Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel, 2014-07-31

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

All times shown according to UTC.

Time Nick Message
00:29 rmoe joined #fuel
00:36 geekinutah joined #fuel
00:48 fhond joined #fuel
00:52 dpyzhov joined #fuel
00:57 fhond joined #fuel
01:07 kodiakF_athome joined #fuel
01:35 xarses joined #fuel
03:44 jobewan joined #fuel
04:06 boris-42 joined #fuel
04:07 geekinutah joined #fuel
04:45 ArminderS joined #fuel
04:57 neophy joined #fuel
05:00 neophy I installed Openstack Havana using Fuel 4.1. Now I trying to manually configure VPNaaS. I am following this doc: https://gist.github.com/cloudnull/8851787. How do I install openstack-neutron-vpn-agent? does yum install openstack-neutron-vpn-agent work?
05:01 neophy I am using Centos 6.4 . Any help?
05:02 neophy I am not able to fine openstack-neutron-vpn-agent rpm for Centos 6.
06:18 Longgeek joined #fuel
06:22 Longgeek joined #fuel
06:47 geekinutah joined #fuel
06:49 e0ne joined #fuel
06:52 pasquier-s joined #fuel
06:59 artem_panchenko joined #fuel
07:01 hyperbaba joined #fuel
07:09 adanin joined #fuel
07:28 pasquier-s joined #fuel
07:34 jaranovich_ joined #fuel
07:40 al_ex12 joined #fuel
07:46 e0ne joined #fuel
08:05 Longgeek joined #fuel
08:14 ykotko_ joined #fuel
08:36 xarses joined #fuel
08:46 e0ne joined #fuel
09:48 al_ex12 joined #fuel
10:05 e0ne joined #fuel
10:28 dilyin joined #fuel
11:04 al_ex12 joined #fuel
11:11 kat_pimenova joined #fuel
11:28 Longgeek_ joined #fuel
11:58 pasquier-s joined #fuel
12:08 adanin joined #fuel
12:18 e0ne joined #fuel
12:58 getup- joined #fuel
13:14 teknomega joined #fuel
13:15 rmoe joined #fuel
13:15 aleksandr_null joined #fuel
13:16 dpyzhov joined #fuel
13:16 GeertJohan joined #fuel
13:16 GeertJohan joined #fuel
13:16 gleam joined #fuel
13:18 jaranovich_ joined #fuel
13:20 mattymo joined #fuel
13:50 al_ex12 joined #fuel
14:05 al_ex12 joined #fuel
14:08 geekinutah joined #fuel
14:09 adanin joined #fuel
14:20 e0ne joined #fuel
14:23 getup- joined #fuel
14:31 mattgriffin joined #fuel
14:38 Longgeek joined #fuel
14:57 angdraug joined #fuel
15:07 adanin joined #fuel
15:50 rmoe joined #fuel
15:55 ArminderS joined #fuel
16:12 tdubyk joined #fuel
16:15 xarses joined #fuel
16:25 mutex joined #fuel
16:31 geekinutah joined #fuel
16:45 TVR_ can I get a sanity check here please? If I edit the "default" access and security group and add any ICMP to the ingress rule... and then spin up an instance with that as the template used, I should then be able to ping the floater, yes?
16:49 angdraug yes
16:50 geekinutah joined #fuel
16:51 bookwar joined #fuel
16:51 TVR_ ok... so that is not happening.. and I have a pattern to recreate this.... do you mind if I bounce some ideas off of you?
16:56 TVR_ so, what services control access and security within OpenStack? what services can I bounce?
16:56 geekinut1h joined #fuel
17:15 rmoe joined #fuel
17:51 ykotko_ joined #fuel
17:51 KodiakF hypothetical here - say I had some crappy blades w/ no disk space and I wanted to use fuel - could I bring those up as compute nodes with fuel, then jump onto the nodes and set up an nfs mount for /var/lib/nova/instances then let fuel start doing it's thing again?
18:02 kat_pimenova_ joined #fuel
18:02 jkirnosova joined #fuel
18:09 ArminderS joined #fuel
18:11 adanin joined #fuel
18:15 adanin joined #fuel
18:19 xarses KodiakF: by default fuel expects to be able to install the OS for you
18:20 KodiakF xarses:  Oh for sure - and I want it to.  I'm just hoping between bare metal provision and node role assignment I can hop on the box and add an nfs mount
18:20 KodiakF does that sound like something that would work out or does puppet manage fstab on the nodes?
