Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel-dev, 2014-09-02

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

All times shown according to UTC.

Time Nick Message
04:22 prmtl joined #fuel-dev
04:26 prmtl joined #fuel-dev
05:00 prmtl joined #fuel-dev
05:35 HeOS joined #fuel-dev
05:58 adanin joined #fuel-dev
06:17 prmtl joined #fuel-dev
06:26 salmon_ joined #fuel-dev
06:36 dancn joined #fuel-dev
06:46 sambork joined #fuel-dev
06:53 tzn joined #fuel-dev
06:59 skolekonov joined #fuel-dev
07:01 pasquier-s joined #fuel-dev
07:20 Longgeek joined #fuel-dev
07:23 Longgeek joined #fuel-dev
07:25 dkaigarodsev joined #fuel-dev
07:32 HeOS joined #fuel-dev
08:13 ddmitriev joined #fuel-dev
08:26 vkozhukalov joined #fuel-dev
08:35 mihgen bogdando: aglarendil mattymo sbog Folks, should be easy to fix: https://bugs.launchpad.net/fuel/+bug/1364083
09:06 e0ne joined #fuel-dev
09:19 mattymo mihgen, oh wow what a simple bug
09:23 mihgen mattymo: sbog let's figure out if 5.0/5.0.1 were affected. If so, I'd vote to have to add this to 5.0/5.0.1 release notes as known issue and what can ppl do with it..
09:26 mattymo mihgen, workaround is reboot it)
09:27 mattymo if it freezes because of out of memory
09:27 mattymo sbog,  took the ticket
09:27 mihgen mattymo: user will see that node has gone offline?
09:29 mattymo yeah because it will freeze and not send beacons
09:29 mattymo so it will be a "down" node
09:29 mattymo err offline bootstrapped node
09:34 msemenov joined #fuel-dev
09:35 aglarendil evgeniyl_: dpyzhov ikalnitsky
09:35 aglarendil please look into https://bugs.launchpad.net/fuel/+bug/1364081
09:35 aglarendil it looks like we do not upload older manifests and repos in case of env rollback
09:35 aglarendil please look
09:36 aglarendil evgeniyl_: dpyzhov ikalnitsky ^^
09:37 bdudko left #fuel-dev
09:37 bdudko joined #fuel-dev
09:44 mihgen mattymo: salmon_ did you guys check if fuelmenu pass change for Fuel UI works?
09:44 mihgen it didn't for me
09:44 mihgen where should I take a look?
09:47 salmon_ mihgen: it only works if you change it during msater node instalation
09:47 mihgen salmon_: that's where I did it
09:48 salmon_ mihgen: hmm, I will check it
09:48 mihgen it didn't pass smooth, UX is still awful there, I had to wait 5-10min before it said that it can't ping google
09:48 mihgen then I had to click another Save& Continue, waited 10 more minutes
09:48 mihgen but looks like other changes were applied
09:49 mihgen salmon_: what config file it supposed to be stored? updated password I mean from fuelmenu, before puppet takes it?
09:49 salmon_ mihgen: /etc/astute.yaml
09:50 mihgen there is no such fiel
09:50 mihgen salmon_: or it's in container?
09:51 evgeniyl_ mihgen: /etc/fuel/astute.yaml
09:51 mattymo mihgen, 5-10 minutes? maybe you're running strange virtual hardware that breaks the python net lib for probing devices
09:51 mihgen I checked that, it had admin/admin defaults
09:51 mattymo that happened before
09:51 salmon_ mihgen: /etc/fuel/astute.yaml
09:51 mihgen mattymo: I run it on my strange mac, as you know ;)
09:51 mattymo yeah strange mac user
09:52 mattymo mihgen, fuelmenu gets no dev time unless it gets time in our feature list
09:52 mattymo still hasn't had any significant improvements since last year
09:53 mihgen mattymo: no improvements needed, just to fix some issues
10:09 adanin joined #fuel-dev
10:15 e0ne joined #fuel-dev
10:33 dkaigarodsev joined #fuel-dev
11:05 teran joined #fuel-dev
11:14 sambork joined #fuel-dev
11:32 vkramskikh hi guys
11:32 vkramskikh i see a thread in fuel-dev ml named Beta milestone of Fuel 5.1 now available!
11:32 vkramskikh xenolog wrote there: * Compute, cinder, ceph nodes has no external (public) network interface and don't get public IP addresses if it's not necessary.
