Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel-dev, 2014-07-09

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

All times shown according to UTC.

Time Nick Message
00:17 IlyaE joined #fuel-dev
01:22 eshumakher joined #fuel-dev
01:25 xarses joined #fuel-dev
01:59 xarses joined #fuel-dev
02:26 xarses joined #fuel-dev
02:37 IlyaE joined #fuel-dev
03:06 IlyaE joined #fuel-dev
03:40 IlyaE joined #fuel-dev
04:12 ArminderS joined #fuel-dev
04:14 ArminderS left #fuel-dev
04:39 Arminder left #fuel-dev
05:03 IlyaE joined #fuel-dev
05:27 dshulyak_ joined #fuel-dev
05:29 ArminderS- joined #fuel-dev
05:29 xarses joined #fuel-dev
05:37 ArminderS joined #fuel-dev
05:41 ArminderS joined #fuel-dev
05:42 ArminderS joined #fuel-dev
05:44 ArminderS joined #fuel-dev
05:45 ArminderS joined #fuel-dev
05:47 ArminderS joined #fuel-dev
05:48 ArminderS joined #fuel-dev
05:49 ArminderS joined #fuel-dev
05:51 ArminderS joined #fuel-dev
05:52 ArminderS joined #fuel-dev
05:53 ArminderS joined #fuel-dev
05:55 ArminderS joined #fuel-dev
05:58 IlyaE joined #fuel-dev
06:23 ArminderS joined #fuel-dev
06:24 ArminderS joined #fuel-dev
06:25 artem_panchenko1 left #fuel-dev
06:25 ArminderS joined #fuel-dev
06:26 ArminderS joined #fuel-dev
06:27 ArminderS joined #fuel-dev
06:28 ArminderS joined #fuel-dev
06:29 ArminderS joined #fuel-dev
06:29 sambork joined #fuel-dev
06:30 ArminderS joined #fuel-dev
06:30 al_ex8 joined #fuel-dev
06:31 e0ne joined #fuel-dev
06:36 Longgeek joined #fuel-dev
06:36 ArminderS joined #fuel-dev
06:36 e0ne joined #fuel-dev
06:46 e0ne joined #fuel-dev
06:50 xarses guys, fuellib CI snapshots are still empty, please fix https://bugs.launchpad.net/fuel/+bug/1338686
06:52 e0ne joined #fuel-dev
07:06 IlyaE joined #fuel-dev
07:08 skolekonov joined #fuel-dev
07:15 artem_panchenko joined #fuel-dev
07:21 pasquier-s joined #fuel-dev
07:33 e0ne joined #fuel-dev
07:35 xarses aglarendil: please review https://review.openstack.org/#/c/103280/ its started to pass CI
07:35 xarses holser: ^
07:37 xarses alex_didenko: ^^
07:40 ArminderS joined #fuel-dev
08:03 ArminderS joined #fuel-dev
08:09 amaretskiy joined #fuel-dev
08:09 holser xarses: I am looking
08:14 guillaume__1 joined #fuel-dev
08:17 aglarendil xarses: we'll look into it. at first glance you will get objections on composition model.
08:21 izinovik hi
08:21 izinovik could you review and merge this one: https://review.openstack.org/#/c/104130/
08:22 zynzel hello guys
08:22 zynzel anybody knows method how i can regenerate astute.yaml?
08:22 zynzel i want to add new field, and i want to test this without redeploy
08:22 zynzel fuel 4.1
08:36 ArminderS joined #fuel-dev
08:43 AndreyDanin joined #fuel-dev
08:49 salmon_ joined #fuel-dev
08:52 brain461 joined #fuel-dev
09:06 AndreyDanin joined #fuel-dev
09:10 e0ne joined #fuel-dev
09:14 igajsin Hello. What about me https://review.openstack.org/#/c/104942/ ?
09:14 igajsin I have reacted to all comment and repaired it.
09:18 ArminderS joined #fuel-dev
09:23 e0ne joined #fuel-dev
09:46 vkramskikh hi salmon_ sambork
09:46 vkramskikh what api do you want me to use to update password?
