Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel-dev, 2016-01-27

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

All times shown according to UTC.

Time Nick Message
00:47 xarses joined #fuel-dev
00:56 zhangjn joined #fuel-dev
01:14 zhangjn joined #fuel-dev
01:34 Liuqing joined #fuel-dev
02:09 zhangjn joined #fuel-dev
02:24 dims joined #fuel-dev
02:28 xarses joined #fuel-dev
02:45 xarses joined #fuel-dev
02:48 e0ne joined #fuel-dev
05:06 xarses joined #fuel-dev
05:12 zhangjn joined #fuel-dev
05:13 javeriak joined #fuel-dev
05:43 xarses joined #fuel-dev
05:43 xarses joined #fuel-dev
06:24 fzhadaev1 joined #fuel-dev
06:42 dshulyak joined #fuel-dev
06:54 javeriak joined #fuel-dev
07:10 javeriak joined #fuel-dev
07:14 javeriak_ joined #fuel-dev
07:42 javeriak joined #fuel-dev
07:45 zhangjn joined #fuel-dev
07:47 xarses joined #fuel-dev
07:48 xarses joined #fuel-dev
07:49 xarses joined #fuel-dev
07:50 xarses joined #fuel-dev
07:51 xarses joined #fuel-dev
07:58 javeriak_ joined #fuel-dev
08:05 pbelamge joined #fuel-dev
08:29 zhangjn joined #fuel-dev
08:37 ekosareva_ joined #fuel-dev
08:49 jaypipes joined #fuel-dev
09:07 kertis joined #fuel-dev
09:17 amnk joined #fuel-dev
09:22 zhangjn joined #fuel-dev
09:24 IvanBerezovskiy joined #fuel-dev
09:27 tzn joined #fuel-dev
09:44 amnk joined #fuel-dev
09:58 rvyalov joined #fuel-dev
10:02 salmon_ joined #fuel-dev
10:09 amnk joined #fuel-dev
10:14 tzn joined #fuel-dev
10:25 tkhno- joined #fuel-dev
10:30 _nadya_ joined #fuel-dev
10:33 zhangjn joined #fuel-dev
11:08 zhangjn joined #fuel-dev
11:21 vkaplov left #fuel-dev
11:22 e0ne joined #fuel-dev
11:46 igajsin1 left #fuel-dev
12:20 amnk joined #fuel-dev
12:23 dims joined #fuel-dev
12:33 zhangjn joined #fuel-dev
12:35 zhangjn joined #fuel-dev
12:40 tzn joined #fuel-dev
12:46 obcecado joined #fuel-dev
12:49 ekosareva_ joined #fuel-dev
12:52 e0ne joined #fuel-dev
12:56 zhangjn joined #fuel-dev
13:30 _nadya_ joined #fuel-dev
13:58 ekosareva_ joined #fuel-dev
14:33 krobzaur joined #fuel-dev
14:33 DanSmithEric joined #fuel-dev
14:33 mkwiek hi guys, I want to mount additional folder to containers, is it enough to add a line here: https://github.com/openstack/fuel-library/blob/master/files/fuel-docker-utils/dockerctl_config#L82 , or is there any more magic to it?
14:34 mkwiek mattymo aglarendil bpiotrowski ^
14:35 DanSmithEric Hello out there Fuelers!   We are seeing an "weird" issue with FUEL8 that we didnt have in the versions prior (same hardware, same options in rollout, etc)..  When we use Reset or Delete environment, we are seeing an EXT4 kernal panic being thrown as the slave nodes are being ordered to reboot.  This kernel panic at first seemed related to having a delta in the plugins vs the actual repo, but now happens with almost every "delete
14:35 DanSmithEric Environment" command issued.
14:35 DanSmithEric anyone else there observe this behaviour (we are running on HP Gen460blc's)..
