those repo's they would have to be compiled and added in by the fuel team
20:58 kupo24z nullme: http://fuel-repository.mirantis.com/fwm/5.1/ubuntu/pool/main/neutron-vpn-agent_2014.1.1-fuel5.1~mira17_all.deb http://fuel-repository.mirantis.com/fwm/5.1/ubuntu/pool/main/neutron-plugin-vpn-agent_2014.1.1-fuel5.1~mira17_all.deb
20:58 kupo24z 5.1
20:58 kupo24z http://fuel-repository.mirantis.com/fwm/5.0.2/ubuntu/pool/main/neutron-plugin-vpn-agent_2:2014.1.1-fuel5.0.2~mira3_all.deb http://fuel-repository.mirantis.com/fwm/5.0.2/ubuntu/pool/main/neutron-vpn-agent_2:2014.1.1-fuel5.0.2~mira3_all.deb
20:58 kupo24z 5.0.2
20:59 nullme kupo24z: Thank You so much for you help, i made my bookmarks!!!
21:00 kupo24z np
21:13 eshumkher joined #fuel
21:14 xarses kupo24z: 5.1 should not have public ip's on nodes that don't need it (by the looks of it) I'll paste the bug numbers if i find them again
21:15 kupo24z Is there an option to enable that for users that dont mind using the extra IP's?
21:18 xarses I would guess it could be possible with advanced-networking bp implementation but you would probably have to do something to the network requirements of the role to give it back a public ip
21:18 xarses imho you don't want public ip's on nodes that dont need them, regardless of your monitoring solution
21:19 xarses it just opens up too much attack surface
21:22 kupo24z Yeah I can just setup a localized collector (Zenoss), i was more asking for curiosity
21:25 kupo24z Would assigning local ip's to non-fuel deployed machines disrupt anything with future deployments in the enviornment?
21:33 nullme joined #fuel
21:34 nullme is there a upgrade documentation from 5.0 to 5.0.2 ?
21:36 aepifanov joined #fuel
21:40 kupo24z xarses: bug created https://bugs.launchpad.net/fuel/+bug/1361383
21:40 xarses kupo24z: thanks
21:41 xarses nullme: 5.0.2 is not a released version
21:41 kupo24z http://docs.mirantis.com/openstack/fuel/fuel-5.0/user-guide.html#upgrade-from-an-earlier-version
21:46 kupo24z btw the default kernel for ubutu 12 will not work with the docker tasks you guys are doing, it panics quite a bit
21:46 kupo24z ubuntu*
21:59 blahRus joined #fuel
22:12 tuvenen_ joined #fuel
22:18 fhond joined #fuel
22:26 kupo24z angdraug: Have you tested with rbd cache = true with Fuel environments yet? I've tried enabling it and it takes forever to start new instances vs instant with it off
22:43 angdraug kupo24z: no, not yet. just had the bug about it raised an hour ago
22:43 angdraug https://bugs.launchpad.net/bugs/1361391
22:44 kupo24z Yeah I enabled that just for kicks and vm creation took ages, diddnt really look into it any more than that though
22:45 angdraug did you just enable rbd cache?
22:45 angdraug should also set "rbd cache writethrough until flush = true"
22:45 kupo24z the [client] changes on all nodes
22:48 xarses joined #fuel

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