11:33 mihgen vkramskikh: yep. it's both openstack-dev and fuel-dev
11:33 vkramskikh could anyone tell me how it is implemented? i thought there must be UI changes
11:33 vkramskikh so on interface configuration screen there should be a "placeholder" for unnecessary networks
11:34 vkramskikh so these networks could be drag&dropped there
11:34 vkramskikh but as i can see this was implemented in another way
11:34 mihgen vkramskikh: we did a quick workaround for 5.1. akasatkin is the best guy to explain I believe
11:34 vkramskikh could anyone tell me how?
11:35 mihgen I believe all done on nailgun & puppet side only, without UI changes
11:35 mihgen so that public is still on iface is just ignored
11:35 mihgen and IP is not allocated to node when node doesn't really need it
11:35 akasatkin vkramskikh: the user-visible changes are in fixture only
11:35 vkramskikh ok, got it
11:36 akasatkin vkramskikh: "Assign public networks to all nodes" parameter
11:36 vkramskikh we are going to reimplement this correctly, are we?
11:36 koder joined #fuel-dev
11:36 akasatkin vkramskikh: within adv-networking
11:37 koder в fuel(или у кого-нить) есть готовое API, которое по конфигу (yaml/json/etc) разворачивает кластер в один вызов?
11:37 akasatkin all networks will be able
11:40 koder does fuel has API, which allows deploy cluster from config file (yaml/json/etc) within a single call?
11:40 koder or two )
11:41 mihgen koder: not really. But there is a plan to: https://blueprints.launchpad.net/fuel/+spec/environment-reproduce-script
11:42 mihgen no sure though when someone get time to do it
11:43 koder > This script should upload yaml file to Nailgun and trigger deployment.  - where I can find example of such file?
11:47 dkaigarodsev joined #fuel-dev
11:48 akasatkin koder: you can create env, add nodes (don't start deployment) and execute "fuel —env=### deployment —default" . It will generate files for orchestrator.
11:48 akasatkin You can use them as examples.
11:50 koder thx
11:52 teran joined #fuel-dev
11:53 akasatkin koder: but cluster object with nodes added into it are to be in Nailgun DB to run deployment from Nailgun so you cannot omit steps: 1. create cluster, 2. add nodes
11:57 cylee_ joined #fuel-dev
12:00 akasatkin https://bugs.launchpad.net/fuel/+bug/1360230 - I have env with this
12:08 teran joined #fuel-dev
12:09 omelchek joined #fuel-dev
12:20 sambork joined #fuel-dev
12:28 vkozhukalov1 joined #fuel-dev
12:36 ai joined #fuel-dev
12:39 e0ne joined #fuel-dev
12:53 asheplyakov joined #fuel-dev
12:59 tuvenen joined #fuel-dev
13:02 manashkin_ joined #fuel-dev
13:04 teran_ joined #fuel-dev
13:10 vkozhukalov joined #fuel-dev
13:46 adanin joined #fuel-dev
13:46 tuvenen joined #fuel-dev
13:58 teran joined #fuel-dev
13:59 ikalnitsky joined #fuel-dev
13:59 jobewan joined #fuel-dev
14:06 jobewan joined #fuel-dev
14:07 angdraug joined #fuel-dev
14:43 kaliya joined #fuel-dev
15:00 kaliya joined #fuel-dev
15:22 teran joined #fuel-dev
15:28 vkozhukalov left #fuel-dev
15:45 angdraug christopheraedo: https://review.openstack.org/116422
15:45 angdraug please compare with the email discussion of the same, I think it's still getting it wrong
15:53 tzn joined #fuel-dev
16:00 albs joined #fuel-dev
16:05 rmoe joined #fuel-dev
16:12 artem_panchenko joined #fuel-dev
16:19 angdraug joined #fuel-dev
16:36 BillTheKat joined #fuel-dev
17:08 manashkin__ joined #fuel-dev
17:15 kaliya_ joined #fuel-dev
17:37 e0ne joined #fuel-dev
17:39 christopheraedo angdraug: will do
17:45 tzn joined #fuel-dev
17:47 teran joined #fuel-dev
17:47 xarses joined #fuel-dev
17:52 e0ne joined #fuel-dev
17:54 kaliya_ joined #fuel-dev
17:54 prmtl joined #fuel-dev
18:06 kaliya_ joined #fuel-dev
18:06 HeOS joined #fuel-dev
18:07 prmtl joined #fuel-dev
18:12 e0ne joined #fuel-dev
18:15 prmtl joined #fuel-dev
18:43 prmtl joined #fuel-dev
18:46 mihgen dpyzhov: why not to do while list from the very beginning?
18:46 mihgen should be same complexity
18:46 mihgen https://review.openstack.org/#/c/118377/1
18:46 dpyzhov mihgen: do you mean ‘white list'?
18:47 mihgen white of course, sorry..
