Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel, 2014-06-26

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

All times shown according to UTC.

Time Nick Message
00:37 rmoe joined #fuel
00:39 e0ne joined #fuel
01:32 xarses joined #fuel
01:39 e0ne joined #fuel
02:09 jonathan4 joined #fuel
02:39 e0ne joined #fuel
03:11 mattgriffin joined #fuel
03:39 e0ne joined #fuel
03:41 xarses joined #fuel
04:34 jonathan4 i give up.. :-P
04:35 jonathan4 i did a stock install of fuel on a fresh host and when i pxeboot it installs my controller fine but my compute nodes barf with that python error
04:39 e0ne joined #fuel
05:39 e0ne joined #fuel
06:22 evt_ joined #fuel
06:23 evt_ left #fuel
06:26 e0ne joined #fuel
06:43 guillaume_ joined #fuel
06:51 e0ne joined #fuel
06:55 e0ne_ joined #fuel
06:59 e0ne joined #fuel
07:26 ddmitriev joined #fuel
07:38 JY_ joined #fuel
07:52 odyssey4me joined #fuel
08:12 e0ne joined #fuel
08:16 brain461 joined #fuel
08:26 neophy joined #fuel
08:37 neophy I am using Fuel 4.1 to deploy OpenStack Havana. I am getting this error in network verification step.
08:37 neophy but all my networking configuration are correct. because I used the same network setting earlier
08:38 neophy it was working fine
08:38 neophy any one faced similar issue?
08:56 brain461 joined #fuel
09:12 taj joined #fuel
09:39 hyperbaba joined #fuel
09:41 sbartel joined #fuel
09:41 sbartel hello
09:41 sbartel i've got a question
09:42 sbartel after some fuel node instalaltion I have a problem when trying to access to fuel services (UI and CLI)
09:42 sbartel 502 badgateway
09:42 sbartel when access to nginx container (dockerctl shell nginx) I can see connection refused on error log of nginx
09:42 sbartel any clue of what can be  the origin of this problem (for info fuelmaster node is deployed on virtualbox with  launch.sh script)
09:44 mattymo sbartel, nailgun container has an issue
09:45 mattymo run dockerctl logs nailgun
09:47 brain461 joined #fuel
09:50 neophy some is there?
09:50 neophy any help?
09:50 mattymo neophy, create a bug on bugs.launchpad.net/fuel with a diagnostic snapshot
09:52 sbartel mattymo: thank, it was it thank you i'll debug
09:52 mattymo the logs might show what went wrong
09:54 sbartel yes
09:54 sbartel it was a problem of db migration
09:54 sbartel about update we made for vxlan support
09:55 sbartel had already this kind of error for 4.1 but without docker problem in logs were a little bit easier to find
09:55 mattymo sbartel, you can just dockerctl shell nailgun, make changes in place, then try to rerun start.sh
09:55 mattymo yes, it makes debugging harder, but upgrades are easier
09:55 sbartel ok i'll try that
09:58 e0ne joined #fuel
09:58 sbartel dumb question what editor to use in dockerctl shell nailgun
09:58 sbartel vi seems to not been accessible
09:59 mattymo yum -y install vi vim
09:59 mattymo these are very minimal containers with no extra packages
10:00 mattymo there's a feature request to install editors and diagnostic tools that's in progress
10:00 sbartel yes thanks !
10:04 hyperbaba Hi,
10:05 hyperbaba I am trying to get functional HA openstack on a 4 node cluster with just two controllers.
10:06 hyperbaba the problem is the quorum in crm. I am ready to risk and disable quorum on crm but i am getting this after crm configure property no-quorum-policy=ignore
10:07 hyperbaba ERROR: cib-bootstrap-options: attribute mysqlprimaryinit does not exist
10:07 hyperbaba ERROR: cib-bootstrap-options: attribute mysqlepoch does not exist
10:07 hyperbaba Do you still want to commit?
10:07 hyperbaba does anybody knows is it safe to continue?
10:13 mattymo hyperbaba, looks like we made a bug setting up 2-node HA
10:13 mattymo it was never tested, but it worked with 1 or 2 HA nodes without any errors in the past
10:14 mattymo hyperbaba, could you make a diagnostic snapshot and file a bug with this error? it will take a bit of investigation to solve
10:17 hyperbaba i've already made a related bug report #1333209 .Diagnostic snapshot provided and marked private.
