Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel, 2015-03-24

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

All times shown according to UTC.

Time Nick Message
00:33 rmoe joined #fuel
00:45 HeOS joined #fuel
00:50 LiJiansheng joined #fuel
00:51 Longgeek joined #fuel
00:59 xarses joined #fuel
01:11 emagana joined #fuel
01:18 emagana joined #fuel
01:52 Rodrigo_US joined #fuel
01:57 mattgriffin joined #fuel
02:28 claflico joined #fuel
02:46 mattgriffin joined #fuel
02:59 claflico joined #fuel
03:16 ddmitriev joined #fuel
03:22 Longgeek_ joined #fuel
03:28 mattgriffin joined #fuel
03:37 gongysh joined #fuel
05:19 zerda joined #fuel
05:45 emagana joined #fuel
07:22 emagana joined #fuel
07:42 pbeskow joined #fuel
07:49 alecv joined #fuel
09:28 HeOS joined #fuel
09:30 e0ne joined #fuel
10:16 alexz joined #fuel
10:28 e0ne joined #fuel
11:03 vtzan MiroslavAnashkin, I understand. the problem is the same even without mongo,ceilometer,osd mon on controllers enabled.
11:17 hyperbaba joined #fuel
11:20 e0ne joined #fuel
11:35 adanin joined #fuel
11:35 bazilza joined #fuel
11:42 bazilza hi, i'm having problems after installing openstack with fuel 6.0
11:42 bazilza vms don't get meta-data
11:42 bazilza 2015-03-24 11:22:44,710 - util.py[WARNING]: Failed fetching metadata from url http://169.254.169.254/2009-04-04/meta-data/
11:43 bazilza all i see is neutron reconnecting to amqp
11:43 bazilza but ERROR oslo.messaging._drivers.impl_rabbit [-] Failed to publish message to topic 'reply_0941f6d5244343f78eea288a46815ad2': [Errno 32] Broken pipe
11:46 bazilza if i restart neutron-server it works for first 5-10 minutes. then meta-data problems start again
11:47 bazilza http://paste.openstack.org/show/195910/
11:48 bazilza upgraded python-oslo.messaging  to 1.4.1-fuel6.0~mira18 , but it doesn't help
11:49 bazilza meta-data server is functional
11:49 bazilza any ideas how to debug this?
11:53 bazilza on rabbitmq side i see closing AMQP connection <0.32289.11> (10.191.4.12:39060 -> 10.191.4.3:5673): {heartbeat_timeout,running}
11:53 e0ne joined #fuel
12:07 dkusidlo joined #fuel
12:07 jaypipes joined #fuel
12:33 aliemieshko <bazilza>hi, try to change vm.swappiness : https://bugs.launchpad.net/fuel/+bug/1413702/ and check /etc/haproxy/haproxy.cfg on the controllers  add ' option http-server-close ' to the 'defaults' section of /etc/haproxy/haproxy.cfg
12:33 aliemieshko Since this setting affects the local HAProxy daemon only and since your cluster may be in production - please do not restart the whole HAProxy with `pcs resource ...` command.
12:33 aliemieshko Instead, you may apply this setting with the one by one controller (one controller at time) restart when convenient, or, if not critical, in scope of the future maintenance window.