09:46 vkramskikh https://github.com/openstack/python-keystoneclient/blob/master/keystoneclient/v2_0/users.py#L61-L77
09:49 salmon_ vkramskikh: update_own_password
09:49 vkramskikh ok thanks
09:51 ArminderS- joined #fuel-dev
09:58 ddmitriev joined #fuel-dev
09:58 Longgeek joined #fuel-dev
10:00 ARaza joined #fuel-dev
10:09 odyssey4me joined #fuel-dev
10:11 vkramskikh salmon_: so as far as i understand, all we need to do from UI is to update the password in keystone
10:11 vkramskikh and nothing else should be modified
10:11 vkramskikh am i rite?
10:15 brain461 joined #fuel-dev
10:15 ArminderS joined #fuel-dev
10:17 ArminderS- joined #fuel-dev
10:25 albs joined #fuel-dev
10:28 salmon_ vkramskikh: yup
10:28 vkramskikh nice
10:34 artem_panchenko joined #fuel-dev
10:37 e0ne joined #fuel-dev
10:39 msemenov_ joined #fuel-dev
10:50 vkramskikh salmon_: https://review.openstack.org/#/c/105683/ here we go
10:50 vkramskikh works with keystone from ISO
10:50 salmon_ vkramskikh: you are amazing :)
10:51 salmon_ ot you are a robot, I'm not sure ;)
10:52 guillaume__1 joined #fuel-dev
10:57 tzn joined #fuel-dev
11:16 pasquier-s joined #fuel-dev
11:20 ikalnitsky mattymo: hi. are you here? :)
11:23 apodrepniy joined #fuel-dev
11:25 mattymo ikalnitsky, I am now
11:28 ikalnitsky mattymo: currently in master we have rsync container with /puppet resource and we have links from /etc/puppet/* to /puppet/*. that's bad. we need to mount /puppet directly into /etc/puppet, like we did it in nginx container.
11:29 ikalnitsky mattymo: do you know anything about this? maybe the fix is already exist?
11:33 teran joined #fuel-dev
11:33 mattymo ikalnitsky, why do you need this so desperately?
11:34 mattymo it means dropping the middle containers
11:34 mattymo what exactly is breaking because of this?
11:34 igajsin I have cleared away unnecessary changes in ::cinder class and  superfluous file. Please, review https://review.openstack.org/#/c/104942/  now.
11:35 igajsin alex_didenko: ^^
11:36 igajsin mattymo ^
11:38 Longgeek joined #fuel-dev
11:39 ikalnitsky mattymo: we're installing new puppets in /etc/puppet/<fuel-version> (host-system). currently new puppets are in /puppet, but not in /etc/puppet (rsync container). so we can't deliver new puppets to slave nodes and openstack patching is broken now (with new docker images - after upgrading - it works, btw).
11:40 alex_didenko igajsin: I'll review it shortly
11:41 mattymo ikalnitsky, why do you need to log into the container to change puppet?
11:42 mattymo just simply replace /etc/puppet on master
11:43 ikalnitsky mattymo: we don't want replace puppets. we want to add new one for new openstack.
11:43 mattymo just replace it in /etc/puppet on fuel master. that's the original plan and why is it changing?
11:43 mattymo ignore the path used inside containers
11:44 mattymo (and this today only applies to rsync container)
11:47 ikalnitsky mattymo: one more time - I don't want to replace it. Why I should do that? We have different openstack releases which should be deployed by different puppets (because it may be important). So I definitely should install it in /etc/puppet/<fuel-version> and expect that rsync container will pick that changes.
11:49 mattymo yes it will get the changes
11:49 ikalnitsky mattymo: but it don't
11:49 mattymo rsync only
11:50 mattymo the other containers don't see anything from the host in /etc/puppet
11:50 mattymo show me an env and we can debug it
11:58 albs joined #fuel-dev
11:58 apodrepniy_ joined #fuel-dev
12:23 sambork joined #fuel-dev
12:25 salmon_ joined #fuel-dev
12:26 izinovik https://review.openstack.org/#/c/104130/10
12:26 izinovik anyone?