10:18 hyperbaba mattymo: bug is related to deployment it self but it is related
10:20 brain461 joined #fuel
10:47 jonathan4 no matter what i do i still get failures right at the end when setting it up
10:48 jonathan4 the "pecan state... thread._local.... HTTPNotFound: The resource could not be found."
10:48 jonathan4 i want to get this working and demo it so i can sign up for support :-(
10:48 jonathan4 it's been three days of trying every possible combination and it fails on installation
10:48 jonathan4 i get "Deployment has failed.  check these nodes:"
10:51 jonathan4 i'm very discouraged.  i know nothing is perfect in the open source world, and i want to sign up for support but if i don't get this set up by friday morning i suspect they are going to go with something crappy like microsoft hyper V
10:54 jonathan4 2014-06-24 16:46:34 ERROR
10:54 jonathan4 (hooks) Pecan state <thread._local object at 0x164edb0>
10:54 jonathan4 Traceback (most recent call last):
10:54 jonathan4 File "/usr/lib/python2.6/site-packages/pecan/core.py", line 570, in __call__
10:54 jonathan4 self.handle_request(req, resp)
10:54 jonathan4 File "/usr/lib/python2.6/site-packages/pecan/core.py", line 421, in handle_request
10:54 jonathan4 controller, remainder = self.route(req, self.root, path)
10:54 jonathan4 File "/usr/lib/python2.6/site-packages/pecan/core.py", line 257, in route
10:54 jonathan4 node, remainder = lookup_controller(node, path)
10:54 jonathan4 File "/usr/lib/python2.6/site-packages/pecan/routing.py", line 60, in lookup_controller
10:54 jonathan4 raise exc.HTTPNotFound
10:54 jonathan4 this is deploying to actual hardware
10:55 jonathan4 it fails all the way near the end
11:02 hyperbaba jonathan4: did you setup the switch for needed vlans from the network settings?
11:02 jonathan4 no, just used flatvlan i think
11:03 jonathan4 i selected "neutron with gre" because the other one was failing too
11:03 jonathan4 is it because my controller is running on a virtual vm and my compute nodes are physical hardware?
11:03 hyperbaba if you have 2 interfaces per node , just move all services from first interface to second leaving just admin role on first. Then you need atleast vlan 101 and 102 on second interface for all the nodes
11:04 hyperbaba this must be the case
11:04 hyperbaba if you don't propagate vlans to vm instance
11:04 jonathan4 ok so if i use physical hardware instead of VM it might just work?
11:04 hyperbaba regardless of neutron with_gre you still need vlans for management and storage
11:05 jonathan4 getting new server in today
11:05 hyperbaba if you configure the network part correctly
11:05 jonathan4 so the defaults won't work?
11:06 hyperbaba defaults presume use of vlan 101 and 102 . Look at the network config page on fuel for your environment
11:06 jonathan4 i guess i gotta study up on the networking stuff then... i'm a noob for this.. 20+ years unix&linux but not much cloud stuff
11:06 jonathan4 ok
11:06 jonathan4 maybe for my proof of concept i should just take my two physical nodes and make one controller and one compute even though 16 cores is overkill for controller :)
11:06 hyperbaba I've got puzzled myself
11:07 hyperbaba In a production envirnoment as i see it from expirience 12 cores per controller and 16GB ram is ok
11:07 jonathan4 so you have seen that error message bfore?
11:07 hyperbaba no
11:08 hyperbaba But if you did not configure network  part as needed, that's a possibility
11:08 hyperbaba that nodes can't communicate using managment and /or storage channels
11:08 odyssey4me joined #fuel
11:08 jonathan4 but if i take all the defaults shouldn't it just work?
11:08 hyperbaba the admin network is just for communicating with fuel
11:09 jonathan4 so is it possible that the reason it failed is because my controller is virtualized but my compute nodes are physical?
