| *** Hefeweiz1n has quit IRC | 00:02 | |
| *** eghobo has quit IRC | 00:03 | |
| *** dwalleck has quit IRC | 00:08 | |
| NobodyCam | off to look for dinner | 00:11 |
|---|---|---|
| JayF | It seems like every so often, it just takes hours upon hours to get tests to run for things | 00:25 |
| JayF | is there some kind of problem? Not enough capacity? | 00:25 |
| JayF | We've had a patch set queued in jenkins for 90m or so now | 00:25 |
| *** matsuhashi has joined #openstack-ironic | 00:27 | |
| *** JayF has quit IRC | 00:40 | |
| *** JayF has joined #openstack-ironic | 00:41 | |
| NobodyCam | lol tests keep getting longer and longer :) | 00:43 |
| *** datajerk has joined #openstack-ironic | 00:49 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Add BackOffLoopingCall with jitter https://review.openstack.org/84303 | 00:58 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Add BackOffLoopingCall with jitter https://review.openstack.org/84303 | 00:59 |
| *** zigo has quit IRC | 00:59 | |
| *** zigo has joined #openstack-ironic | 01:01 | |
| devananda | JayF: http://status.openstack.org/zuul/ | 01:08 |
| *** nosnos has joined #openstack-ironic | 01:26 | |
| *** dwalleck has joined #openstack-ironic | 01:28 | |
| NobodyCam | devananda: do you know of a good how to use the .po files doc you can point me to? | 01:36 |
| devananda | nope | 01:36 |
| devananda | actualy | 01:36 |
| * devananda goes to retest it | 01:36 | |
| NobodyCam | https://wiki.openstack.org/wiki/I18NSupport | 01:36 |
| NobodyCam | found that one | 01:37 |
| NobodyCam | :-p | 01:37 |
| *** dwalleck has quit IRC | 01:40 | |
| devananda | urgh | 01:45 |
| devananda | still not working | 01:45 |
| devananda | NobodyCam: wow, that's really old | 01:46 |
| *** dwalleck has joined #openstack-ironic | 01:48 | |
| Shrews | ancient, even | 01:53 |
| devananda | so | 01:53 |
| Shrews | la | 01:54 |
| devananda | the manual import of i18n'd strings is not, actually, enough to get translation working for me | 01:54 |
| devananda | and combined with haomeng's patch, still not doing it | 01:54 |
| devananda | or I'm doing it wrong | 01:54 |
| devananda | but, for example I can issue | 01:54 |
| devananda | curl -X GET -H 'Accept-Language: pt_BR' -H 'Content-Type: application/json' http://127.0.0.1:6385/v1/nodes/?limit=-10 | 01:54 |
| openstackgerrit | A change was merged to openstack/ironic: Manually update all translated strings https://review.openstack.org/84259 | 01:54 |
| devananda | which should be returning the translated message string | 01:55 |
| devananda | msgstr "Limite deve ser positivo" | 01:55 |
| devananda | but isn't | 01:55 |
| Shrews | ugh, already with the april fool's stuff: http://blog.freenode.net/2014/04/googleplusfreenode/ | 02:01 |
| devananda | mordred: hi! around? got a sec? | 02:05 |
| lifeless | devananda: CI is back up | 02:30 |
| devananda | lifeless: \o/ | 02:31 |
| devananda | lifeless: so. should we be treating that job as voting? | 02:31 |
| lifeless | devananda: e.g. http://status.openstack.org/zuul/ see 84043 | 02:32 |
| lifeless | devananda: I know of no reason for it to fail other than code bugs [somewhere :P] | 02:32 |
| devananda | heh. great | 02:32 |
| lifeless | devananda: the undercloud job we don't know if it works yet, but we'll have more data when that one finishes | 02:33 |
| devananda | lifeless: random ? - you have knowledge of openstack's i18n / gettext hooks? I'm trying to figure out why it's not working right ... | 02:33 |
| lifeless | sorry no | 02:33 |
| lifeless | I haven't looked at that yet | 02:33 |
| Haomeng|2 | devananda: let me try to test my patch with non-en language, and show you the result | 02:36 |
| devananda | Haomeng|2: hi there! | 02:36 |
| Haomeng|2 | yes:) | 02:37 |
| devananda | Haomeng|2: i have been testing your patch with current master (the i18n strings landed just a bit ago) and have not been able to get it to work | 02:37 |
| devananda | Haomeng|2: please let me know how it goes. I'll be around for a while | 02:37 |
| Haomeng|2 | devananda: my patch have two fixes, one is to enable lazy translation to follow oslo change, another one is we do the formatting in common exception class to make sure the error message is in correct charset and encode | 02:38 |
| Haomeng|2 | devananda: ok, I will take other project's lazy translation enablement as reference | 02:39 |
| Haomeng|2 | devananda: I think I need some time to investigate why it is not working, so dont wait me for long time if you want to go to bed:) | 02:42 |
| devananda | Haomeng|2: I have a lot still to do tonight... will be up anyway for several hours more | 02:43 |
