Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel, 2014-03-31

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

All times shown according to UTC.

Time Nick Message
00:01 e0ne joined #fuel
00:03 e0ne_ joined #fuel
00:43 piersdd joined #fuel
00:49 piersdd Hi, all.  I'm not sure where to look to troubleshoot Fuel 4.1 error.
00:50 piersdd Compute nodes are HP Blades which have Broadcom bnx2x nics.  PXE bootstrapped linux does not appear to support them. "bnx2x 0000:02:00.0: firmware: requesting bnx2x/bnx2x-e1h-7.2.51.0.fw"
00:53 piersdd So I get the raw nics, but no VLAN interfaces.  Any idea where I should look to find out how get these working? TIA
01:01 e0ne joined #fuel
01:25 ilbot3 joined #fuel
01:25 Topic for #fuel is now Fuel 4.1 for Openstack: http://fuel.mirantis.com/ | Paste here http://paste.openstack.org/ | IRC logs http://irclog.perlgeek.de/fuel/
01:44 thehybridtech joined #fuel
01:59 piersdd xdeller, How did you resolve your HP BL460c bnx2x firmware issue from bqck in October?
02:01 e0ne joined #fuel
02:23 piersdd sorry xdeller, it was in-fact dhblaz's HP Broadcom nic problem
03:01 e0ne joined #fuel
03:41 mihgen joined #fuel
04:01 e0ne joined #fuel
04:27 IlyaE joined #fuel
04:56 piersdd joined #fuel
05:01 e0ne joined #fuel
05:23 Ch00k joined #fuel
05:32 e0ne joined #fuel
05:38 novel does fuel have a proper REST API documentation?
05:38 mirrorbox joined #fuel
05:38 mirrorbox http://docs.mirantis.com/fuel-dev/develop/api_doc.html#module-nailgun.api.handlers.node
05:38 mirrorbox this one misses request bodies format
05:56 Arminder joined #fuel
06:01 e0ne joined #fuel
06:02 piersdd joined #fuel
06:12 Arminder joined #fuel
06:19 alex_didenko joined #fuel
06:25 mihgen mirrorbox: that's the only REST API docs we have at the moment I believe..
06:25 mihgen anything specific we could help with?
06:29 mirrorbox I've already found https://github.com/stackforge/fuel-web/blob/master/nailgun/nailgun/fixtures/sample_environment.json
06:29 mirrorbox which I guess is almost the same
06:30 mihgen yep
06:30 mihgen NodeHandler is pretty tricky though
06:31 mirrorbox why is that?
06:36 IlyaE joined #fuel
06:37 mihgen it is used for nodes registration, and extensive validation is performed to avoid double-registration of node
06:37 mihgen there was https://blueprints.launchpad.net/fuel/+spec/nailgun-agent-handler
06:37 mihgen don't know, likely we have another handler now for nodes registration
06:40 mirrorbox NodeAgentHandler it seems
06:40 mirrorbox the doc I linked above doesn't seem to reflect that though
06:41 mihgen hmmm that is autogenerated doc
06:45 mihgen weird. docstring is in place, docs build is green..
06:45 mirrorbox 138     r'/nodes/?$',
06:45 mirrorbox 139     NodeCollectionHandler,
06:45 mirrorbox 140     r'/nodes/agent/?$',
06:45 mirrorbox 141     NodeAgentHandler,
06:45 mirrorbox that's what you're talking about, right?
06:45 mirrorbox NodeAgentHandler?
06:45 mihgen yep, separated handler was introduced
06:45 mihgen to register nodes
06:46 Arminder- joined #fuel
06:46 mihgen and update metadata periodically, and to keepalive
06:50 al_ex joined #fuel
06:53 akislitsky joined #fuel
06:54 piersdd joined #fuel
07:10 Arminder joined #fuel
07:12 Ch00k joined #fuel
07:12 asledzinskiy joined #fuel
07:32 mihgen joined #fuel
07:54 dburmistrov joined #fuel
07:59 geogdog joined #fuel
08:03 Ch00k joined #fuel
08:07 topochan joined #fuel
08:12 akislitsky joined #fuel
08:17 e0ne joined #fuel
08:49 mattymo joined #fuel
09:07 topochan joined #fuel
09:08 e0ne_ joined #fuel
09:16 evg joined #fuel
09:20 piersdd mihgen: are you aware of how one might get non-standard kernel stuff into Fuel?  I'm trying to evaluate Mirantis 4.1 using HP Blade servers with 10GBe Broadcom based NICs
09:22 pbrooko joined #fuel
09:23 mirrorbox why does NodeAgentHandler PUT handler uses db().commit() almost after each step?