14:35 DanSmithEric Gen8-460blcs*
14:36 dims joined #fuel-dev
14:41 ron___ joined #fuel-dev
14:42 ron___ Hi
14:42 ron___ I am facing an issue in Fuel 7.0 deployment
14:42 ron___ wherein when I add a controller node to an already existing deployment; my deployment on compute node fails
14:43 ron___ https://usercontent.irccloud-cdn.com/file/hzS3vhAR/Error%20snapshot
14:44 ron___ Since I donot make use of ml2 plugin; i have overriden some Fuel manifests using Fuel hiera overide option on the compute node
14:45 ron___ in this fashion
14:45 ron___ openstack::network::agents: []
14:46 samuelBartel joined #fuel-dev
14:49 javeriak joined #fuel-dev
14:50 ron___ joined #fuel-dev
14:50 ron___ Hi
14:51 ron___ sorry..I got disconnected
14:52 ron___ as mentioned earlier, I am hitting an issue on Fuel 7.0 deployment
14:53 ron___ when i try to add a controller node to an already existing deployment
14:53 e0ne joined #fuel-dev
14:54 ron___ i hit this on the already deployed compute node
14:55 rvyalov joined #fuel-dev
14:57 ron___ i have pasted the error snapshot in the chat box above
14:57 aglarendil ron___: smakar ^^
15:01 smakar hi
15:01 smakar so accordinfg to picture the problem with package "openvswitch-switch" installation
15:01 smakar ron___: could you check manually whether all is ok with this package
15:02 ron___ so i purge out the default ovs and install my own as a part of the initial deployment; so this failure is expected..is there any way I can bypass this Fuel check for default ovs module being present?
15:03 ron___ so at the time of first deployment when i configure my compute node; i make sure i get rid of the default ovs and install my own..as a result of this when i re-run the deployment after adding a controller node; Fuel puppet manifests complain on this default ovs package being not present
15:04 ron___ so this error is expected but is there a way i can override this?
15:05 e0ne joined #fuel-dev
15:05 smakar ok I have got this /
15:05 rvyalov joined #fuel-dev
15:05 smakar let me a look into
15:06 ron___ many thnx
15:07 ron___ i feel the issue seems to be this Fuel manifest /etc/puppet/modules/l23network/manifests/l2.pp
15:07 ron___ wherein Fuel checks for these packages and services
15:11 rvyalov joined #fuel-dev
15:12 smakar there are two ways to fix it
15:12 smakar 1. use the same name for your custom package
15:12 smakar 2. pass your package name during l23network module initialization https://github.com/openstack/fuel-library/blob/stable/7.0/deployment/puppet/l23network/manifests/init.pp
15:12 smakar the parameter is *ovs_common_package_name*
15:12 smakar the place you have to put this is https://github.com/openstack/fuel-library/blob/stable/7.0/deployment/puppet/osnailyfacter/modular/netconfig/netconfig.pp#L5-L7
15:12 smakar
15:15 ron___ can you explain 2?
15:16 ron___ also what should be done in netconfig.pp?
15:16 ron___ didnot get that
15:19 ron___ i think i get your solution 2) but not the part about netconfig.pp
15:21 ron___ but i dont think 2) will fix the issue..it might lead to another issue in the first deployment itself..wherein if i pass my own package name while initializing the l23 module..then compute node deployment will fail..since all my puppet manifests run post Fuel base deployment
15:22 ron___ so if i pass my own package name during l23 mod init; when Fuel is in the base deployment it would not detect my package since my package gets installed after Fuel base deployment ends
15:23 smakar ron___: we initialize class l23network in  netconfig.pp
15:23 smakar you have to add parameter there
15:23 smakar the code should look like
15:23 smakar class { 'l23network' :
15:23 smakar use_ovs => hiera('use_neutron', false),
15:23 smakar ovs_common_package_name =>"your_package_name",
15:23 smakar }
15:24 ron___ but did you understand my concern on using this approach?
15:25 ron___ if i pass my own package name; the initial Fuel deployment would fail right?