| Haomeng|2 | devananda: ok | 02:43 |
| Haomeng|2 | Haomeng|2: devananda: and I think maybe our ironic i18n not working without the lazy translation enablement, let me try to verify, now facing some env issue | 02:44 |
| Haomeng|2 | Haomeng|2: ok | 02:45 |
| Haomeng|2 | devananda: did you change po files to enable non-en strings, by default, all language string are in english I think | 02:54 |
| devananda | Haomeng|2: hm? | 02:54 |
| devananda | Haomeng|2: what do you mean, "enable non-en strings"? | 02:54 |
| Haomeng|2 | devananda: let me try to enable non-en strings in po files and test it, means to set non-en string for non-en po files, let me do it, and let you know the result:) | 02:55 |
| devananda | Haomeng|2: there are now many non-en strings, eg, http://git.openstack.org/cgit/openstack/ironic/tree/ironic/locale/zh_CN/LC_MESSAGES/ironic.po | 02:59 |
| Haomeng|2 | devananda: ok, let me fix my env and test them out | 02:59 |
| *** harlowja is now known as harlowja_away | 02:59 | |
| *** dwalleck has quit IRC | 03:02 | |
| *** killer_prince has quit IRC | 03:08 | |
| *** harlowja_away is now known as harlowja | 03:09 | |
| *** rameshg87 has joined #openstack-ironic | 03:10 | |
| *** radsy has quit IRC | 03:12 | |
| *** matsuhashi has quit IRC | 03:22 | |
| devananda | Haomeng|2: if I change ironic/openstack/common/gettextutils.py: gettext.translation(... fallback=True) to False | 03:28 |
| devananda | Haomeng|2: then I get: IOError: [Errno 2] No translation file found for domain: 'ironic' | 03:28 |
| Haomeng|2 | devananda: "gettextutils.install('ironic', lazy=True)" this code is suggestted by oslo, and I take othere project as reference, they have same code change | 03:29 |
| devananda | Haomeng|2: yes. I think that is a valid change. there is another problem too | 03:29 |
| Haomeng|2 | Haomeng|2: my env is back, let me continue to test this | 03:29 |
| devananda | i just dont know what the other problem is, yet | 03:29 |
| Haomeng|2 | devananda: yes | 03:29 |
| devananda | I am experimenting with manually setting | 03:30 |
| devananda | export IRONIC_LOCALEDIR=/opt/stack/ironic/ | 03:30 |
| Haomeng|2 | devananda: let me do some error message and to see if it can be display with non-en language | 03:30 |
| devananda | Haomeng|2: even starting the conductor service should produce non-en messages in the log now | 03:30 |
| devananda | that is how i'm testing | 03:30 |
| Haomeng|2 | devananda: yes | 03:31 |
| devananda | by making gettext not 'fallback' to en, it is erroring out quickly | 03:31 |
| Haomeng|2 | devananda: do you remove this line 'gettextutils.enable_lazy()' to rollback the code? | 03:32 |
| devananda | no | 03:33 |
| Haomeng|2 | devananda: I think both L18, L19 in this file https://review.openstack.org/#/c/74626/10/ironic/cmd/__init__.py enable the lazy mode to support 'Accept-Language' request I think | 03:35 |
| devananda | Haomeng|2: i have changed LC_ALL in the host | 03:35 |
| devananda | Haomeng|2: so translation is being done for log messages as well | 03:35 |
| Haomeng|2 | devananda: great | 03:35 |
| devananda | it makes testing faster :) | 03:35 |
| Haomeng|2 | devananda: good idea:) | 03:36 |
| Haomeng|2 | devananda: let me continue to test the api call with Accept-Language | 03:36 |
| Haomeng|2 | devananda: if log message is translated, that mean our code is working | 03:36 |
| *** nosnos has quit IRC | 03:37 | |
| devananda | Haomeng|2: well. it attempts to translate log message, but fails to find locale_dir and falls back to english | 03:38 |
| Haomeng|2 | devananda: this should be an issue I think | 03:40 |
| devananda | ? | 03:40 |
| Haomeng|2 | devananda: but I think oslo gettext code should get the relativepath in our ironic/locale | 03:41 |
| devananda | i have overridden local dir to /opt/stack/ironic/ironic/locale | 03:41 |
| devananda | but still get | 03:41 |
| devananda | IOError: [Errno 2] No translation file found for domain: 'ironic' | 03:41 |
| devananda | even though files exist in /opt/stack/ironic/ironic/locale/zh_CN/LC_MESSAGES/ironic.po | 03:42 |
| Haomeng|2 | did we remove this line "gettextutils.install('ironic', lazy=True)" which is registing our ironic domain | 03:42 |
| devananda | no. should I ? | 03:43 |
| devananda | ahh | 03:43 |
| devananda | Haomeng|2: so if I remove both of the lines you added to cmd/__init__ | 03:43 |
| devananda | then ironic-conductor starts | 03:44 |
| devananda | but does not translate anything | 03:44 |
| devananda | so your patch definitely causes ir-cond to *attempt* to translate | 03:44 |
| Haomeng|2 | devananda: the orig code has only one line " gettextutils.install('ironic')", which has not enable lazy i18n | 03:44 |