12:34 aliemieshko and then try to to start new instance
12:39 f1gjam joined #fuel
12:44 Miouge joined #fuel
13:03 pal_bth joined #fuel
13:03 omolchanov joined #fuel
13:05 Miouge joined #fuel
13:25 bazilza aliemieshko, thanks. seems like neutron doesn't connect through haproxy, it uses 5673 not 5672(haproxy port)
13:31 bazilza aliemieshko, swappiness is not an issue. there is abundance of memory
13:42 BobBall How can I debug notthe fuel screen not loading?  Trying to access over a socks proxy (with ssh -D 8080 root@fuel-master) yet it gets to the welcome screen and gets stuck on the cloud progress spinner
13:47 hyperbaba joined #fuel
13:48 bazilza BobBall, check fuel server still has free space left on disk
13:49 BobBall 8.1gb free
13:50 BobBall 29gb free on /var
14:00 bazilza hm
14:00 bazilza BobBall, check the nginx container
14:00 bazilza you can see the log with "docker logs <container name>"
14:00 BobBall hmmm - curiously I only seem to have 4 containers running - mcollective, nginx, astute and keystone
14:01 BobBall I was certain with my previous installation there were 6-8 containers
14:01 kaliya BobBall: `dockerctl check`
14:01 BobBall logs from nginx container seem to be OK to me
14:03 BobBall lxc-attach failed to get the init pid?? Certainly the nailgun container does not seem to be there... Reboot again? (I did only just reboot recently) Can't do "docker logs nailgun" since it needs a container ID and the nailgun container doesn't seem to have started... *digs*
14:03 BobBall Ah - it's because I changed the IP address with fuelmenu
14:03 BobBall somewhere it's still got 10.20.0.2 hardcoded
14:04 BobBall Looks like it has it hardcoded in dozens of places :) I guess therefore that changing the IP addresses post-install isn't really supported ;)
14:05 julien_ZTE joined #fuel
14:06 BobBall Is there a way to update things?  or do I have to use the 10.20.0.0/24 range?  (a different range was allocated to me by my corporate IT)
14:12 ddmitriev joined #fuel
14:14 bazilza BobBall, i'd like to know that as well. i had to reinstall fuel when i needed to change the IP. otherwise it didn't work for me.
14:15 bazilza since then i don't even touch fuelmenu. i just set showmenu=yes on installation boot and configure everything there
14:16 BobBall Oh - so if showmenu=yes then we can configure on first boot then not touch? That's useful.  Easier than setting the static IPs in the boot command line
14:17 bazilza yeap
14:17 BobBall coolio
14:22 bazilza any way to pull updates for 6.0 to already installed fuel? i tried to use upgrade package but it doesn't want to. says i need to have 5.1.1
14:29 BobBall bazilza: See http://docs.mirantis.com/openstack/fuel/fuel-6.0/user-guide.html#changing-pxe-network-parameters-during-installation "Warning: This section must be configured only in scope of Fuel Master node first boot! Setting new network settings for the already installed master node requires that all Docker containers be rebuilt and possibly further manual reconfiguration!"
14:29 BobBall Looks like changing network settings isn't supported :)
14:30 claflico joined #fuel
14:38 e0ne joined #fuel
14:42 f1gjam2 joined #fuel
14:54 getup joined #fuel
14:56 jobewan joined #fuel
15:10 mattgriffin joined #fuel
15:22 LiJiansheng joined #fuel
15:28 emagana joined #fuel
15:42 f1gjam2 I guess, my first question was my mis-understanding of how things should work. So my question is, once everything is setup, should a ops person be able to SSH to all nodes directly via ssh on the admin network, or is the only means to access the shell on nodes via hopping from the fuel master node. This includes if you have intern VLAN  routing setup so you can access the admin network directly.
