Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel, 2015-06-03

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

All times shown according to UTC.

Time Nick Message
00:01 LiJiansheng joined #fuel
00:08 xarses joined #fuel
00:57 lee__ joined #fuel
01:02 LiJiansheng joined #fuel
01:06 lee__ joined #fuel
01:16 teran joined #fuel
02:17 teran joined #fuel
03:20 teran joined #fuel
03:54 Longgeek joined #fuel
03:57 Longgeek_ joined #fuel
03:58 fedexo joined #fuel
04:02 Longgeek joined #fuel
04:08 Longgeek_ joined #fuel
04:18 Longgeek joined #fuel
04:19 Longgeek joined #fuel
04:20 teran joined #fuel
04:20 Longgeek joined #fuel
04:23 Longgeek_ joined #fuel
04:25 Longgee__ joined #fuel
04:26 Longgeek joined #fuel
04:28 Longgeek_ joined #fuel
04:29 Longge___ joined #fuel
05:21 teran joined #fuel
05:27 youellet__ joined #fuel
06:22 teran joined #fuel
06:24 saibarspeis joined #fuel
06:35 Longgeek joined #fuel
06:40 teran joined #fuel
07:40 samuelBartel joined #fuel
08:17 aarefiev joined #fuel
08:38 tzn joined #fuel
08:46 e0ne joined #fuel
08:47 tzn joined #fuel
09:14 HeOS joined #fuel
09:26 monester_laptop joined #fuel
09:52 lquince joined #fuel
09:53 lquince_ joined #fuel
09:54 kutija joined #fuel
10:00 DevStok joined #fuel
10:00 DevStok hi all
10:01 DevStok I'm killing myself trying to understand why I get BAD GATEWAY response
10:01 DevStok 502 or 504
10:01 DevStok from HAPROXY
10:01 DevStok I got 3 controllers in HA
10:01 DevStok ICEHOUSE
10:02 DevStok from last friday I get that problems
10:02 DevStok nova list , sometimes is executed and sometimes goes in timeout
10:04 lee__ joined #fuel
10:05 DevStok @kaliya ci sei?
10:11 kutija If I have 3 controllers in HA, and two of them are down, what should be the expected behavior?
10:11 kutija I ask because I deliberately killed 2 controller nodes
10:11 kutija and my OpenStack has gone crazy
10:11 kutija the last controller did not took control
10:11 kutija and network was down, just as dashboard and all of nova services
10:12 DevStok yes with icehoue i noticed the same behavior
10:13 DevStok you need 2 ctrl active
10:13 DevStok for the net
10:13 DevStok u have to be lucky that the qrouter is deployed on the controller alive
10:13 kutija joined #fuel
10:14 DevStok have u read?
10:14 kutija nope my connection broke
10:14 kutija please repeat :)
10:15 DevStok yes with icehoue i noticed the same behavior
10:16 DevStok you need 2 ctrl active
10:16 DevStok for the net, u have to be lucky that the qrouter is deployed on the controller alive
10:16 kutija lucky? that does not sound good
10:17 kutija I will try again
10:17 kutija to see what is going to happen
10:18 DevStok lucky yes
10:18 DevStok fake HA
10:20 DevStok if you loose the ctrl where the l3 agents works u'll loose the connectivity
10:20 DevStok @xarses
10:20 kutija but shouldn't l3 agent be available on all ctrl nodes?
10:21 kutija I mean why then I need 3 nodes
10:21 DevStok ahaha yes because is a fake HA
10:21 DevStok only balancing
10:21 kutija yeah I get it
10:21 DevStok try to stop corosync on the ctrl with l3 agents
10:24 kutija I'll try
10:27 kaliya DevStok: uh?
10:30 DevStok hey
10:30 DevStok can u help me?
10:30 kaliya write, if I cannot someone else maybe can
10:32 DevStok I wrote
10:32 DevStok can u read?