15:25 ron___ since all my packages and configurations come as Fuel post-deployment stage in my plugin
15:28 smakar if you add your packages repo at the end of deployment - it will fail
15:28 smakar you have to add your repo before deployment and with higher priority then fuel repos
15:28 smakar If I am not mistaken we have such possibility
15:29 ron___ also, this is already defined as class { 'l23network' :
15:29 ron___ use_ovs => hiera('use_neutron', false)
15:29 ron___ } in netconfig.pp in the link you shared
15:29 rvyalov joined #fuel-dev
15:30 smakar yes it defined but youhave to add extra parameter
15:31 ron___ in the fuel manifest or as override?
15:31 ron___ i think i cannot touch Fuel code
15:32 ron___ also how will adding my packages repos before deployment help?
15:32 ron___ can you provide pointers on doing this?
15:33 smakar https://www.mirantis.com/blog/using-fuel-control-repositories-openstack-deployment/
15:35 smakar about not touching fuel code -   first option works here then
15:36 ron___ so did you suggest me to make the netconfig change in the Fuel node?
15:36 ron___ Fuel suggested us not to touch their base image for us to get ceritified for the plugin
15:36 ron___ and suggested us to execute all our tasks as either pre/post deployment
15:37 ron___ should this not be a Fuel deploy issue..why are even the compute node packages being verified when we are adding a controller node to the deployment?
15:38 ron___ the compute node is in ready and deployed state right?
15:38 xarses joined #fuel-dev
15:38 ron___ should Fuel not be leaving the deployed nodes as is
15:38 ron___ especially the compute node in this case when i add a controller node to an existing deployment
15:39 mwhahaha depends, we do have to do some updates on other nodes when nodes are added to a cluster
15:40 ron___ but how will adding our packages above Fuel repos help? for that we would need to rename our packages right?
15:41 ron___ to what Fuel expects
15:41 smakar yeah
15:41 ron___ that is not a feasible option for us
15:42 smakar why ?
15:43 ron___ To begin with; we were advised by Fuel at the very beginning that we would not be certified if we touch Fuel image; that is the reason we went this approach
15:43 mwhahaha you can provide the packages via a plugin as well
15:43 mwhahaha which doesn't require touching the fuel image
15:44 ron___ we are doing it that way
15:45 ron___ we are placing our packages on the Fuel node at /var/www/nailgun/plugins/<plugin name>/repositories/ubuntu/
15:45 ron___ when we deploy
15:45 mwhahaha what packages are you providing?
15:46 ron___ all controller and compute ndoe debians
15:46 ron___ node
15:46 mwhahaha what
15:46 ron___ so we have converted all our packages into debians..
15:47 ron___ since all nodes are Ubuntu
15:47 ron___ and we copy all these debains to the location i mentioned above
15:48 mwhahaha are you providing an openvswitch-common?
15:48 ron___ yes..we are
15:48 ron___ but the name is different
15:48 ron___ it is not "openvswitch-common"
15:49 ron___ i assume there would be other plugins which donot use the exact same name as Fuel expects to be present as manifests right?
15:49 mwhahaha so you need to be able to pass your package name into l23network::l2
15:50 mwhahaha and if that is currently not available, your options are 1) rename your package and provide openvswith-common or 2) disable the netconfig and implement it the way you need
15:50 mwhahaha 1 is the easiest
15:50 ron___ so option 2 requires changing Fuel?
15:50 mwhahaha no you could do 2 but you'd now be responsible for the entire network configuration process
15:53 ron___ when you said responsible for entire network config process it means changes in Fuel right?
15:53 ron___ or via hiera override?
15:53 mwhahaha no, provide a similar function via plugin
15:54 mwhahaha basically you'd need to disable the netconfig task and implement your own which is now getting way complex and i'm still not sure why you can't just rename your package?
15:54 mwhahaha is the whole point of your plugin to just use debian packages?
15:54 ron___ so only renaming the package names as Fuel expects and placing them at /var/www/...... will fix the issue?