15:44 blahRus joined #fuel
15:48 kaliya f1gjam: you can SSH-access the nodes not from the Master. There is an option in the Fuel UI ('Settings') to input a ssh-key
15:50 f1gjam the problem isnt that
15:50 f1gjam the problem is i can ping the nodes but cant ssh
15:50 f1gjam when i looked, the gateway on the controller and the compute nodes is set to the master
15:50 f1gjam and the master gateway is set to the router
15:51 f1gjam so trying to ssh fails as the nodes dont know how to get back to the client
15:53 Miouge joined #fuel
16:01 f1gjam here is the controllers routing table
16:01 f1gjam Kernel IP routing table
16:01 f1gjam Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
16:01 f1gjam 0.0.0.0         192.168.20.2    0.0.0.0         UG    100    0        0 br-fw-admin
16:01 f1gjam 192.168.20.0    0.0.0.0         255.255.255.0   U     0      0        0 br-fw-admin
16:01 f1gjam 192.168.40.0    0.0.0.0         255.255.255.0   U     0      0        0 br-mgmt
16:01 f1gjam 192.168.50.0    0.0.0.0         255.255.255.0   U     0      0        0 br-storage
16:01 f1gjam 192.168.122.0   0.0.0.0         255.255.255.0   U     0      0        0 virbr0
16:01 f1gjam sorry thats a compute node
16:02 emagana joined #fuel
16:03 ppradhan joined #fuel
16:03 julien_ZTE joined #fuel
16:04 rmoe joined #fuel
16:06 f1gjam I just wanted confirmation of this is expected behaviour
16:20 Miouge joined #fuel
16:21 f1gjam kaliya
16:21 f1gjam kaliya - can you confirm this ?
16:32 Miouge joined #fuel
16:38 GustavoJoseRuiz1 joined #fuel
16:41 jaypipes_ joined #fuel
16:49 GustavoJoseRuiz1 holaaa
16:50 angelberrios joined #fuel
16:50 kaliya f1gjam: for me is working, SSH from a host on the admin segment which is not the Master
16:51 GustavoJoseRuiz1 left #fuel
16:53 angelberrios I do not know the channel policy to post a question, could somebody help me.
16:55 kaliya angelberrios: IRC rules (ask, if someone can help will answer) apply here as well
16:56 angelberrios Thanks
16:56 angelberrios I have an operational environment running in production:
16:57 angelberrios 8  | operational | HLPRCLOUD01 | multinode | 2          | [{u'node_id': 18, u'name': u'disks'}, {u'node_id': 19, u'name': u'disks'}, {u'node_id': 20, u'name': u'disks'}] | None
16:58 angelberrios after the upgrade to 5.1.1 from 5.1 the deployment changes button was enable
16:59 angelberrios There is any way to reset or undo changes so I could add more nodes?
17:01 kaliya angelberrios: isn't the Add Nodes button available?
17:02 angelberrios yes, but if I add the nodes, won't it tried to apply the "changes"?
17:03 kaliya it will after you click Deploy Changes
17:04 angelberrios The information about the disks are not equal to what is currently running and I will loose the actual setup
17:04 angelberrios right?
17:05 julien_ZTE joined #fuel
17:05 kaliya Disks partitioning doesn't need to be equal on all the nodes
17:07 angelberrios so you could apply changes selectively?
17:07 kaliya You can add the new nodes with the desired partitioning yes
17:08 angelberrios So you could select to apply changes only to the new node?
17:08 kaliya Yes, and even it's not HA
17:11 angelberrios performing .....
17:17 angelberrios ok, I see now my problem
17:18 xarses joined #fuel
17:18 angelberrios the deploy changes button is visible, but the discard changes icon is not
17:19 f1gjam kaliya where can i post details if not here?
17:19 angelberrios how can perform the same function of the discard icon from shell?
17:19 f1gjam it seems as though the admin network is not being configured correctly
17:19 kaliya f1gjam: see the topic
17:19 kaliya angelberrios: discard?
17:20 angelberrios yes
17:20 f1gjam acutally i get 30 days free support might as well use it :)
17:20 kaliya f1gjam: please ;) I'm awaiting for your cases
17:20 kaliya angelberrios: please explain, don't get your point
17:20 f1gjam ok i will do it once I finish with this VPN issue :)
17:21 angelberrios right now I have the changes button available but the information is wrong, I want to discard those changes
17:21 angelberrios but do not have the icon vailable at the gui
17:22 * kaliya is thinking about the meaning of discard changes
17:23 angelberrios where I could post screen shot?
17:23 kaliya angelberrios: on any image sharing site
17:26 angelberrios http://tinypic.com/r/u8t1e/8
17:27 angelberrios http://tinypic.com/r/24ouk2e/8
17:27 angelberrios The information of those "changes" is wrong.