| *** eghobo has joined #openstack-ironic | 03:44 | |
| devananda | right. this does not work. | 03:45 |
| devananda | I modified the oslo gettextutils code to add some logging | 03:45 |
| devananda | if I revert your patch, that logging is not triggered | 03:45 |
| Haomeng|2 | devananda: strange | 03:46 |
| devananda | I am learning a lot about the translation code | 03:46 |
| devananda | but it seems odd that lazy=True is required for log messages when the system LC_CLANGUAGE is already set | 03:46 |
| Haomeng|2 | devananda: I understand "lazy=True" is just for api which support on-demand with the input language | 03:48 |
| devananda | right | 03:49 |
| devananda | except it looks like we need that for the LOG messages too .... | 03:49 |
| devananda | oh. one sec | 03:49 |
| Haomeng|2 | devananda: log message language is depending on the current /etc/sysconfig/i18n language configuration I think | 03:49 |
| devananda | may be an artefact of where i added logging :) | 03:49 |
| Haomeng|2 | devananda: :) | 03:50 |
| devananda | Haomeng|2: ye! that was it. I can revert your patch and get translation to try to run. it still fails to find the localedir though :) | 03:51 |
| devananda | Haomeng|2: so you did not intend to enable lazy translation for the conductor? | 03:53 |
| Haomeng|2 | devananda: the lazy enablement is on globel level I think for both conductor and api, it is in our ironic/cmd/__init__.py code | 03:54 |
| devananda | correct | 03:54 |
| devananda | but that seems unnecessary | 03:54 |
| devananda | shouldn't lazy only be enabled for the api? | 03:55 |
| devananda | ahh. nvm | 03:55 |
| devananda | ahh | 03:56 |
| devananda | hmmmm | 03:56 |
| Haomeng|2 | devananda: yes, if lazy is used for api only, it is not required for conductor | 03:56 |
| Haomeng|2 | Haomeng|2: :) | 03:56 |
| Haomeng|2 | devananda: let me check other project code change | 03:56 |
| devananda | Haomeng|2: if it should only be lazy in the api,then your change should be applied to ironic/cmd/api.py and ironic/api/app.wsgi | 03:58 |
| Haomeng|2 | devananda: yes, let me have try | 04:00 |
| Haomeng|2 | devananda: looks like other project are enableed lazy mode on globel level - https://github.com/openstack/glance/blob/master/glance/cmd/__init__.py#L17 | 04:00 |
| Haomeng|2 | devananda: not sure if they have good testing as well:) | 04:01 |
| Haomeng|2 | devananda: my env still have problem... | 04:02 |
| Haomeng|2 | devananda: will get back for a while... | 04:06 |
| *** lazy_prince has joined #openstack-ironic | 04:11 | |
| *** lazy_prince is now known as killer_prince | 04:11 | |
| *** matsuhashi has joined #openstack-ironic | 04:14 | |
| *** nosnos has joined #openstack-ironic | 04:23 | |
| adam_g | *sad face* | 04:25 |
| adam_g | devananda, i guess this is what we'll need to do till tempest provides a better way http://paste.ubuntu.com/7188082/ | 04:26 |
| devananda | adam_g: heh. it's not as bad as I expected .... | 04:27 |
| devananda | adam_g: admittedly i'm not that familiar with tempest's test suite yet. so maybe this is everything and i just dont realize it | 04:28 |
| adam_g | still pretty fugly | 04:28 |
| devananda | i'd almost prefer one line per category | 04:28 |
| adam_g | devananda, whadya mean? | 04:29 |
| devananda | but i dont know how infra would feel | 04:29 |
| devananda | adam_g: # skip tests that do <thing> ; r="$r|..." | 04:29 |
| devananda | pretty much just split up L13 | 04:30 |
| adam_g | devananda, oh, yeah unfortunately tempest.api.compute.* ends up stressing other non-nova stuff | 04:30 |
| devananda | ah | 04:30 |
| adam_g | ill see if i can clean it up some more before submitting it tomorrow | 04:31 |
| adam_g | running it against devstack now to confirm we at least get all green | 04:31 |
| devananda | compute.test_images_one_server sounds like the thing we'd want to run | 04:31 |
| devananda | but i haven't looked at the actual tests | 04:32 |
| adam_g | devananda, it creates an image from a running server, ie snapshot | 04:32 |
| devananda | ooh | 04:32 |
| devananda | heh | 04:32 |
| adam_g | :) | 04:32 |
| devananda | right. skip it :) | 04:32 |
| adam_g | while you're around... how do i get https://review.openstack.org/#/c/84043/ to depend on a more recent patchset in 84033? | 04:33 |
| *** killer_prince2 has joined #openstack-ironic | 04:35 | |
| devananda | adam_g: start with the latest 84033, cherry-pick 05d3b3ee3, git review -R | 04:35 |
| adam_g | devananda, ah thanks | 04:36 |
| devananda | -R will avoid potentially rebasing 84033, so even though git review will prompt you "are you sure you want to upload multiple thinsg" it actually wont | 04:36 |