15:55 ron___ is what you say
15:55 ron___ are there no other plugins that have hit thi issue?
15:56 mwhahaha not that i'm aware of
15:56 mwhahaha basically you're error is because you're not providing proper packages & their dependencies for what you're doing
15:56 ron___ looking at available upstream plugins; the way of configurong compute nodes looks similar to me
15:56 mwhahaha so yes, in theory providing a package with the same name fuel expect will satisify the error
15:57 ron___ and place the renamed packages at /var/www/... location right? no re-arranging Fuel repos and that stuff right?
15:57 ron___ only renaming packages to my current deployment style
15:57 mwhahaha it would be advantageous to file a bit requesting that these package names be overridable for future versions
15:58 javeriak joined #fuel-dev
15:58 mwhahaha the plugin framework provides a repository structure in the url that automatically gets added I believe, i don't think you need to manually place packages (but i'm not 100% on that)
15:58 mwhahaha er file a bug not file a bit
16:00 ron___ i donot have a local repository as such and so i have been copying debians manually to the /var/www/plugins/...../repositories/ubuntu/. as mentioned above
16:01 ron___ for deployment purposes
16:02 mwhahaha hmm ok
16:03 mwhahaha but anyway yes, if you are providing an alternate set of packages for fuel, currently for anything less than 9.0 you need to match whatever the naming conventions used by the MOS package set. For 9.0 we are working to add in the ability to leverage Ubuntu Cloud Archive based packages so that would be an alternative naming convention that could be used as
16:03 mwhahaha well
16:05 rmoe joined #fuel-dev
16:05 ron___ ok, Thanks
16:12 amnk joined #fuel-dev
16:18 ron___ btw..what is the l23 network module used for?
16:18 ron___ can you explain?
16:19 ron___ as far as the compute node deployment is concerned..
16:19 javeriak hi everyone, i hit this on my mos7 deploymet: https://bugs.launchpad.net/mos/+bug/1508489 and i used the workaround of commenting out the public endpoint and cli commans now go through; but python client calls are still failing with 'nova.flavors.fin fails Max retries exceeded with url: /v2/04df02729e4443e5a9a4f3db35f44f3c/flavors \(Caused by <class 'socket.gaierror'>: [Errno -2] Name or service not known) '
16:19 mwhahaha configures the network
16:19 mwhahaha all the interfaces and bridges and patches
16:20 IvanBerezovskiy left #fuel-dev
16:20 ron___ can you provide more detail?
16:20 ron___ as to what it exactly does on the compute node wrt to ovs?
16:21 mwhahaha https://github.com/xenolog/l23network
16:21 ron___ also how it this related to ovs config as such?
16:22 ron___ in simple terms
16:22 mwhahaha smakar or xenolog13 would have to provide those answers
16:23 yantarou joined #fuel-dev
16:25 amnk_ joined #fuel-dev
16:26 mwhahaha javeriak: that error points to a configured url for which it can't resolve, probably because it's pulling down the public url from the service endpoints and still tries to use it. You could try a host file entry for the public url
16:28 javeriak mwhahaha can't we force the pythonclient to use internal ip instead ?
16:29 mwhahaha i'm not sure, would have to dig into the code to see what it's actually doing
16:31 kertis joined #fuel-dev
16:33 javeriak mwhahaha i tried putting an enrty in the hosts file but then error changes to SSL ones _GET_SERVER_CERTIFICATE:certificate verify failed
16:34 mwhahaha oh you need to grab the CA bundle as well
16:34 javeriak the what?
16:37 mwhahaha the ssl CA certificates that are used for the self signed certs
16:37 javeriak mwhahaha this is unknown territory for me, could you elaborate a little please? what do i need to put where?
16:37 smakar ron___: l23network module configures basic network for openstack nodes
16:51 amnk joined #fuel-dev
17:05 mwhahaha javeriak: the tl;dr is that when you use a self signed cert, you have to add the certificate for the authority to any machine you wish to trust that cert for (to avoid such errors). I'm looking for where we keep our CA cert so you could grab it and add it to your machine to resolve such issue but i'm not having any luck at the moment.