18:20 xarses Fuel's process is broken into three steps, discovery, provision, and deployment. You could conceivably allow the discover to occur (or do it your self), you can do the provision your self, and then let fuel do the deployment
18:22 KodiakF Well in my mind I'd hopefully have Fuel discover, provision (then I ninja in and add the nfs mount) then I assign the role of compute and Fuel is none the wiser about the nfs mount I added to the compute node
18:22 xarses the provision and deployment steps can be executed by hand using the cli so they don't have to be coupled together
18:23 xarses none of the mounts are mangled during the puppet run, it's all done during provisioning
18:23 xarses So it could work
18:26 xarses TVR_: first ensure that one of the other endpoints on the private network can ping the instance to ensure that it's not a transport issue. you should be able to use either the dhcp namespace or qrouter to do this
18:27 KodiakF Cool.  Filing this away for an upcoming capabilities discussion
18:27 angdraug joined #fuel
18:27 KodiakF mgmt wants to compare Nebula One vs Fuel vs (possibly) Dell "Crowbar" which looks practically defunct
18:28 xarses assuming that the private network is transporting properly, then you probably have issues with the l3-agent service
18:28 KodiakF oh the questions are all theoretical at this point
18:28 KodiakF mgmt is trying to pick a deployment product since manual deployment takes too long as we've seen in our first iteration of openstack from scratch
18:29 angdraug last I checked Nebula was behind the curve by 1.5 openstack releases, and had a much narrower functional scope
18:29 KodiakF That's what I was seeing - oh don't you just get soooo much for 100k hahaha
18:29 KodiakF narrower functional scope at least - the being so far behind is a revelation
18:30 KodiakF surprised neutron even works to an acceptable degree if it's on Griz
18:32 angdraug http://www.openstack.org/marketplace/distros/distribution/nebula/nebula-one-cloud-controller
18:32 angdraug still on grizzly
18:33 TVR_ so here is what I am seeing.. FYI: we are using heat templates to deploy stacks
18:36 TVR_ We build an environment... if we use the default unmodified, and use heat to deploy, all instances after, no matter what, keep the settings for default, which is no access in.... if we build an environment fresh, but we modify the default adding all tcp and all udp and all icmp in... then spin a heat template, all instances thereafter have all access.... even if we remove the all rules from default and spin up manually....
18:36 TVR_ it seems the rules get pulled into memory and don't get flushed or something
18:37 adanin joined #fuel
18:59 KodiakF angdraug:  Thanks much for the link that's good ammo
19:32 f13o joined #fuel
19:39 wrale_ joined #fuel
19:39 e0ne joined #fuel
19:42 wrale_ my instance times out when trying to contact the metadata service .. 5.0.1, ceph, neutron gre, ubuntu cloud 12.04.4 image
19:42 wrale_ + HA... any tips to debug?
19:42 wrale_ horizon is altogether slow with console and instance logs
19:44 wrale_ nova-oslo.messaging._drivers.impl_rabbit ERROR: Failed to consume message from queue: Socket closed <?
19:44 TVR_ 5.01? Do you have an image for that?
19:44 wrale_ i made one
19:45 wrale_ it's over a week old now
19:45 TVR_ damn it jim..
19:45 wrale_ buggy
19:45 TVR_ (a  little star trek refference)
19:45 wrale_ yup.. lol
19:45 TVR_ I need 5.01 as 5.0 doesn't have the drivers for my HP's
19:46 TVR_ 4.1 is rock solid... but I need to use it a bit non-standard... so I run into issues
19:47 wrale_ https://gist.github.com/josh-wrale/72ac65127cfdf4f319f5 < doc i wrote about building 5.0.1
19:49 TVR_ thx
19:50 TVR_ we are a heavy AWS shop here.... we just installed 7 DC's... we want to use heat and cloud formation to uniformly scale out either environment...
19:51 TVR_ heat keeps breaking my HA environments and I keep having to stabilize them.... the fun of it all
19:54 wrale_ i'm about fed up with openstack as a whole, tbh.. lol..