11:09 hyperbaba yes. If you have network layer to support the defaults
11:09 jonathan4 i see
11:09 jonathan4 ok, trying again :)
11:10 jonathan4 i guess getting stumped by technology is what keeps us humble
11:10 hyperbaba configure the secondary interface ports on switch to support vlans 101 and 102 and also native vlan untagged
11:10 jonathan4 you mean on physical switch?  i don't have access to that  but i think it is just a dell 2748
11:11 hyperbaba As my friend said "Linux is operating system designed for configuring"
11:11 hyperbaba yes
11:11 hyperbaba phisical switch needs to be configured
11:11 odyssey4me_ joined #fuel
11:11 jonathan4 hmmm
11:11 jonathan4 ok
11:12 jonathan4 i'll work on that
11:25 Kupo24z joined #fuel
11:35 odyssey4me joined #fuel
11:42 hyperbaba Anyone successfully used murano deployments? I've got working openstack with ceph and murano installed but unable to deploy the murano environments. No errores reported in dashboard (just empty logs for components) and no instances deployed
11:47 jonathan4 hyperbaba, do you know if proxmox handles vlan tagging?  i am redeploying without using my proxmox VE and am hoping the problem was due to the proxmox vif ignoring openstack tagging
11:47 jonathan4 wheneverthing was 100% within proxmox VE and virtualbox it worked but when i hybridized it broke
12:01 ddmitriev joined #fuel
12:02 odyssey4me_ joined #fuel
12:10 hyperbaba jonathan4: dunno , never used proxmox
12:23 ddmitriev joined #fuel
12:23 e0ne joined #fuel
12:26 e0ne_ joined #fuel
12:55 brain461 joined #fuel
13:37 fandi joined #fuel
13:53 mattgriffin joined #fuel
14:23 jobewan joined #fuel
14:39 xarses joined #fuel
14:42 wrale_ an option to sort the unallocated nodes by name would be awesome.. also, if i point /etc/resolv.conf toward my ipmi dnsmasq server, would the names auto-populate?
14:42 wrale_ (from fuel node)
14:48 mattymo wrale_, you should configure it in fuelmenu as your upstream dns server
14:48 mattymo assuming it can also do reverse lookups for all other domains, like google.com
14:48 wrale_ mattymo: cool.  thank you.. and i will test to be sure before changing
14:51 mattymo for sorting, no, I believe there's nothing like that
15:03 aepifanov joined #fuel
15:12 guillaume_ joined #fuel
15:26 blahRus joined #fuel
15:47 geekinutah joined #fuel
15:54 xarses joined #fuel
15:57 jonathan4 nope.. still same error deploying no matter what i do
15:59 jonathan4 deployment still fails on my two node cluster:
16:00 jonathan4 (hooks) Pecan state <thread._local object at 0x164edb0>
16:00 jonathan4 Traceback (most recent call last):
16:00 jonathan4 File "/usr/lib/python2.6/site-packages/pecan/core.py", line 570, in __call__
16:00 jonathan4 self.handle_request(req, resp)
16:00 jonathan4 File "/usr/lib/python2.6/site-packages/pecan/core.py", line 421, in handle_request
16:00 angdraug joined #fuel
16:00 jonathan4 controller, remainder = self.route(req, self.root, path)
16:00 jonathan4 File "/usr/lib/python2.6/site-packages/pecan/core.py", line 257, in route
16:00 jonathan4 node, remainder = lookup_controller(node, path)
16:00 jonathan4 File "/usr/lib/python2.6/site-packages/pecan/routing.py", line 60, in lookup_controller
16:00 jonathan4 raise exc.HTTPNotFound
16:00 jonathan4 HTTPNotFound: The resource could not be found.
16:10 taj joined #fuel
16:20 jonathan4 Node NameNode MAC addressNode InterfaceExpected VLAN (not received)
16:20 jonathan4 Untitled (13:16)00:25:90:1c:13:16eth0101,102,103
16:20 jonathan4 Untitled (17:de)00:25:90:1c:17:deeth0101,102,103
16:20 jonathan4 i have no control over the switch.. it's a dell 2748.. is it supressing the vlan tagging?
16:34 dilyin joined #fuel
16:38 dmitryilyin joined #fuel
16:39 dilyin joined #fuel
16:44 albionandrew joined #fuel
16:47 mattgriffin joined #fuel
16:53 ikalnitsky|home joined #fuel
16:54 ikalnitsky|home left #fuel
16:57 jonathan4 ok, so i turned on tagging on my 2748...  i'm building this in a $3,000 environment.. two 1U L5520-based systems and a dell 2748... my hope is to subscribe to mirantis support and run my small public cloud and basically just make a living :)
16:57 jonathan4 not looking to compete with AWS but if i can replace an IT salary running my own cloud for small customers, i'll be thrilled.