17:05 mwhahaha the best practices is to not use the self signed certs for deployment for which you want external access via https.
17:08 bhaskarduvvuri joined #fuel-dev
17:09 javeriak mwhahaha oh okay; i guess it would just be easier if the keystone config would work as is
17:11 HeOS joined #fuel-dev
17:14 mwhahaha Yes that's a shortcoming with the default ssl config
17:17 aarefiev joined #fuel-dev
17:27 ogelbukh joined #fuel-dev
17:29 pbelamge joined #fuel-dev
17:47 _nadya_ joined #fuel-dev
17:55 ron___ Hi "mwhahaha", regarding the compute re-deployment issue we talked about some time back..would Fuel not use my hiera overriden "use_ovs" as False..if i set it via my plugin instead of me re-writing the netconfig/l23 module
17:56 ron___ currently i have surpassed Fuel's ml2 config checks with my hiera overide manifests like:
17:56 rvyalov joined #fuel-dev
17:56 ron___ openstack::network::agents: []
17:57 ron___ and calling this from deployment_tasks.yaml file as follows
17:57 ron___ - id: compute-hiera-override
17:57 ron___ type: puppet
17:57 ron___ groups: ['compute']
17:57 ron___ required_for: [globals]
17:57 ron___ requires: [hiera]
17:57 ron___ parameters:
17:57 ron___ puppet_manifest: puppet/manifests/compute-hiera-override.pp
17:57 ron___ puppet_modules: puppet/modules:/etc/puppet/modules
17:57 ron___ timeout: 720
17:58 ron___ this helps me surpass all ml2 related checks in Fuel
17:58 ron___ will Fuel not override the use_ovs from the netconfig with my hiera override parameter
17:59 ron___ i was referring to this
17:59 ron___ class { 'l23network' :
17:59 ron___ use_ovs => hiera('use_neutron', false)
17:59 ron___ }
18:00 dshulyak joined #fuel-dev
18:13 mwhahaha compute-hiera-override is for something else
18:13 mwhahaha you would need to put the use_neutron false in a place prior the the running of netconfig
18:14 mwhahaha is your plugin source available somewhere? as I don't have any idea what you're actually trying to accomplish it's hard to tell you where you could address the issues you might be running into
18:19 ron___ i can share the plugin rpm on dropbox and grant you acccess
18:20 ron___ if you can provide me your email id i can do that
18:21 tzn joined #fuel-dev
18:21 ron___ or i can email it to you
18:24 ron___ have emailed you
18:27 bhaskarduvvuri joined #fuel-dev
19:16 amnk joined #fuel-dev
19:18 ron___ Does Fuel 7.0 enable to disable Fuel netconfig and allow plugin developers to write or tweak their own?
19:18 ron___ confusing question:)
19:18 ron___ i meant..is there an option on Fuel 7.0 to disable Fuel netconfig
19:19 ron___ and let plugin inherit the Fuel netconfig.pp and tweak it a bit to fit needs
19:19 ron___ Fuel plugin
19:25 e0ne joined #fuel-dev
19:30 tkhno- joined #fuel-dev
19:53 dims joined #fuel-dev
19:55 _nadya_ joined #fuel-dev
19:55 javeriak joined #fuel-dev
20:06 amnk joined #fuel-dev
20:12 bhaskarduvvuri joined #fuel-dev
20:13 _bd_ joined #fuel-dev
20:21 salmon_ joined #fuel-dev
20:34 fzhadaev1 joined #fuel-dev
22:51 dims joined #fuel-dev
23:02 e0ne joined #fuel-dev
23:04 e0ne_ joined #fuel-dev
23:20 amnk joined #fuel-dev
23:22 amnk_ joined #fuel-dev
23:30 dims_ joined #fuel-dev

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