09:28 bookwar joined #fuel
09:38 evgeniyl mirrorbox: it's legacy, we have a lot such old, bad, places in our code, Nick is working on refactoring of such messy places.
09:39 mihgen hi piersdd
09:39 piersdd hi mihgen
09:39 mihgen we actually gonna have 3.10 for bootstrap
09:40 mihgen and for target nodes, we already provide an option to install newer kernel for centos
09:40 mihgen what' s the issue do you experience now?
09:40 mihgen is it in bootstrap that your NICs can't be discovered now?
09:41 piersdd yes
09:42 piersdd trying to dig up where it was referenced by an earlier user.. in version 3.1
09:43 mihgen piersdd: take a look at this email thread, feel free to participate as well: https://lists.launchpad.net/fuel-dev/msg00719.html
09:43 piersdd I note MiroslavAnashkin commented on the 13th "We already got 4 different change requests addressed to fix issues with Fuel 4.1 OS boot on HP hardware."
09:43 piersdd thanks
09:43 mihgen do you think 3.10 kernel will work for it?
09:44 piersdd not sure
09:52 piersdd joined #fuel
10:12 piersdd mihgen: Can you please point me in the direction of the 'Fuel Utility', to workaround; Bug #1260492 (https://bugs.launchpad.net/fuel/+bug/1260492)
10:17 piersdd mihgen: also Bug #1275663, as I've both Broadcom 1GBe and 10GBe blades slated for our compute lab
10:22 piersdd mihgen:  How would you recommend I test 3.10 kernel?  Install CentOS 6.5 then upgrade kernel to 3.10?
10:26 piersdd does anyone know if the Mirantis roadmap looks to putting OpenStack components in LXC containers?
10:50 warpig joined #fuel
10:58 vkozhukalov joined #fuel
11:04 mihgen piersdd: emm
11:04 mihgen aglarendil: ^^^
11:07 mihgen piersdd: let me find folks who know more about it
11:07 mihgen according to https://review.openstack.org/#/c/62121/5 - we have required pieces on bootstrap
11:08 mihgen it still doesn't see your NICs, does it?
11:09 piersdd PXE Bootstrap sees the NICs.. but cannot implement VLANs on them.. dmesg:
11:09 piersdd bnx2x 0000:02:00.0: firmware: requesting bnx2x/bnx2x-e1h-7.2.51.0.fw
11:09 piersdd bnx2x 0000:02:00.0: eth0: using MSI-X  IRQs: sp 57  fp[0] 59 ... fp[2] 61
11:09 piersdd ADDRCONF(NETDEV_UP): eth0: link is not ready
11:09 mihgen > roadmap looks to putting OpenStack components in LXC containers?   - we don't plan this at the moment. We will see how it works for Fuel Master node first - https://blueprints.launchpad.net/fuel/+spec/fuel-upgrade
11:10 piersdd (lxc) cool.. thanks
11:10 mihgen oh that's interesting
11:10 mihgen can't implement vlans == openvswitch?
11:11 mihgen or Linux?
11:11 piersdd ah.. OK.. probably my ignorance of linux then.  so sorry
11:11 piersdd not sure
11:11 piersdd which logs to check?
11:11 mihgen is it on installation phase, right?
11:12 mihgen when you deploy openstack, not in bootstrap
11:13 piersdd pre-deploy.  nodes enrolled.  controllers (on Supermicro) have a bunch of relevent VLANs.. computes, in HP Blade do not.
11:13 mihgen wait, in pre-deploy we don't configure vlans
11:13 mihgen only network verification does only for checks
11:13 mihgen it enables Linux vlans, and then reverts all network settings back
11:14 piersdd OH!  really my bad then..
11:14 mihgen so the actual configuration is done by l23network puppet module when base OS provisioning is done
11:14 xdeller <piersdd>: with installation of *both* newer fw and kernel(3.10 from elrepo plus new mips fw)
11:15 piersdd OK.  Must be some switch VLAN issue at my end.  please get back to your excellent work on this awesome project.
11:18 piersdd xdeller: So said "installation of *both* newer fw and kernel" is included by default in Mirantis 4.1?