16:58 jonathan4 or "stoked" as they say where i live.
17:05 dhblaz joined #fuel
17:08 dhblaz On our production controller nodes /var/lib/mysql/ibdata1 is 26G.  This problem can be mitigated by using the configuration option innodb_file_per_table.  Should I file this on launchpad or is it address in 5.x or perhaps already on the roadmap?
17:08 dhblaz /\ xarses ?
17:12 geekinutah joined #fuel
17:15 xarses hi
17:15 evg dhblaz: hi, there was bug related: https://bugs.launchpad.net/fuel/+bug/1316003
17:16 xarses dhblaz: 5.0 and 4.1 should already have innodb_file_per_table
17:29 keyz182 joined #fuel
17:29 jaypipes joined #fuel
17:32 keyz182 Hey, got a quick question about Fuel. Just installed 5.0 from the isos on the Mirantis website, testing it out. Is there any way to password protect the Fuel-ui?
17:37 xarses keyz182: There is not an easy way at this time, There is an open blueprint for it though and we are actively working on it https://blueprints.launchpad.net/fuel/+spec/access-control-master-node
17:40 wrale_ I'm running into this isssue.  Is there any way to patch my system in place to fix it? https://bugs.launchpad.net/fuel/+bug/1328200
17:44 MiroslavAnashkin wrale_: we are testing if this patch may be directly applied to the 5.0 master node or we need an additional back port.
17:49 wrale_ MiroslavAnashkin:  thank you
17:52 wrale_ Am I unable to continue until a patch is released and tested?
17:57 MiroslavAnashkin This patch unfreezes already frozen environments among the other changes, but you may work with the new envs, if you have additional idle nodes.
17:58 wrale_ Thank you.  I'll try that.
18:11 angdraug MiroslavAnashkin: wrale_: https://review.openstack.org/102267
18:11 angdraug is that the backport you're looking for?
18:15 MiroslavAnashkin this patch is for stable/5.0, which has gone somewhere ahead from 5.0 release tag
18:16 wrale_ i just made my fuel dvd using the stable/5.0 tag a few days back.. hmmm
18:31 jonathan4 question:  in reading all of the openstack documentation it talks about needing three nodes (controller, network, and compute) but it seems like that is not the case.. does mirantis do something that combines network with one of the other ones (rightfully so, IMO)?  or am i just using legacy networking?
18:36 dilyin we use network on controller, so you only need controller and compute.
18:36 jonathan4 cool
18:36 jonathan4 my controller is going to be a solid state dual core jetway with 4 gig ethernets
18:36 jonathan4 the next thing is figuring out how to use my iscsi targets on my freebsd ZFS box
18:37 jonathan4 and have all the storage sitting on that.. i assume adding storage with cinder is very easy if i have an iscsi target?
18:37 dilyin you need just install iscsi tagrget daemon there, it's not that hard to do
18:37 jonathan4 i'm between jobs but have 20+ years unix/linux and just trying to ace openstack to get a kickass job
18:37 jonathan4 yeah
18:38 jonathan4 well, not really between jobs.. i own a small ISP (ASN #54380 ;) but it's not enought o pay the bills
18:38 dilyin i've never tried to use iscsi with cinder. check the docs
18:38 jonathan4 roger that.