| devananda | Haomeng|2: I'm going to toss up a slight change to your patch, in another patchset | 04:41 |
| Haomeng|2 | devananda: that is fine for me | 04:41 |
| devananda | Haomeng|2: i also noticed that your commit has a poor author line -- Author: root <whaom@cn.ibm.com> | 04:41 |
| devananda | Haomeng|2: i dont think your name is root :) | 04:41 |
| Haomeng|2 | devananda: :) | 04:42 |
| Haomeng|2 | devananda: that is default os user name:) | 04:42 |
| devananda | Haomeng|2: git config --global user.name XXX | 04:42 |
| Haomeng|2 | devananda: let me fix the name, yes | 04:42 |
| devananda | :) | 04:42 |
| Haomeng|2 | devananda: thanks for your reminder:) | 04:42 |
| openstackgerrit | Adam Gandelman proposed a change to openstack/ironic: Update Ironic Nova drivers' unit tests https://review.openstack.org/84043 | 04:43 |
| Haomeng|2 | devananda: I have checked our team guys which focus on i18n, he think the lazy mode does not effect the log message, so why othere project set it on globel I think | 04:44 |
| Haomeng|2 | devananda: but we can just enable lazy mode for api only | 04:45 |
| *** sabah has joined #openstack-ironic | 04:47 | |
| *** subah has joined #openstack-ironic | 04:57 | |
| *** subah has quit IRC | 04:58 | |
| *** zul has quit IRC | 05:00 | |
| *** matsuhas_ has joined #openstack-ironic | 05:02 | |
| *** matsuhas_ has quit IRC | 05:02 | |
| devananda | Haomeng|2: ok. so even with that fix, api is not attempting to translate strings | 05:03 |
| devananda | Haomeng|2: right now we can only get logged translations, not api translations | 05:03 |
| devananda | Haomeng|2: we are missing a translation hook in the api middleware | 05:03 |
| *** zul has joined #openstack-ironic | 05:04 | |
| *** matsuhashi has quit IRC | 05:05 | |
| Haomeng|2 | devananda: but i understand the _("") hook is installed by our code gettextutils.install('ironic') on global level | 05:05 |
| devananda | yes | 05:05 |
| devananda | it's not quite that simple though | 05:05 |
| Haomeng|2 | devananda: yes | 05:05 |
| *** matsuhashi has joined #openstack-ironic | 05:05 | |
| devananda | the api service isn't changing the requested language | 05:05 |
| devananda | when calling Message.translate, the desired_locale defaults to None. | 05:06 |
| devananda | this needs to be passed in by the API | 05:06 |
| devananda | or it will default to the system locale | 05:06 |
| Haomeng|2 | devananda: ok, we can call " gettextutils.install('ironic') " in our api code to have try | 05:07 |
| devananda | which was already working | 05:07 |
| lifeless | devananda: hey so | 05:08 |
| Haomeng|2 | devananda: yes, we have to change /etc/sysconfig/i18n to set non-en language | 05:08 |
| lifeless | ironic is failing on deploys | 05:08 |
| devananda | lifeless: eh? | 05:08 |
| lifeless | devananda: in CI; I'm filing a bug now - but AFAICT its no long an infrastructure issue | 05:08 |
| lifeless | longer that is | 05:08 |
| lifeless | devananda: https://bugs.launchpad.net/ironic/+bug/1300589 | 05:10 |
| *** harlowja is now known as harlowja_away | 05:11 | |
| lifeless | devananda: note that the seed doesn't actually *deploy* anything - if we add a baremetal cirros test or something like that then we'd be able to show this failure in the seed job | 05:12 |
| lifeless | devananda: [thats a known thing-to-fix] | 05:12 |
| *** killer_prince has quit IRC | 05:14 | |
| devananda | lifeless: ERROR Timeout reached when waiting callback for node | 05:17 |
| devananda | lifeless: so. why did it time out? | 05:17 |
| lifeless | devananda: I don't know | 05:17 |
| devananda | lifeless: this is suspicious: ERROR Request returned failure status | 05:17 |
| devananda | from glance | 05:17 |
| lifeless | 2014-04-01 03:31:17.878 3865 WARNING keystoneclient.middleware.auth_token [-] Authorization failed for token | 05:18 |
| lifeless | 2014-04-01 03:31:17.878 3865 WARNING keystoneclient.middleware.auth_token [-] Authorization failed for token | 05:18 |
| devananda | lifeless: in the ir-cond log? | 05:18 |
| lifeless | devananda: its odd that you'd get an error from glance and then still timeout rather than erroring immediately :) that might point to a bug | 05:18 |
| lifeless | devananda: thats glance log | 05:18 |
| lifeless | devananda: the logs are in the seed-logs.tgz | 05:18 |
| devananda | so that ought to have failed when ir-cond first pulled the images down during prepare() | 05:19 |
| lifeless | timestamps may be important here | 05:19 |
| lifeless | the 31:17 lines up exactly | 05:19 |
| devananda | Buffer I/O error on device loop0p1 | 05:20 |