11:18 xdeller afaik no, only there was only kernel-ml
11:19 piersdd OK. thanks.  I'll try and figure out how to do that then.  thanks again
11:19 xdeller https://packages.debian.org/wheezy/firmware-bnx2x disassemble this and put files to /lib/firmware
11:20 xdeller CI folks, it is necessary to provide two firmware slices, one with 'stock' kernel and newer one with 3.10 w/ appropriate deps
11:20 xdeller in the next release I suppose
11:23 xdeller sorry, for link from above wheezy-backports or sid sounds more appropriate. I remember that only bleeding edge fw worked for us
11:26 piersdd thanks xdeller
11:33 e0ne joined #fuel
11:41 aglarendil|nb joined #fuel
11:44 e0ne joined #fuel
11:45 topochan joined #fuel
11:51 saju_m joined #fuel
12:06 saju_m joined #fuel
12:19 evg joined #fuel
12:53 vk joined #fuel
13:04 e0ne joined #fuel
13:07 vkozhukalov joined #fuel
13:19 aglarendil|nb joined #fuel
13:23 justif joined #fuel
13:26 Dr_Drache joined #fuel
13:31 BillTheKat joined #fuel
14:08 thehybridtech joined #fuel
14:39 mattymo1 joined #fuel
14:58 IlyaE joined #fuel
15:09 anotchenko joined #fuel
15:15 anotchenko joined #fuel
15:34 mihgen joined #fuel
15:38 jobewan joined #fuel
15:40 MiroslavAnashkin joined #fuel
15:50 jhurlbert joined #fuel
15:51 jhurlbert Can you use Fuel to deploy additional nodes after the initial cluster has been deployed?
15:54 warpig jhurlbert: yes, you can...  Anything other than controller nodes from what I've tested.
15:54 anotchenko joined #fuel
15:54 jhurlbert warpig: awesome, thanks.
15:54 justif joined #fuel
15:54 jhurlbert warpig: will fuel also update the existing nodes with information about the new node? like hosts, conf files, etc?
15:56 the_hybrid_tech joined #fuel
15:56 warpig jhurlbert: I did an initial deployment of a controller and two compute/storage nodes and added an additional one later and it all looked good.  New node resources were reported in Horizon...
15:57 jhurlbert warpig: great, thank you for the info.
15:58 warpig no prob
15:58 mihgen > Anything other than controller nodes from what I've tested.  - it's rather a bug which should be addressed..
15:58 mihgen adding controllers in HA mode should not cause issues
16:01 e0ne_ joined #fuel
16:04 crandquist joined #fuel
16:05 rmoe joined #fuel
16:12 e0ne joined #fuel
16:27 MiroslavAnashkin joined #fuel
16:30 IlyaE joined #fuel
17:02 dburmistrov joined #fuel
17:06 Ch00k joined #fuel
17:17 xarses joined #fuel
17:26 xarses warpig: you are having problems deploying additional controllers?
17:26 vkozhukalov joined #fuel
17:30 jhurlbert joined #fuel
17:32 angdraug joined #fuel
17:41 e0ne joined #fuel
17:49 anotchenko joined #fuel
17:49 MiroslavAnashkin joined #fuel
18:02 jhurlbert joined #fuel
18:08 IlyaE joined #fuel
18:27 IlyaE joined #fuel
18:31 Dr_Drache hate to ask
18:31 xarses ask what?
18:31 Dr_Drache is there a way to maintance the openstack database to trim it down
18:31 Dr_Drache LOL
18:31 Dr_Drache hate to ask because my google-foo is sucking
18:32 xarses I'm sure there is
18:32 xarses all or nearly all of the OpenStack services don't prune data that has been marked as deleted
18:33 Dr_Drache that sucks,
18:33 Dr_Drache my database is @ 1.3M instances, which maybe 300 are live.
18:33 xarses It's an operators prerogative. I understand why OS doesn't prune them for you
18:34 xarses but yes "that sucks"
18:35 xarses There are likely conversations on the openstack operators mailing list that you could find, or you could ask on #openstack or the fore-mentioned mailing list
18:35 xarses lol, look at your keystone db
18:36 xarses we where/are going to include something to clean up the expired tokens table, you can search the fuel-dev ML for that
18:36 xarses I cant remember what package we where going to use, but some toolkit can do some fancy delete stuff while not locking the db all the time
18:48 tatyana joined #fuel
19:05 Dr_Drache xarses, yea.
19:06 Dr_Drache thougfht maybe I was missing something.
19:06 e0ne joined #fuel
19:06 Dr_Drache I will search.
19:44 e0ne joined #fuel
20:06 jhurlbert joined #fuel
20:35 aleksandr_null joined #fuel
20:44 jhurlbert joined #fuel
20:46 BillTheKat joined #fuel
20:51 e0ne joined #fuel
20:56 toha1 joined #fuel
21:21 e0ne joined #fuel
21:24 e0ne joined #fuel
21:38 ruhe left #fuel
21:44 jhurlbert joined #fuel
21:59 e0ne joined #fuel
22:58 piersdd joined #fuel
23:27 justif joined #fuel

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