10:32 DevStok btw
10:32 DevStok 3 controllers Icehouse HA
10:33 DevStok nova list -> sometimes get answer sometimes bad gateway
10:33 DevStok 502 or 504
10:33 kutija joined #fuel
10:33 kaliya vip up in crm resources? other openstack commands receive 50x also?
10:33 DevStok is possible that a qrouter with similar config gives problems to haproxy config
10:34 DevStok ip netns exec haproxy ip a -------> 192.168.0.4
10:34 DevStok ip netns exec qrouter-xxxxxxx ip a ----> 192.168.0.1
10:42 DevStok via crm
10:42 DevStok what command
10:42 DevStok via crm i can type command and get fast response
10:46 kaliya glance image-list, keystone user-list or so
10:47 kutija build #259 @ 6.1 works very very well
10:47 kutija it's a huge improvement over builds from last two weeks
10:47 kutija deployed without single issue
10:48 kutija great job guys
10:48 DevStok kaliya glance image-list have the same issue
10:48 DevStok like nova list
10:48 DevStok neutron net-list
10:48 DevStok etc
10:49 DevStok in nova-api.log
10:49 DevStok 2015-06-03 12:48:25.494 20746 ERROR glanceclient.common.http [req-1036ed83-210f-46f5-a266-9761f45c46f9 ] Request returned failure status. 2015-06-03 12:48:25.495 20746 ERROR nova.api.openstack [req-1036ed83-210f-46f5-a266-9761f45c46f9 None] Caught error: <html><body><h1>504 Gateway Time-out</h1> 2015-06-03 12:48:26.144 20747 ERROR nova.api.openstack [req-edd78021-28c0-4a2e-9b98-49a9e7c4b5aa None] Caught error: <html><body><h1>502 Bad G
10:53 kaliya thanks kutija
10:54 kaliya DevStok: `crm resource` status?
10:54 kutija I have a question about expanding storage - if I add two more storage nodes, will it just increase replication level or I will get 24TB more of storage?
10:55 kutija (I have 2 nodes now)
10:58 DevStok u can check the replication factor in ceph
10:58 DevStok probably u got 2
10:58 DevStok as factor
10:58 DevStok crm status is ok
10:58 kaliya crm resource
10:59 kaliya and yes kutija if you haven't increased the replication ratio, you'll get your space
10:59 kutija kaliya: is there a way to "control" where instances will land, I mean on which storage in that case?
11:01 kaliya on which compute makes sense, and you can
11:01 samuelBartel joined #fuel
11:01 kaliya storage, not sure, since there are also replicas around
11:02 kutija I know for compute
11:03 DevStok @kaliya crm resource .... then?
11:03 LiJiansheng joined #fuel
11:03 kutija thanks kaliya
11:03 DevStok status is ok
11:03 DevStok started for all
11:06 kaliya nova service-list
11:07 DevStok is all ok
11:07 DevStok same for neutron agent-list
11:07 DevStok friday we did a mod to neutron and neutronclient python
11:07 kaliya is the python-nova client updated in regards of the fuel-repository.mirantis.com?
11:07 DevStok but I restored the old files
11:07 kaliya which mod
11:10 DevStok yes
11:11 DevStok a mod to use another 4 table regarding the management of qos in neutron
11:11 DevStok but i restored the files
11:32 gongysh joined #fuel
11:58 e0ne joined #fuel
12:05 Akshik__ joined #fuel
12:05 Akshik__ Hi team
12:05 Akshik__ would like to know if the mentioned patch is available part of Fuel 5.1
12:05 Akshik__ https://git.openstack.org/cgit/openstack/nova/commit/?id=66721eb2c0f53fc4260b2f0aa9a3811da0f7ddbd
12:06 Akshik__ im trying to boot instance and create new volume
12:06 Akshik__ in the process since the volumen is taking long time is getting failed
12:06 Akshik__ when i tried configuring the block_device_allocate_retries it is not honored
12:06 Akshik__ need your inputs
12:14 DevStok joined #fuel
12:33 lquince joined #fuel
12:36 lquince_ joined #fuel
12:46 hyperbaba joined #fuel
12:49 lquince joined #fuel
12:57 rmoe joined #fuel
13:05 mattgriffin joined #fuel
13:42 evg joined #fuel
14:19 flexi joined #fuel
14:42 DevStok joined #fuel
14:42 DevStok haproxy
14:43 DevStok doesnt work
15:01 newbism joined #fuel
15:03 newbism Hi. I have a Force10 10Gbe switch and a quad node with Mellanox ConnectX3 cards Id like to use for storage network... I have my nodes discovered, configured interfaces... It shows it Green at 10Gb... I click verify networks but it fails saying it was expecting "untagged"
15:03 newbism What do I need to do to the switch to make it untagged? I didn't add any vlans so im confused...