| devananda | any concern there? | 05:20 |
| lifeless | anything after 02:55 is when we have started the deploy (so are assuming the seed is fully configured, images loaded etc) | 05:20 |
| lifeless | dib triggers that loop0p1 fairly routinely | 05:20 |
| lifeless | we've not tracked any negatives to it at this point- believed to be a udev race (udev reading a device we've removed) | 05:21 |
| Haomeng|2 | devananda: this is heat code change - https://review.openstack.org/#/c/69133/5/heat/common/wsgi.py which call gettextutils.translate in wsgi | 05:21 |
| Haomeng|2 | devananda: maybe we need such call in our ironic wsgi, let me find | 05:21 |
| devananda | lifeless: provisioning starts at | 05:23 |
| devananda | [01/Apr/2014 02:55:50] "PUT /v1/nodes/f36bc755-df36-40fc-8d57-9504e194eadd/states/provision | 05:23 |
| devananda | lifeless: there's no issue with glance auth at that time | 05:23 |
| devananda | lifeless: so what changed between 2:55 and the glance auth failure? | 05:24 |
| lifeless | devananda: nothing - last line in os-collect-config is | 05:24 |
| lifeless | [2014-04-01 02:40:46,791] (os-refresh-config) [INFO] Completed phase migration | 05:24 |
| lifeless | devananda: so all our deploy glue was idle | 05:25 |
| openstackgerrit | lokesh s proposed a change to openstack/python-ironicclient: adding ironic command line interfaces https://review.openstack.org/84346 | 05:25 |
| lifeless | keystone log : | 05:25 |
| lifeless | 2014-04-01 02:40:48.286 3794 WARNING keystone.common.wsgi [-] Could not find user, ironic. | 05:25 |
| lifeless | 2014-04-01 03:31:17.876 3794 WARNING keystone.common.wsgi [-] Could not find token, admin. | 05:25 |
| lifeless | that 2:40 one is a bit odd | 05:25 |
| *** loki_ has joined #openstack-ironic | 05:25 | |
| Haomeng|2 | devananda: maybe we need a customized wsgi to enable the exception translation | 05:30 |
| lifeless | devananda: could possibly be just that ironic used an expired token | 05:33 |
| devananda | lifeless: are you guys expiring tokens quickly? | 05:33 |
| lifeless | devananda: defaults :) | 05:34 |
| devananda | lifeless: which is? | 05:34 |
| lifeless | as much as possible | 05:34 |
| lifeless | elements/keystone/os-apply-config/etc/keystone/keystone.conf | 05:34 |
| lifeless | I believe the default is one hour | 05:34 |
| devananda | 86400 | 05:35 |
| devananda | is what's commented out in that keystone.conf | 05:35 |
| lifeless | yeah | 05:35 |
| devananda | which is far longer than 22 minutes | 05:35 |
| lifeless | which would have come from keystone.conf.sample | 05:35 |
| lifeless | where are you getting 22 minutes from ? | 05:35 |
| devananda | bad math | 05:36 |
| devananda | 02:40 - 03:31 | 05:36 |
| devananda | 51 minutes, not 22 | 05:36 |
| devananda | i was looking at a different number | 05:37 |
| devananda | still should not have caused the token to expire | 05:37 |
| lifeless | 2:55 is when the deploy started | 05:37 |
| lifeless | could you have gotten the token before the deploy started? | 05:37 |
| lifeless | (when does Ironic acquire tokens? and does Ironic automatically retry when a token failure is encountered ?) | 05:38 |
| *** sabah has quit IRC | 05:41 | |
| devananda | hmm | 05:46 |
| devananda | it's conceivable that clean_up happened with a context that wasn't auth'd for glance access | 05:47 |
| devananda | lifeless: so it's not an issue with keystone token | 05:47 |
| devananda | lifeless: both prepare and clean_up call _get_tftp_image_info, which calls out to glance using our glance_service wrapper | 05:48 |
| devananda | they both pass in the existing context, which may include an auth token | 05:48 |
| devananda | but if clean_up is called from a periodic task (which it was in this case) | 05:48 |
| devananda | then taht token was generated by ironic, not by nova (or nova's user) | 05:49 |
| devananda | so | 05:49 |
| devananda | that's potentially a bug | 05:49 |
| devananda | but not the cause of your problem | 05:49 |
| lifeless | clean_up is called from nova delete? | 05:49 |
| lifeless | devananda: so - this should be reproducible | 05:50 |
| lifeless | devananda: pushing up patches should let you gather more info in the same way as with e.g. devstack-gate | 05:50 |
| devananda | clean_up may be called from 'nova delete' or from ironic.conductor.manager | 05:50 |
| devananda | _check_deploy_timeouts | 05:51 |
| lifeless | devananda: if I deploy two instances with the same root image, and a single conductor, will it be shared between them ? | 05:51 |
| devananda | i believe so, yes | 05:51 |
| lifeless | cool | 05:51 |
| lifeless | so, I did some basic timing with Ironic the other day | 05:52 |