19:55 jaypipes joined #fuel
19:55 wrale_ and to the fuel devs, shipping minimum viable product should entail disabling features if they don''t work as advertised
19:58 TVR_ I like openstack in general.... I just sometimes wish it was designed more like a skyscraper....you build the first 5 floors and then build the 6'th.... what you have when done is access (still) to the first 5 floors, and when done you can access the 6'th.... what you DON't do is redesign the first 5 floors to add the 6'th
20:10 wrale_ any reason why tenant 1 can reach metadata server, but tenant 2 cannot, using neutron ?  tenant networks and router are setup.. i can attach a floating ip
20:11 rmoe normally when a vm can't reach the metadata service it's because the vm didn't get an IP from DHCP
20:16 e0ne joined #fuel
20:28 e0ne joined #fuel
20:29 blahRus joined #fuel
20:29 wrale_ i'm tempted to try this to speed everything up.. as an aside https://www.digitalocean.com/community/tutorials/how-to-setup-additional-entropy-for-cloud-servers-using-haveged
20:29 aleksandr_null Guys, do we use memcached for keystone in 5.0 ?
21:21 geekinutah joined #fuel
21:34 teknomega hey rmoe any idea on how to wipe that partition a bit better
21:34 teknomega what we are doingis not working
21:34 teknomega dd command does not wipe sdb4
21:35 rmoe what are you seeing that indicates the filesystem is still there?
21:35 teknomega the ceph partition
21:35 teknomega parted /dev/sdb show xfs partition
21:35 teknomega parted /dev/sdb list
21:35 teknomega print
21:35 rmoe did ceph-deploy fail?
21:35 teknomega let me check
21:35 rmoe if you couldn't mount that partition I can't believe that the file system is still there, so ceph-deploy should succeed
21:35 teknomega i just rebooted the Fuel server
21:36 teknomega let me get some logs and post them
21:36 teknomega on patebin
21:36 teknomega its just not installing and keeps erroring out
21:37 teknomega i am going to wipe the sdb4 partition
21:37 teknomega reboot the servers
21:37 teknomega and then redeploy using the deploy button on Fuel
21:39 teknomega http://pastebin.com/RdEVhEV6
21:40 teknomega here is the entire log file
21:40 teknomega http://pastebin.com/zpur8mk3
21:52 teknomega ???
21:54 teknomega left #fuel
21:54 teknoprep joined #fuel
22:05 rmoe yeah, all of those failures are because the OSd failed
22:05 rmoe and it failed because of the existing filesystem on it
22:05 rmoe if that's gone the OSD will be created successfully and everything else should work
22:21 geekinutah where is source code for Mirantis OpenStack made available?
22:22 angdraug geekinutah: it's built out of gerrit.mirantis.com
22:22 geekinutah I'll ask a different sort of question
22:23 geekinutah If I wanted to diff Mirantis Openstack against tree.....
22:23 geekinutah errr, not tree, trunk
22:23 geekinutah damn woody works
22:23 geekinutah *words
22:24 xarses well, opestack uses git, so there is no trunk... =p
22:25 geekinutah xarses so snarky
22:25 xarses =D
22:25 geekinutah but I wasn't asking about opestack *smirks*
22:25 geekinutah OpenStack
22:26 xarses You would have to extract the python from the packages and compare it against what ever branch you are looking for
22:29 geekinutah omg
22:52 angdraug geekinutah: if you have access to gerrit.mirantis.com, you could just use git :)
22:52 angdraug but you'll need vpn for that
22:53 angdraug also recently we started putting git commit ids into changelogs of binary packages
22:53 angdraug so you should be able to track down which exact commit a particular binary was built from
22:56 adanin joined #fuel
23:06 geekinutah for an open distribution of OpenStack, it sure is convoluted to get the source
23:06 geekinutah just saying
23:31 angdraug hey, at least we're moving in the right direction, opening more stuff with time vs closing stuff down
23:34 angdraug geekinutah: https://lwn.net/Articles/430098/
23:35 angdraug keep reminding people that it's a problem and eventually we'll fix it :)
23:35 f13o joined #fuel
23:57 adanin joined #fuel

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