15:05 mwhahaha newbism:  https://www.force10networks.com/CSPortal20/TechTips/0051B_HowDoIConfigureVLANs.aspx
15:06 newbism mwhahaha: so the storage network needs to be in a vlan but untagged?
15:07 mwhahaha no it can be tagged
15:07 mwhahaha did you configure it as tagged?
15:07 mwhahaha in fuel
15:08 newbism mwhahaha: I deslected the vlan, figure since the whole switch is for storage why bother with it
15:08 mwhahaha ok so you need to make sure the interfaces on the switch are untagged
15:09 mwhahaha the url i pasted is an example of how to do that on the force10 FTOS
15:10 mwhahaha so you'd need to put all the ports you're using in a single vlan untagged
15:10 mwhahaha and then it should work
15:10 mwhahaha for force 10 it's backwards from cisco where you configure the vlan with interfaces and not interfaces with vlans
15:11 mwhahaha i'm not sure but if you haven't already configured the switch it may ship with all the interfaces in a unconfigured state
15:18 lquince joined #fuel
15:20 lquince_ joined #fuel
15:23 lquince__ joined #fuel
15:47 newbism mwhahaha: set the ports up as described untagged... click verify, still complains expected untagged
15:49 newbism mwhahaha: I checked node fuel list for the discovered nodes... ssh'd into them applied an IP on eth0's ... they can ping each other
15:50 newbism so close :(
15:50 newbism ugh
15:50 mwhahaha well you could still try deploying
15:50 mwhahaha and if it works file a bug against the network checker
15:53 newbism arg ok
15:55 lquince_ joined #fuel
16:21 darkhuy joined #fuel
16:27 xarses joined #fuel
16:32 teran joined #fuel
16:53 rmoe joined #fuel
16:57 kutija joined #fuel
17:31 ub joined #fuel
17:36 ub joined #fuel
17:41 hyperbaba joined #fuel
17:55 deckardkain joined #fuel
17:58 newbism joined #fuel
17:58 newbism So when a release is made... generally speaking it should work that is... the release is considard 'stable' and the features within 'work'?
17:59 flexi hi what is the best way train for openstack which training is these days going the best
17:59 flexi ?
17:59 tatyana joined #fuel
18:00 flexi http://www.openstack.org/marketplace/training/
18:00 flexi Cloud enabled or mirantis?
18:00 newbism What's your budget?
18:01 flexi no budget
18:01 flexi i am looking for any online videos for the mirantis bootcamp
18:02 flexi ideally Mirantis OS100, OS50 or if possible OS200
18:03 kaliya flexi: maybe you also want that we present you some certificate
18:03 newbism yeah idk of any comprehensive training for free
18:04 flexi oh no
18:04 flexi no free
18:04 newbism its really annoying I cant change settings after resetting the environment
18:04 flexi *not
18:04 kaliya I nominate you, flexi, sir Mirantis Cloud certified
18:04 kaliya you happy now?
18:04 deckardkain this might not be best place to pose this question. But i was wondering where i could find a matrix of mirantis pricing for support.