| lifeless | and it was slower than our nova-baremetal runs | 05:52 |
| lifeless | which surprised me a little | 05:52 |
| lifeless | but I haven't dug in | 05:52 |
| lifeless | so I need to context switch | 05:52 |
| lifeless | if you need anything admin privileged to help debug this please let me know | 05:53 |
| devananda | lifeless: afaict the VM did not DHCP boot | 05:53 |
| devananda | lifeless: which is why it eventually timed out | 05:53 |
| lifeless | we're running nobodycams patch for ci_commands | 05:53 |
| lifeless | or it would have errored on power management much earlier and not tried the deploy | 05:53 |
| devananda | i'm not sure how that relates to it not DHCPBOOTing | 05:54 |
| *** pradipta_away is now known as pradipta | 05:54 | |
| lifeless | devananda: the VM configuration is static | 05:54 |
| devananda | ... | 05:54 |
| lifeless | devananda: and identical for all of them; only way it coud not DHCP boot is if it wasn't powered on | 05:54 |
| devananda | how can it get tftp targets then? | 05:54 |
| devananda | i dont see any requsets to tftpd in the logs | 05:55 |
| devananda | (or what ever tftpd was replaced with) | 05:55 |
| devananda | or is there a static tftpd elsewhere too? | 05:55 |
| lifeless | Apr 1 03:00:38 ubuntu dnsmasq-dhcp[4833]: DHCPDISCOVER(tap1e5579ec-96) 52:54:00:b5:63:b7 | 05:56 |
| lifeless | Apr 1 03:00:38 ubuntu dnsmasq-dhcp[4833]: DHCPOFFER(tap1e5579ec-96) 192.0.2.3 52:54:00:b5:63:b7 | 05:56 |
| lifeless | so - thats DHCP boot | 05:56 |
| lifeless | of the VM | 05:56 |
| devananda | that MAC is not the MAC that was enrolled with ironic | 05:56 |
| devananda | er. nvm | 05:56 |
| devananda | maybe it was | 05:56 |
| * devananda wishes POST content was logged somewhere ... | 05:56 | |
| lifeless | https://jenkins03.openstack.org/job/check-tripleo-ironic-undercloud-precise/123/console is doing the same thing | 05:57 |
| lifeless | so I don't think its a one-off virsh fail | 05:57 |
| lifeless | http://logs.openstack.org/43/84043/2/check-tripleo/check-tripleo-ironic-undercloud-precise/a956ec9/console.html has the full testenv descriptor at the top | 05:57 |
| lifeless | 52:54:00:b5:63:b7 is the MAC of the first node in the nodes list | 05:57 |
| lifeless | so that is indeed the VM registerer with Ironic | 05:58 |
| devananda | ok. | 05:58 |
| lifeless | if it wasn't that MAC dnsmasq would not have responded | 05:58 |
| devananda | so. should we be seeing a log from tftp when the VM requested the kernel/ramdisk/token? | 05:58 |
| devananda | I do within my devstack env | 05:58 |
| lifeless | where does that get logged? | 05:59 |
| devananda | var/log/syslog | 05:59 |
| lifeless | ok so possible failures | 06:01 |
| lifeless | we know BM1 -> seed worked for DHCP | 06:01 |
| lifeless | we've no reason to expect TFTP to fail | 06:01 |
| lifeless | TFTPD might not be running | 06:01 |
| lifeless | the port might not have PXE options configured on it | 06:03 |
| lifeless | e.g. if you signal neutron to set pxe | 06:03 |
| lifeless | and then boot the vm before neutron has hupped the dnsmasq process | 06:04 |
| lifeless | yup | 06:05 |
| lifeless | huge race condition | 06:05 |
| devananda | lifeless: in theory, it wouldn't get the dhcp response that we see in the logs if that were the case | 06:05 |
| devananda | but you're using a static dhcp config | 06:05 |
| devananda | i dunno how that will play with tftpd | 06:05 |
| devananda | afaik, that shouldn't affect tftp. ironic has prepared the tfptd files before we power on the machine | 06:06 |
| lifeless | devananda: static dhcp config ? | 06:06 |
| lifeless | devananda: no you misunderstand | 06:06 |
| devananda | ? | 06:07 |
| openstackgerrit | Jenkins proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/83956 | 06:07 |
| lifeless | devananda: neutron is doing the dhcp right | 06:07 |
| lifeless | devananda: _update_neutron is called to pass the needed PXE boot options into neutron | 06:07 |
| lifeless | devananda: and then without waiting we power on the VM | 06:09 |
| lifeless | devananda: see comment 1 in the bug | 06:09 |
| lifeless | devananda: the call to neutron returns when neutron has the request in the DB | 06:09 |
| lifeless | devananda: not when the dnsmasq agent is updated | 06:09 |
| devananda | right | 06:10 |
| devananda | but if that were the problem here | 06:10 |
| devananda | we'd not see the DHCP RESPOSNE | 06:10 |
| devananda | but we do | 06:10 |
| *** loki__ has joined #openstack-ironic | 06:10 | |
| lifeless | devananda: yes you would | 06:10 |
| devananda | ? | 06:10 |
| lifeless | devananda: because the port when created can do DHCP | 06:10 |