18:47 dpyzhov right now we know what we don’t want to see
18:47 dpyzhov and we don’t have a white list
18:48 salmon_ joined #fuel-dev
18:48 dpyzhov mihgen: It will be nice if someone create whitelist
18:49 dpyzhov we can do it later
18:50 mihgen dpyzhov: how did you create black list? I assume there is same way for white one
18:51 mihgen actually while creating black you could miss some package which we don't really want to patch
18:51 dpyzhov mihgen: sure we miss some packages
18:52 dpyzhov right now we know, that presence of ceph in patching can produce pain
18:52 dpyzhov and we don’t want pain. and we don’t want to do major changes right before HCF
18:53 dpyzhov because we can’t be sure that whitelist will not break our upgrade
18:53 clauded joined #fuel-dev
18:53 * dpyzhov summons aglarendil
19:00 clauded Just upgraded from fuel 5.01 to 5.1 and launched "update environment". It succesfully ended but when I run apt-get update/apt-get upgrade on each node, many packages are not updated (nailgun-agent nailgun-mcagents nailgun-net-check python-ceilometerclient python-cinderclient python-glanceclient python-keystoneclient python-neutronclient python-novaclient python-oslo.messaging python-swiftclient python-troveclient). Should I manually run the upgra
19:02 aglarendil clauded: which environment did you have before the update ?
19:02 aglarendil clauded: I mean, version
19:02 clauded aglarendil: version 5.01
19:03 aglarendil clauded: and you started update to 5.0.2 ?
19:05 clauded aglarendil: well it's confusing. I installed the official fuel 5.0.1 then applied the 5.1 upgrade tar. Now in the Releases tab I have 3 versions (5.0.1, 5.0.2 and 5.1).
19:05 aglarendil clauded: so you updated your 5.0.1 env to 5.0.2 ?
19:06 aglarendil clauded: I mean openstack cluster
19:06 clauded aglarendil: I just clicked the update button in Actions tab.
19:07 aglarendil clauded: okay, which version of 5.1 do you have? python-oslo.messaging should have been updated
19:07 aglarendil clauded: as we update it explicitly using puppet.
19:07 aglarendil clauded: speaking of nailgun packages and openstack clients this looks like a bug.
19:08 clauded aglarendil: in the fuel dashboard, it's says Icehouse on Ubuntu 12.04.4 (2014.1.1-5.0.2)
19:08 aglarendil clauded: could you please file a bug with output of http://<master_node_ip>:8000/api/version ?
19:10 clauded aglarendil: sure but for my personal information, why is it upgraded to 5.0.2 and not 5.1?
19:11 aglarendil clauded: because 5.1 is the version of FUEL. it contains code for two new versions of OpenStack clusters: 2014.1.1-5.1 and 2014.1.1-5.0.2
19:11 aglarendil clauded: 5.0.2 contains the same code of OpenStack as 5.1 but other services are deployed differently
19:11 aglarendil clauded: e.g. we have different deployment scheme for RabbitMQ and different OCF script for Galera
19:12 aglarendil clauded: they are much better, but we cannot apply them to existing environment yet
19:12 aglarendil clauded: as it requires very complicated steps to be performed which we cannot do right now
19:12 clauded aglarendil: ok thanks. One more question : how can I test Zabbix (it's not in the gui)?
19:13 aglarendil clauded: well, you need to find IP address of your Zabbix node and go there by http
19:14 adanin_ joined #fuel-dev
19:14 mihgen aglarendil: nailgun-agent, mcagents and all these pkgs are installed during post-install of anaconda/preseed
19:15 mihgen we don't update them during patching, do we?
19:15 igormarnat joined #fuel-dev
19:15 clauded aglarendil: Zabbix node? I tought Zabbix was deployed on the fuel node
19:15 Ch00k joined #fuel-dev
19:16 mattymo joined #fuel-dev
19:16 mihgen clauded: it is not deployed on fuel master node, it is separate node
19:17 mihgen there should be notification like "your env successfully installed, go there and there"
19:17 mihgen is not zabbix server url there?
19:18 dteselkin joined #fuel-dev
19:19 clauded aglarendil: is there any instructions out there to setup Zabbix monitoring?
19:20 mihgen clauded: it's in progress: https://review.openstack.org/#/c/113773/
19:20 aglarendil clauded: well, FUEL configures Zabbix for you along with all openstack monitoring items
19:20 katyafervent joined #fuel-dev
19:23 mihgen clauded: there is separate role actually for zabbix. When you assign roles to nodes, you should see it
19:23 mihgen clauded: unless you are running official Mirantis OpenStack
19:23 mihgen the thing is that zabbix is considered experimental by mirantis
19:23 mihgen so it's not enabled by default
19:24 clauded mihgen: hum, it's the official Mirantis upgraded with the dev upgrade tar so I dont see it :(
19:24 mihgen yeah so it's not supported officially
19:24 mihgen but if you want to hack it, it is pretty simple
19:25 mihgen vkramskikh: did you put that missing piece into docs on how to enable experimental mode?