18:04 newbism Could not find resource 'Service[undef]' for relationship from 'Package[ceilometer-common]' on node node-1.domain.tld
18:04 flexi thanks but still long way to go
18:04 newbism oh also tried to deploy with iser ... failed
18:05 flexi has anyone deployed fuel openstack with vmware?
18:05 flexi vmware workstation
18:05 flexi ?
18:05 newbism no, im not a masochist
18:06 kaliya flexi: nova on vmware, or the masternode and the infrastructure in vcenter?
18:06 newbism its weird multi-node HA and QEMU are defaults
18:06 kaliya newbism: 6.0?
18:06 newbism yes
18:07 newbism kaliya: yes
18:07 kaliya so you have controllers+mongo?
18:07 flexi kaliya: no i have used virtual box on win8+cygwin and then tried in guest (ubuntu) ... but i dont find it useful
18:07 newbism kaliya: yes
18:07 kaliya newbism: and this happens when you reset the env?
18:07 flexi i want to try vmware or kvm which may allow h/w virtualization on guest machine
18:07 kaliya was the env operational and working?
18:07 flexi my host is windows
18:07 newbism kaliya: no no ... during provision
18:07 kaliya mmmh
18:07 newbism forget i said anything about reset env
18:08 kaliya newbism: without mongo, of course it provisions fine right?
18:08 newbism kaliya: we shall see...
18:08 kaliya I would go and run a puppet apply and check the logs
18:09 kaliya newbism: I am leaving the office now, but you can reach us here. In case, if you find on that provisioning error again, would you plese hop onto the failing controller and grab the puppet.log ?
18:09 newbism kaliya: sure
18:09 kaliya flexi: do you want to run an openstack cluster, or do you simply need some desktop virtualisation?
18:10 deckardkain one other question what's driving the deployment of openstack services? i.e rackspace private cloud uses ansible
18:10 kaliya deckardkain: fuel uses puppet now, used chef some time ago
18:10 kaliya newbism: is this? https://bugs.launchpad.net/fuel/+bug/1455029
18:11 newbism kaliya: yep thats it
18:11 deckardkain kaliya: thanks.
18:11 kaliya newbism: so this is in progress. If you have a LP account, please click that this bug affects you. And if you can take the puppet.log, please attach in response to this bug. Thanks!!
18:12 newbism kaliya: ok
18:15 deckardkain Need a little advice with choosing a network topology. I would like to deploy instances on KVM as well as vpshere. as per the fuel-6.0 docs "vmware vcenter and vsphere can only be implemented on nova-network with flatdhcp topology.". My question is will this support 8021q as we will need multi tenancy in our esxi environment. Will this impact our requirement for openstack to build instances on kvm as well?
19:03 kutija could someone recommend a good way to do a stress test, except using Tempest?
19:06 deckardkain kutija: what kind of stress test?
19:06 kutija network for starters
19:07 kutija and overall performance, mostly IO
19:07 kutija to see how Ceph is holding and Neutron with lot of network activity
19:08 kutija I already wrote some scripts to see how it's working with a lot of instances spawning at the same time but I would like to hit it harder
19:08 kutija I also spawned instances running benchmark tests all at the same time (IO, CPU, network) mostly with IOPing, UnixBench and so on
19:10 kutija and it turned out well, I realized what were the weak spots in terms of IO but I would like to see how much my internal network can hold on
19:10 kutija and how really reliable Neutron is
19:10 kutija so I'm open to suggestions
19:10 kutija if any of course
19:13 e0ne joined #fuel
19:49 teran joined #fuel
20:02 lquince joined #fuel
20:07 mattgriffin joined #fuel
20:07 teran_ joined #fuel
20:18 HeOS joined #fuel
20:53 tatyana joined #fuel
20:56 tatyana joined #fuel
20:57 tatyana left #fuel
21:01 angdraug joined #fuel
21:07 xdeller joined #fuel
21:27 youellet_ joined #fuel
21:53 mattgriffin joined #fuel

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