18:43 geekinutah joined #fuel
18:47 mattgriffin joined #fuel
18:47 dilyin joined #fuel
18:47 taj joined #fuel
18:47 angdraug joined #fuel
18:47 xarses joined #fuel
18:47 blahRus joined #fuel
18:47 Kupo24z joined #fuel
18:47 sbartel joined #fuel
18:47 jonathan4 joined #fuel
18:47 rmoe joined #fuel
18:47 omelchek joined #fuel
18:47 homegrown joined #fuel
18:47 obcecado joined #fuel
18:47 boris-42 joined #fuel
18:47 bogdando joined #fuel
18:47 MiroslavAnashkin joined #fuel
18:47 aglarendil joined #fuel
18:47 aedocw joined #fuel
18:47 aleksandr_null joined #fuel
18:47 strictlyb joined #fuel
18:47 GeertJohan joined #fuel
18:47 book` joined #fuel
18:47 dpyzhov joined #fuel
18:47 holser joined #fuel
18:47 wrale_ joined #fuel
18:47 bookwar joined #fuel
18:47 sbog joined #fuel
18:47 saibarspeis joined #fuel
18:47 ogelbukh joined #fuel
18:47 AndreyDanin joined #fuel
18:47 mattymo joined #fuel
18:47 sanek joined #fuel
18:47 izinovik joined #fuel
18:47 mrasskazov joined #fuel
18:47 justif joined #fuel
18:47 nurla joined #fuel
18:47 dburmistrov joined #fuel
18:47 mihgen joined #fuel
18:47 meow-nofer joined #fuel
18:47 MorAle joined #fuel
18:47 akislitsky joined #fuel
18:47 apalkina joined #fuel
18:47 jkirnosova joined #fuel
18:47 kat_pimenova joined #fuel
18:47 evg joined #fuel
18:47 monester joined #fuel
18:47 vkramskikh joined #fuel
18:47 jbellone joined #fuel
18:47 mutex joined #fuel
18:47 alex_didenko joined #fuel
18:47 meow-nofer_ joined #fuel
18:47 jseutter joined #fuel
18:47 scroiset_ joined #fuel
18:47 wayneeseguin joined #fuel
18:47 evgeniyl joined #fuel
18:47 christopheraedo joined #fuel
18:47 TVR_ joined #fuel
18:47 Bomfunk joined #fuel
18:47 ChanServ joined #fuel
18:47 Llew joined #fuel
18:47 bas___ joined #fuel
18:47 dmitryme joined #fuel
18:47 bdudko_ joined #fuel
18:48 Bomfunk joined #fuel
18:48 Llew joined #fuel
18:48 TVR_ joined #fuel
18:49 jonathan4 mirantis 5.0 install is definitely broken it seems.
18:50 jonathan4 Deployment has failed.  Timeout of deployment is exceeded.
18:55 jaypipes_ joined #fuel
19:02 wrale_ mine is stuck at 29%...
19:03 dpyzhov joined #fuel
19:07 xarses would you take "well it works for me"?
19:07 xarses =/
19:08 xarses Please create a ticket and upload a support bundle. If privacy of the data is an issue I can make the tickets private
19:08 wrale_ i'm trying to figure out what's wrong with mine.. not sure where to start.. new to fuel
19:08 xarses It's not nice if it gets stuck like that
19:08 jonathan4 i'm not yet a paying customer though... kinda learning this so i can get myself a job so you probably won't help me unless i pay, right?
19:08 jaypipes joined #fuel
19:09 xarses wrale_: please paste output from 'fuel node'
19:12 xarses jonathan4: this is a community channel. Some of us work for Mirantis, such as my self, some are members of the fuel community at large. We will absolutely help to resolve bugs in fuel as some of us are fuel developers, however we have no SLA like the support group does, we help as possible. If you are looking for SLA based response and closure then you should engage Mirantis through our support organization (For which you can start a 30 day trial)
19:13 jonathan4 oh cool.. i'll def do that.. is it expensive?
19:13 xarses honestly I dont know
19:13 jonathan4 ok
19:13 xarses oh, the 30 day trial is free
19:13 jonathan4 ok
19:14 xarses afterwards, I don't know what our support costs are
19:15 jobewan joined #fuel
19:16 geekinutah joined #fuel
19:17 wrale_ i can't seem to see much for all the errors from the bug i ran into up there (different "environment").. the second environment is stuck at 29%... the logs are probably interwoven .. facepalm
19:17 wrale_ i appreciate your help, either way
19:18 xarses wrale_: if you go into the one cluster, what say the nodes are doing?
19:18 xarses They should all have their own progress bars and status'
19:19 wrale_ xarses: it says the nodes are 'installing ubuntu'...i have ipmiview for these nodes (supermicro).. the install left the nodes back at a "node-xx" login screen so that seems like it went well.
19:20 wrale_ the bars are completely blue
19:21 xarses so the failure is with the node reporting the completed status. Check to see if the nodes can all ping the fuel master node
19:24 xarses you can also goto the cli on the fuel node and do 'fuel env' to list the cluster names and id numbers, and then 'fuel nodes --env <id>' from the previous command.
