Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel, 2013-12-13

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

All times shown according to UTC.

Time Nick Message
23:10 book` joined #fuel
23:13 dnovakovskiy joined #fuel
23:44 IlyaE joined #fuel
00:21 IlyaE joined #fuel
00:31 IlyaE joined #fuel
01:33 fandikurnia01 joined #fuel
01:53 Fecn joined #fuel
02:13 Fecn1 joined #fuel
02:49 Fecn joined #fuel
02:50 xarses joined #fuel
03:21 rmoe joined #fuel
03:29 Fecn joined #fuel
03:32 ArminderS- joined #fuel
03:33 ArminderS joined #fuel
03:34 ArminderS- joined #fuel
05:27 SergeyLukjanov joined #fuel
05:30 vkozhukalov joined #fuel
06:15 mihgen joined #fuel
06:44 Vidalinux joined #fuel
06:45 wputra joined #fuel
06:50 Fecn joined #fuel
07:08 teran joined #fuel
07:19 Fecn joined #fuel
07:19 SergeyLukjanov joined #fuel
07:28 mrasskazov joined #fuel
07:53 vkozhukalov joined #fuel
07:53 Fecn1 joined #fuel
07:55 e0ne joined #fuel
08:02 jouston joined #fuel
08:03 e0ne joined #fuel
08:12 SergeyLukjanov joined #fuel
08:14 vkozhukalov joined #fuel
08:16 e0ne joined #fuel
08:21 Fecn joined #fuel
08:29 rongze joined #fuel
08:41 Fecn joined #fuel
08:53 mrasskazov joined #fuel
08:54 Fecn1 joined #fuel
09:11 SteAle1 joined #fuel
09:13 SteAle joined #fuel
09:24 teran joined #fuel
09:24 rvyalov joined #fuel
09:26 teran_ joined #fuel
09:27 mihgen joined #fuel
09:34 rongze joined #fuel
09:35 tonyha joined #fuel
09:37 mrasskazov1 joined #fuel
09:37 Fecn joined #fuel
09:38 teran joined #fuel
10:00 Fecn joined #fuel
10:09 anotchenko joined #fuel
10:10 SergeyLukjanov joined #fuel
10:12 vk joined #fuel
10:20 anotchenko joined #fuel
10:34 dan_a_ joined #fuel
10:38 dan_a_ Hi all, is there any way to install Fuel other than with the Mirantis ISO? I'm hitting https://bugs.launchpad.net/fuel/+bug/1259276 - which is fixed in master - when trying to do an install from the ISO
10:45 anotchenko joined #fuel
10:46 e0ne Hi dan_a_, you can make your own iso following this guide: http://docs.mirantis.com/fuel-dev/develop/env.html#building-the-fuel-iso
10:47 dan_a_ Thanks e0ne, that's perfect
10:48 mattymo dan_a, we originally had a method of installing from CentOS and EPEL mirrors, but there are so many factors that could break deployment. Fixing the repos to a specific "snapshot" of upstream improves reliability for each dev cycle.
10:56 Fecn1 joined #fuel
11:08 Vidalinux joined #fuel
11:38 rongze joined #fuel
11:43 Axam joined #fuel
12:03 anotchenko joined #fuel
12:12 Fecn1 Is there any way to prevent the puppet scripts from waiting for 2 mins for a link on each disconnected NIC? Our servers have 10 nics each (though only 2 wires), and deploying a controller in 3.2.1 is taking over 34 mins because of all the waiitng around
12:13 ruhe joined #fuel
12:15 rongze joined #fuel
12:22 anotchenko joined #fuel
12:25 mihgen joined #fuel
12:33 SergeyLukjanov joined #fuel
12:44 ArminderS joined #fuel
13:02 MiroslavAnashkin There should be default timeout value, which you may set in resource declaration
14:00 mihgen joined #fuel
14:08 Fecn1 joined #fuel
14:13 teran joined #fuel
14:13 teran joined #fuel
14:18 e0ne_ joined #fuel
14:24 archon1st joined #fuel
14:36 ruhe joined #fuel
14:44 rongze joined #fuel
14:48 mrasskazov joined #fuel
14:52 vk joined #fuel
14:58 IlyaE joined #fuel
15:39 Shmeeny joined #fuel
15:39 rongze joined #fuel
15:56 jkirnosova joined #fuel
16:06 Fecn1 MiroslavAnashkin : Are you able to explain settings it in a resource declaration in a way that mere mortals can understand.. (preferably with a URL to guide me through) - I do have a couple of years of puppet experience, but only a couple of hours of looking beyond the Fuel web interface
16:07 Fecn1 I did already try chaiging the default_ping_timeout in the 123network manifest (the default was 30 secs in there)... but clearly that was not hte right place to change it as it kept on waiting for 60 secs for each nic after I'd made the change anyway
16:09 korn__ Currently Miroslav is offline.
16:09 Fecn1 Thanks - I shall lurk and hope I notice when he's back :)
16:10 korn__ He will be available on monday.
16:12 albionandrew joined #fuel
16:20 teran joined #fuel
16:21 rongze joined #fuel
16:21 teran_ joined #fuel
16:38 jeremydei joined #fuel
16:38 IlyaE joined #fuel
16:38 Vidalinux joined #fuel
16:44 IlyaE_ joined #fuel
16:44 jhurlbert joined #fuel
16:47 jhurlbert joined #fuel
16:51 Vidalinux joined #fuel
16:53 jeremydei joined #fuel
17:01 angdraug joined #fuel
17:04 e0ne joined #fuel
17:23 SergeyLukjanov joined #fuel
17:24 teran joined #fuel
17:25 jeremydei joined #fuel
17:30 xarses joined #fuel
17:32 rmoe joined #fuel
17:46 ruhe joined #fuel
17:48 SergeyLukjanov_ joined #fuel
18:06 xarses Fecn: == Fecn1?
18:08 xarses Fecn: there is a timer in l23networks that waits on an increasing backoff period for the link state to come up because some interfaces are not ready immediately. I think it first appears in 3.2.1
18:11 xarses Fecn: might be from this https://github.com/stackforge/fuel-library/commit/d62083a358a39b64690297956bacb825d0a77199