12:28 e0ne_ joined #fuel-dev
12:48 e0ne joined #fuel-dev
12:57 pasquier-s joined #fuel-dev
13:01 tzn joined #fuel-dev
13:09 pasquier-s joined #fuel-dev
13:20 prmtl joined #fuel-dev
13:22 prmtl_ joined #fuel-dev
13:31 mihgen aglarendil: alex_didenko mattymo As I understand, there is a suggestion to fix an issue on puppet side: https://bugs.launchpad.net/fuel/+bug/1319451
13:31 mihgen should we try it?
13:34 mattymo mihgen, using puppet to delete and recreate queues?
13:34 mattymo it's kind of silly really. we could create all the queues in puppet like we do for mcollective. that's not hard. that still won't fix the complaints from nailgun/astute which want to configure different queue options
13:34 mattymo if we set it the way astute wants it, nailgun will complain
13:35 Longgeek joined #fuel-dev
13:35 mattymo honestly to stay closer to openstack style, we should let nailgun and astute both create the queues and set params
14:12 ilbot3 joined #fuel-dev
14:12 Topic for #fuel-dev is now #fuel-dev Fuel dev docs http://docs.mirantis.com/fuel-dev/ | Paste here http://paste.openstack.org/ | IRC logs http://irclog.perlgeek.de/fuel-dev/ | gerrit traffic @ #fuel-tracker
14:13 xarses we cant keep using the broken iso, we need to upgrade the iso, or apply just the patch to fix snapshots
14:18 dpyzhov joined #fuel-dev
14:19 dmitryme hey people, could you please review a couple CRs for fuel-main and fuel-library for 5.1: https://review.openstack.org/#/c/102253/ and https://review.openstack.org/#/c/103061/
14:20 dmitryme we have these merged into 5.0.1
14:20 dmitryme now need them in 5.1
14:20 dmitryme Ceilometer does not work without them
14:21 jobewan joined #fuel-dev
14:26 mihgen dilyin: did you take a look at https://bugs.launchpad.net/fuel/+bug/1339593 ? this is high priority thing as it's 5.0.1
14:26 mihgen xarses: hi dude
14:26 mihgen xarses: as I understand mattymo was fixing it
14:27 jaypipes joined #fuel-dev
14:27 mattymo mihgen, xarses, which specific issue?
14:27 mihgen mattymo: about "fuellib systests still have empty snapshorts"
14:28 mihgen dmitryme: was it merged into 5.0.1 without merging into master first? doesn't sound very cool...
14:28 mattymo systests are fixed according to xarses's email
14:34 al_ex9 joined #fuel-dev
14:38 mihgen aglarendil: hey dude you sure your -2 is still relevant at https://review.openstack.org/#/c/104197/ ?
14:38 al_ex9 joined #fuel-dev
14:38 Chlorum joined #fuel-dev
14:48 xarses joined #fuel-dev
14:52 al_ex9 joined #fuel-dev
14:54 dpyzhov joined #fuel-dev
15:01 tzn joined #fuel-dev
15:02 IlyaE joined #fuel-dev
15:04 al_ex10 joined #fuel-dev
15:09 AndreyDanin joined #fuel-dev
15:23 dmitryme mihgen: agree, but we actually had a reason for that. Merge into 5.1 was blocked because packages for 5.1 were not ready at the moment. So fuel-ci always set -1 to the CRs
15:25 mihgen dmitryme: ok… I hope we will fix it next time. I guess the reason was that we didn't have branches in gerrit for specs in 5.1?
15:25 xarses bookwar: test_pullrequest just runs for 0.00 sec and returns OK
15:25 dmitryme mihgen: yep, exactly
15:25 mihgen mattymo: aglarendil dilyin , other folks, let's not forget about https://review.openstack.org/#/c/102253/ and https://review.openstack.org/#/c/103061/
15:26 xarses I see that it tries to skip if OPENSTACK_RELEASE = Centos, so I canged it to Ubuntu, and it does the same still
15:26 mihgen especially considering it is already in 5.0.1
15:27 ikalnitsky joined #fuel-dev
15:37 Longgeek joined #fuel-dev
15:39 ArminderS joined #fuel-dev
15:40 Longgeek joined #fuel-dev
15:40 xarses aglarendil: if its an issue, make comment and maybe propose your way to implement
15:41 aglarendil xarses: what are you talking about?