19:24 xarses it might tell us which one is stuck
19:24 wrale_ will try
19:25 xarses nodes that have a state of 'provisioned' are completed and reported back
19:25 xarses nodes with a state of 'provisioning' are considered still in progress
19:26 xarses also if it cant reach fuel, it might have a status of offline
19:28 wrale_ the fuel env works, but I get an HTTP 500 errorwhen i do a 'fuel nodes --env 2' (2 is the environment hanging up)
19:28 Kupo24z I'm having issues installing default-jre even after adding an external (non-fuel mirror) with 'apt-get install default-jre' I dont get this problem on a standard ubuntu 12 install however this occurs on fuel-deployed ones
19:35 albionandrew joined #fuel
19:35 dhblaz xarses: is there a simple way for me to shutdown the entire galera cluster so that I don’t trigger a sync when bringing these nodes back online?
19:35 Kupo24z hmm nvm got it, needed to add backuports and multiverse
19:55 wrale_ " Timeout of provisioning is exceeded "
19:56 dhblaz Does anyone know if it is safe to change wsrep_sst_method in 4.0 installations to rsync?
19:57 xarses dhblaz: not that I'm familiar with, but I'd assume that it can be done
19:57 dhblaz xarses: any setting other than mysqldump that you know is safe?
19:58 dhblaz It is just hard to deal with the 2h of downtime we have now with every controller reboot
19:58 xarses dhblaz: supposedly it should be no problem to change, I've never played with any of the other options
19:58 xarses ya, I know you should be able to force haproxy to take the donor and receipient out of the vip so you can continue working
19:59 xarses I'm also not sure how to do that
19:59 xarses it sounds like its scoped for 6.0 to do that automatically
20:00 dhblaz xarses: if you want I can send you the way I have changed haproxy to cope, but it isn’t perfect.  It seems that some services don’t use haproxy and talk directly to their mysql instance.
20:02 xarses dhblaz: no they do, the problem is handeling "server has gone away" in a timely manner
20:03 dhblaz so if you change any stats line you have in /etc/haproxy/haproxy.cfg to   stats  socket /var/lib/haproxy/stats level admin
20:03 dhblaz and restart haproxy
20:03 xarses you need the patch to throw away the db pool after the first connection error, and to set the timeout value to 60 instead of the default which i think is 900
20:03 dhblaz you can disable mysql nodes like this:
20:03 dhblaz echo "disable server mysqld/node-18" | socat unix-connect:/var/lib/haproxy/stats stdio
20:04 xarses right now you have connection_timeout * connection pool size (usually 10) seconds before the connections will reset when the active host has moved
20:04 wayneeseguin Is a 6GB RAM VM sufficient to sustain Fuel in production?
20:05 wayneeseguin (My understanding is that once the system is online Fuel's involvement is small and sits there quietly)
20:11 wrale_ So I thought I had made an iso of 5.0 stable, but if I followed these instructions, I built just the main 5.0 tag.  I see that 5.0 stable is in its own branch.. what should i do with the config.mk to make a healtier iso (like 5.0 stable)?
20:12 wrale_ http://docs.mirantis.com/fuel-dev/develop/env.html#building-the-fuel-iso
20:17 wrale_ nvm.. the trunk version has what i need.
20:56 xarses wrale_: tag 5.0 will be the 5.0 release
20:56 xarses stabe/5.0 will be the current (likely unreleased) 5.0.x version and could possibly have issues
20:57 xarses master (trunk) is currently pre 5.1 and is likely to have issues
20:58 wrale_ i'm hoping to see if i can apply the patch from https://review.openstack.org/102267 to my running stable/5.0 system.. reading the wiki
21:03 jeremydei joined #fuel
21:08 wrale_ patch command is not installed by default.. bogus :)
21:16 dhblaz xarses: FYI, I’m pretty sure this log message indicates that cinder was using mysql directly instead of through haproxy. node-16 cinder-volume 2014-06-26 21:15:00 WARNING cinder.openstack.common.db.sqlalchemy.session [-]  Got mysql server has gone away: (2006, 'MySQL server has gone away')
21:17 xarses no, thats exactly what I was talking about. Fuel configures you to use the vip
21:17 xarses the problem is that HA proxy is really bad about telling you that the node moved
21:17 xarses so you have to wait for various timout values to kick in
21:18 dhblaz I had already disabled haproxy use of mysql for node-16 (where I am rebuilding the db) nearly an hour ago.