19:26 mihgen I see only http://docs.mirantis.com/fuel-dev/develop/nailgun/customization/settings.html?highlight=experimental
19:26 clauded mihgen: I'm just testing so getting the dev 5.1 iso will be simpler (albeit slower)
19:26 bookwar we have fuel-community builds with experimental features enabled
19:27 mihgen clauded: yeah you can get a build with enabled zabbix, sure. It is good that you tested patching though :)
19:30 christopheraedo mihgen: Documentation for enabling experimental features hasn't been merged yet, but is close: https://review.openstack.org/#/c/116131/
19:32 mihgen christopheraedo: it is too bad that it is in fuel-docs
19:32 mihgen I think it should be in dev docs
19:32 mihgen as it is about hacking, not about using
19:33 christopheraedo Right, but there was discussion on this - experimental features do not require you to build the ISO, and there is benefit for people who want to try them without getting into fuel development (for instance with zabbix), so I think makes a lot of sense to have in regular docs
19:35 christopheraedo But! Soon fuel-dev docs will be in fuel-docs, so it will all be in basically one place, much easier to find, and we can then change the organization a little for the things that are between basic user and most advanced dev.
19:35 mihgen christopheraedo: so why don't we simply put this seciton into dev docs, and make a reference from user docs?
19:35 mihgen repo doesn't matter
19:35 mihgen what matters is that if I'm a developer I can't find this basic important info
19:35 mihgen obviously I'm still and will be looking for it in dev docs, not user docs
19:36 mihgen I hope we do not plan to mess dev docs with user docs
19:36 christopheraedo I still think it belongs in user docs, but I agree having a mention of enabling experimental mode in the dev docs is needed
19:38 christopheraedo The dev docs will still stand on their own, but will have all the docs (user, operator, dev, etc.) in one repo, and available at one URL. Will look nicer, and also easily indexes ALL the docs, so one search covers everything
19:38 mihgen christopheraedo: ok I'm tired for now convincing the way around, so I'll be fine to have just a link from dev docs to user docs
19:38 mihgen though we should link to master version of user docs, as dev docs should always stay on master
19:38 christopheraedo :) OK with me, but happy to keep getting great feedback on the doc stuff for sure
19:48 teran joined #fuel-dev
19:49 teran_ joined #fuel-dev
19:49 mihgen christopheraedo: it was hard not to put -1 there... )
19:54 christopheraedo Haha
19:59 mihgen what does Fuel need to support iSCSI multipath?
20:02 mihgen xarses: any idea/
20:03 mihgen is not it backward compatible with single path, so if we do multipath - singlepath will continue to work if multi is not supported?
20:03 mihgen I see https://review.openstack.org/#/c/17946/15 in nova
20:04 xarses mihgen: ? about iSCSI?
20:04 mihgen yep
20:06 xarses looks like we need multipath binary, and some config in nova.conf.
20:06 xarses but are we looking to multipath a cinder-volume provider from lvm?
20:06 xarses or multipath an array?
20:07 mihgen could be both
20:07 mihgen actually the one can use boot from volume
20:07 xarses for the prior we'd need do some work to the cinder-volume server so that there are multiple nics providing the storage or their won't be any benefit
20:07 mihgen and volujmes served by cinder
20:08 mihgen xarses: could be benefit in reliability but I'm not sure
20:09 xarses we'd need to be able to have two nic-roles for storage (advanced-networking bp)
20:10 xarses for the latter case, it would just be some nova side changes and should be much easier to support
20:13 mihgen yep, I see, thanks
20:23 xarses left #fuel-dev
20:23 xarses joined #fuel-dev
20:24 xarses mihgen: as a note, no one in the US has oper with ChanServ on #fuel-dev, can you please have someone added?
20:25 xarses mihgen:
20:25 xarses oops =)
20:26 mihgen xarses: how can I do it?
20:29 xarses thanks
20:29 xarses lets see if it does it on the re-log
20:29 xarses joined #fuel-dev
20:30 clauded left #fuel-dev
20:44 xarses hmm
20:44 xarses =)
20:44 xarses thanks mike
20:45 mihgen xarses: np :) don't know if you can kick people out though
20:59 koder joined #fuel-dev
21:14 tzn joined #fuel-dev
21:51 xarses joined #fuel-dev
22:03 adanin joined #fuel-dev
22:28 tzn joined #fuel-dev
23:42 teran joined #fuel-dev

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