15:43 aglarendil guys, is it ok if we make a hangout call at 8PM MSK/9AM PDT to discuss xarses neutron module merge ?
15:43 xarses <aglarendil> xarses: we'll look into it. at first glance you will get objections on composition model.
15:43 xarses aglarendil: What we need
15:43 mihgen aglarendil: I can't participate at 8pm, I can do it now though..  you can do it without me of course
15:43 xarses aglarendil: what ever we need =)
15:44 mihgen xarses: we need to discuss further steps with ML2 and how it affects Mellanox and possibly other folks who will need to do rebases
15:45 aglarendil is dmitry angdraug here ?/
15:50 xarses aglarendil: he is on the CI meeting but apparently not on IRC
15:51 xarses I poked him on Skype
15:52 xarses mihgen: yes we do
15:53 xarses mihgen: for Mlnx, I've looked over the code, they don't appear to touch the neutron config files that puppet-neutron uses, so there may be little to no issue there.
15:53 xarses I will need to re-review
15:53 xarses NSX also
15:53 mihgen xarses: good news then
15:54 mihgen xarses: as I won't be avlbl at 8, so two questions here
15:54 mihgen 1) did you test HA ?
15:54 prmtl joined #fuel-dev
15:54 mihgen 2) did you test both GRE & openvswitch under ML2?
15:54 bookwar xarses: https://github.com/stackforge/fuel-main/blob/master/fuelweb_test/tests/test_pullrequest.py#L26 should work
15:55 bookwar xarses: maybe pythonpath is no right
15:55 xarses Works with 1 controller many multiple times, retesting with more since ubuntu appears to be HA for systest
15:55 xarses 2) you mean GRE and VLAN, yes works for me in centos and ubuntu
15:55 xarses both under ml2
15:56 xarses ML2 is nice, we will need to re-decide some net arch, we can have all endpoint types in a single cluster
15:56 aglarendil xarses: the real problem that guys who are working
15:57 aglarendil xarses: on Mlnx and NSX are not so sophisticated in FUEL as we are at the moment
15:57 aglarendil xarses: it will be much more difficult for them to even do a simple rebase
15:57 aglarendil xarses: this can put or already delayed FF under the risk
15:58 aglarendil xarses: I looked over your code. It looks clean and good
15:58 aglarendil xarses: but I am afraid of bugs that could be brought in
15:58 aglarendil xarses: because you removed parameter validation made
15:59 aglarendil xarses: by sanitize_quantum_config function
15:59 aglarendil xarses: and this can lead to lots of bugs and unstable deployment
15:59 aglarendil xarses: while we are hunting down all the forgotten parameters
16:00 xarses Oh, I can share the map off where all the parameters went
16:00 aglarendil xarses: also you modified code for nova-network
16:00 aglarendil xarses: did you test this modification for regression ?
16:02 xarses I modified the entry point of nova network in openstack:: you mean?