18:14 e0ne joined #fuel
18:15 IlyaE joined #fuel
18:31 e0ne joined #fuel
18:38 rmoe_ joined #fuel
19:20 IlyaE joined #fuel
19:48 Fecn joined #fuel
19:56 Fecn joined #fuel
20:02 Fecn xarses: Yes - that's the one... the 'waiting for carrier' message was a new thing for me in 3.2.1
20:03 xarses tsduncan__: whould it be possible for you to share your lscpi info for your network interfaces, we'd like to know the chipset info
20:03 xarses Fecn: ya, it was added for slow carrier up otherwise it's likely for the install to fail
20:06 Fecn xarses: Yeah - I understand why it's there - would just like to be able to tune it a bit... 60 secs is a very long time to wait for a link-detected signal
20:11 Fecn I found a default of 120 in 123network/lib/puppet/type/l3_if_downup.rb .. but I think mine was waiting for 60 secs....  Not an important point anymore as that deployment has completed now and I can get on with playing aroudn with openstack now
20:11 xarses Fecn: It looks like it waits up to wait_carrier_after_ifup_timeout, but checks every 10 sec
20:12 Fecn time to see if vlan splinters helps my strange network dropouts (which I am more inclined to blame out networks team for than anything else)
20:12 xarses Fecn: centos using neutron?
20:12 Fecn Centos, neutron vlan segmentation
20:12 xarses Fecn: what does ethtool -i show for the physical interface?
20:13 Fecn We can't get ubuntu to work because the network modules are loaded in a different order to the centos bootstrap... so our 10gibt nices show up as eth0/1 instead of eth8/9
20:13 Fecn nices == nics
20:16 Fecn xarses: any particular interace you want to see? eth0 = admin, eth1 = private (vlan seg range) , eth8 = 10gvit, maangement, storage, public (tagged vlans)
20:17 Fecn I am fortunate enough to have two racks of gear for our proof-of-concept test setup.... so I've just deployed another vlan-segmentation config into the other rack (where networking seems altogether more reliable)
20:17 Fecn I'll know whether it works in about 30 mins when I've had the chance to throw some VMs on there
20:21 Fecn1 joined #fuel
20:31 rongze joined #fuel
20:31 Fecn joined #fuel
20:51 vkozhukalov joined #fuel
20:56 jhurlbert What is the best way to go about modifying the Cobbler images that Fuel uses for the bare metal OS provisioning?
21:03 tsduncan__ xarses: yes. would you like me to post to paste.openstack.org?
21:04 tsduncan__ xarses: http://paste.openstack.org/show/54978/
21:13 xarses tsduncan__: thanks
21:15 xarses jhurlbert: cobbler only uses one image 'bootstrap' to do node discovery. During cloud install cobbler uses anaconda or preseed (centos, ubuntu respecivly) configurations to prepare the base nodes, they are installed from repo's in this case so there is no image
21:16 xarses jhurlbert: what are you trying to achieve ?
21:17 jhurlbert so cobbler pulls down the ubuntu (in our case) cloud image, then uses preseed to modify?
21:17 jhurlbert i was possibly thinking of changing the root pw, but if they are the ubuntu cloud images, then i probably wont
21:18 jhurlbert Also, is there a way to put auth around the Fuel web UI?
21:19 xarses jhurlbert: no, it uses a preseed initrd image image to run preseed
21:19 xarses jhurlbert: not currently
21:19 jhurlbert Ah, understood.
21:19 jhurlbert Thanks for your help.
21:19 xarses the root password should be set in the pre-seed template, you should be able to change it there
21:20 xarses also, the nodes them selves have password login disabled for root they can only use the ssh keys from root on the fuel node
21:20 xarses i should say password login over ssh
21:20 jhurlbert xarses: ah, sure enough, there it is in the preseed. I'm not sure why i didn't think of that.
21:21 xarses the console still works for password login
21:21 jhurlbert yes, it was mostly for console login if ever needed.
21:21 xarses jhurlbert: there are too many parts to remember every thing =)
21:22 jhurlbert ha, i'm finding that out more and more as i dig into the puppet manifests and how cobbler does it job
21:22 jhurlbert xarses: thanks again!
21:29 albionandrew joined #fuel
21:32 albionandrew HI xarses do you or any of your guys have any docs on finding bottle necks in ceph? We have a cluster up, same hardware configured as before, same network layout, the only thing changed is we are using ceph rather than cinder. We ran bonnie against a volume attached to an instance and the performance was worse. ... just thought you might have something ?
21:41 mihgen joined #fuel
21:53 xarses albionandrew: I don't have anything useful to point you to, I would note that unless you did something to your disk topology you have num_replicas * write_iop in before an write is ack'ed so its common to see some loss there
21:54 albionandrew xarses thanks
21:54 xarses albionandrew: its probably better asked in #ceph on oftc network
21:55 vkozhukalov joined #fuel
21:55 albionandrew xarses sure. Thanks, have a nice weekend.
22:02 IlyaE joined #fuel
22:42 e0ne joined #fuel

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