| lifeless | devananda: just not PXE | 06:10 |
| devananda | ah | 06:10 |
| devananda | urgh. right. | 06:10 |
| devananda | it's late :) | 06:11 |
| lifeless | yes | 06:11 |
| lifeless | so I'm going to add a 10 seconds delay there and push a branch | 06:11 |
| lifeless | and we'll see if it passes | 06:11 |
| devananda | k | 06:11 |
| *** loki_ has quit IRC | 06:11 | |
| devananda | i'm going to spin off the glance issue into a new bug | 06:11 |
| devananda | then sign off for the night | 06:12 |
| Haomeng|2 | devananda: one question, do you know how to add transations for our ironic wsgi api call? we need it to enable the lazy translations, maybe we have to do it in next release | 06:14 |
| Haomeng|2 | devananda: we have no such wsgi.py code to customize the api call | 06:15 |
| devananda | lifeless: fwiw, https://bugs.launchpad.net/ironic/+bug/1300602 | 06:15 |
| Haomeng|2 | devananda: we are different with heat wsgi code I think | 06:16 |
| devananda | Haomeng|2: yes. two places | 06:16 |
| lifeless | devananda: thanks | 06:16 |
| openstackgerrit | lifeless proposed a change to openstack/ironic: Workaround race with Neutron port updates https://review.openstack.org/84361 | 06:16 |
| Haomeng|2 | devananda: ok, I want to enable it if possible | 06:16 |
| Haomeng|2 | devananda: is this one ParsableErrorMiddleware? | 06:17 |
| devananda | Haomeng|2: yep. so take a look at ceilometer's ParsableErrorMiddleware | 06:17 |
| devananda | ours is similar, but lacks the translation hook | 06:17 |
| Haomeng|2 | devananda: ok | 06:17 |
| Haomeng|2 | devananda: thank you | 06:17 |
| Haomeng|2 | devananda: have a good dream:) | 06:17 |
| devananda | Haomeng|2: to enable lazy in the api, you need to add it to both | 06:17 |
| Haomeng|2 | devananda: yes | 06:17 |
| devananda | cmd/api.py and api/app.wsgi | 06:18 |
| Haomeng|2 | devananda: ok, got:) | 06:18 |
| Haomeng|2 | devananda: thanks | 06:18 |
| devananda | api.py is for python WSGI app. api/app.wsgi is for Apache mod_wsgi applications | 06:18 |
| devananda | Haomeng|2: thanks! I made some progress and can post it, but it's not working yet | 06:18 |
| Haomeng|2 | devananda: ok, got | 06:19 |
| Haomeng|2 | devananda: my pleasure:) | 06:19 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Add lazy translation to API tier https://review.openstack.org/84362 | 06:20 |
| devananda | Haomeng|2: feel free to take what you can from ^ and fit it into your patch. | 06:20 |
| Haomeng|2 | devananda: ok, let me see your code | 06:20 |
| Haomeng|2 | devananda: sure, thank you:) | 06:21 |
| devananda | thank you :) | 06:21 |
| devananda | good night | 06:21 |
| Haomeng|2 | devananda: :) | 06:21 |
| *** saju_m has joined #openstack-ironic | 06:21 | |
| Haomeng|2 | devananda: I am in afternoon, good night for you, and have a nice dream:) | 06:21 |
| openstackgerrit | lokesh s proposed a change to openstack/python-ironicclient: adding ironic command line interfaces https://review.openstack.org/84346 | 06:33 |
| *** nosnos has quit IRC | 06:45 | |
| *** nosnos has joined #openstack-ironic | 06:46 | |
| *** matsuhashi has quit IRC | 06:58 | |
| *** matsuhashi has joined #openstack-ironic | 06:58 | |
| *** jistr has joined #openstack-ironic | 07:02 | |
| *** matsuhashi has quit IRC | 07:03 | |
| *** matsuhashi has joined #openstack-ironic | 07:06 | |
| GheRivero | morning all | 07:15 |
| Haomeng|2 | GheRivero: morning:) | 07:15 |
| dtantsur | Morning Ironic | 07:20 |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Handling validation in conductor consistently https://review.openstack.org/84374 | 07:24 |
| *** mrda is now known as mrda_away | 07:29 | |
| *** ndipanov has joined #openstack-ironic | 07:33 | |
| *** eghobo has quit IRC | 07:45 | |
| openstackgerrit | Sergey Nikitin proposed a change to openstack/ironic: Improved coverage for ironic API https://review.openstack.org/84376 | 07:48 |
| *** max_lobur has joined #openstack-ironic | 07:53 | |
| *** athomas has joined #openstack-ironic | 07:54 | |
| *** killer_prince has joined #openstack-ironic | 08:02 | |
| *** matsuhashi has quit IRC | 08:03 | |
| *** matsuhas_ has joined #openstack-ironic | 08:05 | |
| *** lucasagomes has joined #openstack-ironic | 08:13 | |
| *** romcheg has joined #openstack-ironic | 08:21 | |
| *** killer_prince has quit IRC | 08:23 | |
| *** romcheg has quit IRC | 08:23 | |
| *** derekh has joined #openstack-ironic | 08:24 | |
| *** romcheg has joined #openstack-ironic | 08:24 | |
| *** matsuhas_ has quit IRC | 08:33 | |
| *** matsuhashi has joined #openstack-ironic | 08:38 | |
| *** saju_m has quit IRC | 08:43 | |