16:02 xarses I tested it before the nova rebase, I will make a point to retest it again
16:04 aglarendil xarses: regarding parameters - I meant that some of the parameters
16:04 aglarendil xarses: could be filled by sanitize_* functions
16:04 aviramb joined #fuel-dev
16:04 aglarendil xarses: as defaults while not being sent by nailgun
16:04 aglarendil xarses: and now we can have these options missing
16:05 ArminderS left #fuel-dev
16:05 aglarendil xarses: and this can lead to hardly catchable bugs
16:05 xarses If we cant rely on nailgun sending everything we need, then we need to make serious attention there
16:06 xarses it would break everything, not just neutron
16:06 aglarendil I mean, that we parameter validation should not be missing between modules of the system
16:06 aglarendil xarses: ideally, we should validate all the parameters in all the methods and all the classes
16:06 aglarendil xarses: it is not feasible, but we should keep validation at least between big blocks of the system
16:07 aglarendil xarses: this is what was intentionally in our plans - write puppet function to validate
16:07 aglarendil xarses: all the parameters that come from nailgun
16:07 aglarendil xarses: thus allowing deployment to fail, e.g. if we have string in the field that should contain IP address
16:07 xarses I don't understand this point well, if we move to hiera the data would be validated way before puppet runs
16:08 aglarendil xarses: I am speaking of a little bit more complex validation
16:08 aglarendil xarses: and you made a step back abandoning validation for one of our most complex modules
16:08 aglarendil xarses: what I am speaking of is the real data schema with all the datatypes specified
16:09 aglarendil xarses: and validation of parameters along with validation of combinations of parameters
16:10 xarses yes, which should be done by the generation of astute.yaml. or as validation of all of astute.yaml
16:10 aglarendil xarses:  you are talking about nailgun and orchestration part
16:11 xarses correct
16:11 aglarendil xarses: I am talking about puppet-side validation
16:11 aglarendil xarses: ideally, we should not trust external modules such as nailgun
16:11 aglarendil xarses: we should always recheck the parameters that are sent by them
16:14 aglarendil xarses: my gut feeling says that it is very risky to accept your code as
16:14 aglarendil xarses: it is much more invasive and may introduce regressions
16:14 aglarendil xarses: blocking other teams
16:18 aglarendil xarses: and it is less tested now - xenolog's code was tested with all the modes
16:18 aglarendil xarses: also allowing to turn on older OVS behavious in case user wants this
16:19 AndreyDanin joined #fuel-dev
16:19 xarses aglarendil: that should work for mine as well
16:19 xarses xenologs still isn't passing CI
16:19 aglarendil xarses: this is because CI is broken
16:20 aglarendil xarses: it sporadically fails due to https://bugs.launchpad.net/fuel/+bug/1338594
16:21 aglarendil xarses: we are waiting for newer ISO to pass tests and be deployed to our CI
16:24 dshulyak_ joined #fuel-dev
16:25 bookwar aglarendil: https://fuel-jenkins.mirantis.com/job/master_fuellib_review_systest_ubuntu/1575/console
16:28 xarses aglarendil: I don't think validating the parameters again is worthwhile, however I'm not goin to argue that right now. Sanitize is the wrong way to do it. We can't just be magicly appearing new settings and setting the elements back into the upstream modules avoiding the parmater entry points, its not readable. It's not using the same parameter names, its not using a similar data structure as the modules consume. It needs all kinds of work. If removing
16:28 xarses sanatize was a blocker, then it should have been raised a long time ago as such.
16:39 aglarendil xarses: AFAIK, this is what you disagreed on with xenolog. My suggestion is that we review this code again and merge it into 6.0 release cleanly and with no bugs.