17:28 angelberrios but i do not have the discard changes available in the first screen shot
17:29 kaliya angelberrios: you just added 2 nodes, then when you Deploy Changes, it says he wants to apply a new disk configuration to the existing nodes?
17:30 angelberrios no, this is without adding any new nodes
17:30 kaliya you didn't modify the existing nodes partitioning right?
17:30 angelberrios yes
17:31 kaliya yes- you did or yes- you didn't?
17:31 angelberrios i do not modify, it was shown after the upgrade
17:31 angelberrios i did not
17:31 kaliya Ceph?
17:31 angelberrios nope
17:31 angelberrios cinder, but the environment allows for cepth
17:32 kaliya show me a screenshot of the disk partitioning for the 3 cluster nodes
17:32 angelberrios ok
17:32 angelberrios you mean actual or in the gui?
17:32 kaliya in the UI
17:37 angelberrios http://tinypic.com/r/2lt3q50/8  http://tinypic.com/r/2h2fg44/8  http://tinypic.com/r/2yw6592/8
17:37 kaliya angelberrios: has something changed from the initial setup?
17:38 angelberrios nope
17:39 angelberrios i did try a lot of configuration
17:39 kaliya http://tinypic.com/r/24ouk2e/8 appears when you click Deploy Changes?
17:39 championofcyrodi what is the service endpoint for object store backed by ceph (e.g. rados gw) ? although we 'checked' the box to use ceph for the swift object store, we don't see anything running on the controllers related to ceph-radosgw
17:39 championofcyrodi and our swift api endpoint is 503 unavailable
17:39 angelberrios yes
17:40 kaliya angelberrios: why are you Deploying Changes?
17:41 angelberrios I was trying to do some modification but I discard them in v 5.1
17:41 championofcyrodi nvm... appears to be ceph related
17:41 championofcyrodi http://ceph.com/docs/master/radosgw/
17:42 angelberrios when I did the uopgrade the button appears
17:42 angelberrios but no discard icon
17:44 kaliya angelberrios: no need to Deploy Changes
17:44 angelberrios yes
17:45 angelberrios but not discard icon :-(
17:47 f1gjam joined #fuel
17:50 angelberrios sorry, but was I ambiguous in my last answer kaliya?
17:50 angelberrios I do not need to deploy those changes
17:51 angelberrios but no have any way in the UI to discard them
17:54 championofcyrodi looks like i'm stilling missing some concept here.... should i expect to be able to use the swift commandline api w/ ceph as the backend storage?
17:54 championofcyrodi using radosgw-admin i successfully created an s3 user... but the actual 'swift list' command on a controller also has 503 unavail
17:55 championofcyrodi is this expected? and i should proceed to auth w/ ceph directly? or should the openstack swift api work?
18:03 emagana joined #fuel
18:29 emagana joined #fuel
18:54 julien_ZTE joined #fuel
19:11 e0ne joined #fuel
19:19 HeOS joined #fuel
19:30 emagana joined #fuel
19:37 emagana joined #fuel
19:53 ppradhan I have ha setup ( 3 control, 2 computer and 3 ceph) . I had to shutdown one control node. Everything runs but accesing is dashboard is really slow. Where do I start to debug?
20:15 f1gjam2 joined #fuel
20:42 julien_ZTE joined #fuel
21:05 emagana joined #fuel
21:26 mattgriffin joined #fuel
21:29 emagana joined #fuel
21:43 julien_ZTE joined #fuel
21:49 mrasskazov joined #fuel
21:49 tlbr joined #fuel
21:53 kiwnix joined #fuel
21:53 emagana joined #fuel
21:59 mutex joined #fuel
22:28 julien_ZTE joined #fuel
22:35 jfluhmann joined #fuel
22:50 LiJiansheng joined #fuel
23:16 xarses joined #fuel
23:52 julien_ZTE joined #fuel

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