21:19 wrale_ well the patch doesn't seem to match my running system.. i found two mailgun directories inside different docker volumes... trying to edit the patch to force it in..
21:20 dhblaz Perhaps my disablement of haproxy didn’t work...
21:22 wrale_ and my 5.0.1 build failed with a yum error 1.. after "No Match"
21:23 xarses dhblaz: https://review.openstack.org/#/c/77657/ (1 of 2)
21:24 dhblaz xarses: thanks
21:25 xarses dhblaz: https://bugs.launchpad.net/fuel/+bug/1288438 https://review.openstack.org/#/c/81225/
21:26 xarses (2 of 2)
21:26 dhblaz when grastate.dat has uuid: all zeros do I need to rm it to get the node to sync?
21:26 xarses all components will need that db connection string changed, and the oslo code (or equivalent) patched
21:27 xarses dhblaz: not sure, never had to muck with it
21:27 xarses dhblaz: once you get rid of the mysql lockup, you will find that rabbit is locking up
21:28 dhblaz you sent me before the steps to recover a failed node.  I can’t find it do you have it readily available?  If not I’ll keep looking in my transcripts
21:29 xarses dhblaz: here is the catalog of everything we've changed related to ha resiliency recently https://lists.launchpad.net/fuel-dev/msg01057.html
21:29 xarses I probably gave you https://gist.github.com/xarses/2703f00cf09805e918bc
21:33 wrale_ is there any site that builds the iso's each day?
21:33 wrale_ a jenkins?
21:37 dhblaz xarses: thanks, that was it
21:39 xarses wrale_: there isn't one available yet, but it should be coming soon
21:39 wrale_ xarses: thank you
21:42 Kupo24z xarses: is it ill-advised to add the public ubuntu mirrors to the apt sources list and update the system? will it have package conflicts with fuel's versions?
21:43 xarses Kupo24z: it could
22:20 wrale_ i dropped the trunk version of nailgun in at the site-packages level.. rebooted... now my first env is unfrozen.. and fuel nodes --env 1 works.... trying deploy soon
22:23 wrale_ is an outbound nic to public on the fuel node required?  i'd like to disable mine in /etc/sysconfig/network-scripts to hide my fuel web from the world.. i looked at iptables to block inbound everything from public, but looks dangerous to edit
22:23 xarses you don't need public access to fuel node
22:24 xarses only that the the nic you are using from PXE is accessable to all the nodes
22:25 wrale_ cool.. and if my upstream dns is on the ipmi network, and the fuel node has access to it, is that okay?  or do the compute nodes and all need access to the upstream dns?
22:25 wrale_ thanks
22:26 wrale_ i was thinking fuel node would be the upstream to any other node
22:26 wrale_ henceforth
23:13 GeertJohan joined #fuel
23:13 GeertJohan joined #fuel
23:24 jonathan4 ok, i'm going to sign up and open a support ticket.. nodes get all the way done before timing out and failing on deployment with a python error...
23:24 jonathan4 is 5.0 just broken?
23:26 wrale_ i'm thinking i might try a late 4.x version..
23:26 jonathan4 same problem?
23:26 wrale_ similar... plus iso problems
23:26 jonathan4 how do i download 4.x?
23:26 wrale_ dunno.. just a thought
23:27 jonathan4 are you getting the "Deployment has failed.  Timeout of deployment exceeded." message too?
23:27 mattgriffin joined #fuel
23:28 wrale_ my problems are all over the place.. some of them are because i built my own iso based on the moving target known as stable/5.0
23:28 wrale_ it built the other day.. now i can't get it to even build
23:29 wrale_ i bet 5.0.1 will be awesome when they get it done in july (crosses fingers)
23:29 wrale_ but i can't wait, literally
23:29 wrale_ trying to build 5.0 now .. long shot
23:30 wrale_ i'll try building the master branch if that doesn't work
23:32 wrale_ 5.0 build failed
23:34 wrale_ almost 70 nodes to go into this..

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