16:41 aglarendil xarses: alternative risk is to merge it and get broken master for a while, blocking other teams and jeopardizing feature freeze
16:42 aglarendil xarses: if we had already working code for Mlnx and NSX rebased on your code than it would be another pair of shoes
16:45 aglarendil xarses: I am strongly for merge of your code, but not the day before the FF
16:45 xarses Umm thats kind of the point of feature freeze
16:47 aglarendil xarses: I mean, that it will require additional work from Mlnx and VMware teams and they will miss FF
16:48 xarses AFAICT Mlnx does nothing with the neutron module
16:49 xarses which might be a problem for them, I don't even see them updating the mechanisms setting
16:49 aglarendil xarses: Mlnx uses ML2
16:49 aglarendil xarses: they are waiting for ML2 to introduce their code, I guess
16:49 aglarendil xarses: also the bigger problem is with NSX
16:58 xarses aglarendil: yep, It looks like it uses legacy openvswitch
17:06 evgeniyl mattymo: guys from docker have a page https://www.docker.com/resources/usecases/ maybe we can describe our use cases, what do you think? :)
17:07 guillaume__1 left #fuel-dev
17:10 e0ne joined #fuel-dev
17:16 xarses aglarendil: I think that NSX might have problems with either patch, it looks like it intends to use legacy OVS, but dosn't explicitly set it, so when either ML2 is submitted, it will be broken
17:17 angdraug joined #fuel-dev
17:27 dshulyak_ joined #fuel-dev
17:28 pasquier-s joined #fuel-dev
17:34 aglarendil xarses: and that's why xenolog's patch is better as it allows user to use legacy OVS
17:34 e0ne joined #fuel-dev
17:35 xarses aglarendil: no, NSX will fail anyway because it uses generated value from sanatize
17:35 xarses when ml2 becomes default it will break
17:37 aglarendil xenolog's patch allows you to specify legacy OVS
17:37 xarses yes, but the NSX patch just uses the current default
17:38 xarses https://review.openstack.org/#/c/86329/15/deployment/puppet/neutron/manifests/init.pp
17:40 brain461 joined #fuel-dev
17:53 sanek joined #fuel-dev
17:53 artem_panchenko joined #fuel-dev
17:57 IlyaE joined #fuel-dev
18:30 IlyaE joined #fuel-dev
18:42 brain461 joined #fuel-dev
18:47 brain461 joined #fuel-dev
19:08 brain4611 joined #fuel-dev
19:15 e0ne joined #fuel-dev
19:17 tzn joined #fuel-dev
19:19 e0ne joined #fuel-dev
19:26 teran joined #fuel-dev
19:33 eshumakher joined #fuel-dev
19:34 IlyaE joined #fuel-dev
19:45 xarses angdraug: https://review.openstack.org/#/c/105791
19:46 e0ne joined #fuel-dev
20:05 enikanorov__ joined #fuel-dev
20:06 msemenov__ joined #fuel-dev
20:06 IlyaE joined #fuel-dev
20:07 mirrorbox joined #fuel-dev
20:10 msemenov_ joined #fuel-dev
20:15 dpyzhov joined #fuel-dev
20:25 e0ne joined #fuel-dev
20:29 e0ne joined #fuel-dev
20:30 IlyaE joined #fuel-dev
20:46 e0ne joined #fuel-dev
20:57 dpyzhov joined #fuel-dev
21:00 e0ne joined #fuel-dev
21:02 IlyaE joined #fuel-dev
21:52 IlyaE joined #fuel-dev
22:02 jay-house-hunter joined #fuel-dev
22:12 IlyaE joined #fuel-dev
22:16 bookwar stable/5.0 fuel-ci seems to be fixed now, empty tests pass on iso 111
22:21 angdraug bookwar: 5.1 seems still broken
22:22 bookwar for master i am waiting for this test https://fuel-jenkins.mirantis.com/job/master_fuellib_review_systest_ubuntu/1594/console to finish
22:23 bookwar it is run on 307 iso, with fixed galera package and rsyslog fix
22:24 aglarendil ubuntu bvt_2 failed for some weird amqp timeout
22:24 aglarendil there is bug about contents of snapshot
22:24 aglarendil https://bugs.launchpad.net/fuel/+bug/1339914
22:24 aglarendil guys, please look into it
22:25 aglarendil also, please look into ubuntu bvt_2 environment. may be you will find why comput nodes get this error
22:25 aglarendil though controller nodes easily connect to AMQP server
22:25 xarses aglarendil: isnt 1339914 the syslog issue mattymo was talking about?
22:25 aglarendil may be
22:26 aglarendil but our QA team committed a workaround for this into master before the tests were run
22:26 aglarendil the problem was with additional syslog server added to config
22:26 aglarendil this server was not accessible
22:27 aglarendil and rsyslog hung forever trying to send messages to this server
22:27 aglarendil QA team blanked the secondary syslog server field
22:27 aglarendil I am not sure if it is blank for bvt tests though
22:28 xarses bookwar: I see non zero byte snapshots, thanks
22:28 xarses very helpful
22:28 aglarendil but they do not contain /var/log/remote, though :(
22:28 xarses it contains node/var/log though
22:29 xarses which is good enough for deployment errors
22:36 salmon_ joined #fuel-dev
22:37 IlyaE joined #fuel-dev
22:39 apodrepniy joined #fuel-dev
22:40 bookwar deploy has finished, OSTF now, we've almost there...
22:45 bookwar and passed
22:48 aglarendil i restarted ubuntu bvt_2
22:49 aglarendil anyway, this ISO is better
22:53 bookwar yeah, i'm switching ci test to 307 iso now
23:00 xarses looks like controllers without vip cant reach vip again
23:09 xarses bookwar: is there a reason that so many jobs are in queue, but alot of the build server slots appear to be idle?
23:11 bookwar xarses: when i need to regenerate the environmnt with new iso, i'll take server out of the pool and first run master node setup, then empty test, and then put it in the pool again
23:11 bookwar it takes ~ 40 minutes
23:12 bookwar 40 minutes for centos and 60-70 for ubuntu
23:13 bookwar and normal tests are waiting in line
23:13 bookwar thus it gets crowded
23:14 bookwar and the second issue is that, currently we can't run two ubuntu-master env on the same server
23:15 bookwar we can run ubuntu-master and ubuntu-5_0, or ubuntu-master and centos-master in parallel, but not two similar environments
23:16 aglarendil xarses: controllers are fine
23:16 aglarendil the problem is with computes
23:16 aglarendil they do not register their services
23:16 aglarendil either it is intermittent bug with rabbitmq
23:16 aglarendil or I do not know, what could be the reason
23:16 aglarendil also, please consider this fix
23:16 aglarendil https://review.openstack.org/#/c/105912/
23:17 aglarendil this fixes Centos HA
23:17 aglarendil as it magically stopped working
23:19 aglarendil yep. it is because we added mira1 suffix to package release version
23:20 aglarendil please, merge it ASAP
23:23 aglarendil also, this one. it would be great if you investigated it
23:23 aglarendil https://bugs.launchpad.net/fuel/+bug/1339928
23:37 xarses aglarendil: root@node-2:~# ping 192.168.0.2
23:37 xarses PING 192.168.0.2 (192.168.0.2) 56(84) bytes of data.
23:37 xarses From 192.168.0.4 icmp_seq=1 Destination Host Unreachable
23:37 aglarendil what's that ?
23:38 xarses hapr-m: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
23:38 xarses inet 192.168.0.2/24 scope global hapr-m
23:38 xarses which is on node-1
23:39 xarses angdraug: https://review.openstack.org/#/c/104633/
23:39 xarses node-1:~# ip netns exec haproxy ping 192.168.0.4
23:39 xarses PING 192.168.0.4 (192.168.0.4) 56(84) bytes of data.
23:39 xarses From 192.168.0.2 icmp_seq=1 Destination Host Unreachable
23:39 aglarendil hmm. is it ubuntu or centos
23:39 xarses ubuntu
23:40 aglarendil and 0.4 is which IP address ?
23:40 aglarendil is routing configured normally
23:40 aglarendil + is ip_forward turned on ?
23:40 xarses its node-2
23:41 aglarendil 2 reasons: proxy_arp is turned off
23:41 aglarendil ip_forward is turned off
23:41 aglarendil iptables is not allowing traffic between the nodes
23:41 xarses well its breaks ha
23:42 xarses node-1:~# ping node-2
23:42 xarses PING node-2 (192.168.0.4) 56(84) bytes of data.
23:42 xarses 64 bytes from node-2 (192.168.0.4): icmp_req=1 ttl=64 time=0.672 ms
23:42 xarses 64 bytes from node-2 (192.168.0.4): icmp_req=2 ttl=64 time=0.221 ms
23:42 xarses it works outside of the namespace
23:42 xarses its just through the vip again
23:43 aglarendil I do not face this issue on bvt_2 test env
23:49 aglarendil may be you have the issues with deployment and some resources did not configure appropriately
23:50 aglarendil the most probable reason for compute nodes not login into amqp cluster
23:50 aglarendil is rabbitmq cluster partitioning

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