| yuriyz | morning Ironic | 08:46 |
| Haomeng|2 | yuriyz: morning:) | 08:46 |
| romcheg | Morning guys | 08:53 |
| *** saju_m has joined #openstack-ironic | 08:56 | |
| *** athomas has quit IRC | 08:59 | |
| *** yonglihe_ has quit IRC | 09:00 | |
| lucasagomes | morning yuriyz Haomeng|2 romcheg | 09:04 |
| Haomeng|2 | lucasagomes: morning:) | 09:04 |
| romcheg | lucasagomes: My US visa was approved. So I will go to the summit, if nothing special happens :) | 09:05 |
| lucasagomes | romcheg, w00t!!! nice! | 09:05 |
| Haomeng|2 | romcheg: congts:) | 09:06 |
| *** overlayer has joined #openstack-ironic | 09:07 | |
| *** pradipta is now known as pradipta_away | 09:19 | |
| *** martyntaylor has quit IRC | 09:26 | |
| *** martyntaylor has joined #openstack-ironic | 09:26 | |
| *** jistr has quit IRC | 09:39 | |
| *** athomas has joined #openstack-ironic | 09:50 | |
| *** athomas has quit IRC | 09:56 | |
| *** athomas has joined #openstack-ironic | 09:59 | |
| *** jistr has joined #openstack-ironic | 09:59 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Replace sfdisk with parted https://review.openstack.org/84396 | 10:12 |
| *** matsuhashi has quit IRC | 10:14 | |
| rameshg87 | hello | 10:21 |
| rameshg87 | when does ironic open for juno ? we were a few items which were moved from icehouse and was asked to resubmit when juno reopens | 10:22 |
| lucasagomes | rameshg87, hi, I don't know exactly :/ but I'm expecting it to open this week | 10:23 |
| *** ndipanov has quit IRC | 10:23 | |
| lucasagomes | I think that after Ironic get's tagged as RC1 | 10:23 |
| lucasagomes | we are good to go | 10:23 |
| rameshg87 | thanks :-) | 10:24 |
| *** overlayer has quit IRC | 10:26 | |
| *** overlayer has joined #openstack-ironic | 10:31 | |
| *** overlayer has quit IRC | 10:34 | |
| *** ndipanov has joined #openstack-ironic | 10:36 | |
| *** nosnos has quit IRC | 10:40 | |
| *** overlayer has joined #openstack-ironic | 10:40 | |
| *** overlayer has quit IRC | 10:42 | |
| *** overlayer has joined #openstack-ironic | 10:45 | |
| *** overlayer has quit IRC | 10:59 | |
| *** rameshg87 has left #openstack-ironic | 11:06 | |
| *** mdurnosvistov1 has joined #openstack-ironic | 11:14 | |
| *** sphoorti has joined #openstack-ironic | 11:29 | |
| *** lucasagomes is now known as lucas-hungry | 11:34 | |
| *** loki__ has quit IRC | 11:39 | |
| *** saju_m has quit IRC | 11:44 | |
| openstackgerrit | Jenkins proposed a change to openstack/ironic: Updated from global requirements https://review.openstack.org/83471 | 11:50 |
| *** saju_m has joined #openstack-ironic | 11:56 | |
| *** saju_m has quit IRC | 12:02 | |
| romcheg | lucas-hungry: Your last patch is intended to replace 83399 and 83396, isn't it? | 12:04 |
| romcheg | I like KISS methodology :) | 12:05 |
| lucas-hungry | romcheg, for the rc1 release only, 83399 and 83396 are plumbings for other stuff (like getting rid of the swap partition, GPT partitions, layout consistency) | 12:05 |
| lucas-hungry | but it's too big to land for the rc | 12:05 |
| lucas-hungry | so the new one is just a short-term fix | 12:06 |
| romcheg | Ah, I see | 12:06 |
| romcheg | I thought we planned to get rid of low level operations like disk partitioning in Ironic and move them to the agent? | 12:06 |
| romcheg | *in the next release* | 12:07 |
| lucas-hungry | yeah, the agent it another driver | 12:07 |
| lucas-hungry | the default pxe will still need some partitioning logic I think | 12:07 |
| lucas-hungry | is another driver* | 12:07 |
| romcheg | Makes sense | 12:07 |
| romcheg | I regret missing the meet up :( | 12:07 |
| lucas-hungry | :) no worries | 12:08 |
| lucas-hungry | yeah in the logs there's a bunch of talks related to the disk stuff | 12:08 |
| *** overlayer has joined #openstack-ironic | 12:10 | |
| *** romcheg has left #openstack-ironic | 12:10 | |
| *** romcheg has joined #openstack-ironic | 12:10 | |
| lucas-hungry | brb lunch is done :) | 12:11 |
| romcheg | lucas-hungry: Bon appetite! | 12:11 |
| *** ndipanov has quit IRC | 12:18 | |
| *** overlayer has quit IRC | 12:18 | |
| *** saju_m has joined #openstack-ironic | 12:21 | |
| *** saju_m has quit IRC | 12:23 | |
| *** ndipanov has joined #openstack-ironic | 12:30 | |
| *** rloo has joined #openstack-ironic | 12:34 | |
| *** lucas-hungry is now known as lucasagomes | 12:36 | |
| *** saju_m has joined #openstack-ironic | 12:39 | |
| *** linggao has joined #openstack-ironic | 12:39 | |
| *** jdob has joined #openstack-ironic | 12:40 | |
| *** jbjohnso_ has joined #openstack-ironic | 12:41 | |
| *** mtrisoline has quit IRC | 12:42 | |
| *** jgrimm has joined #openstack-ironic | 12:43 | |
| *** yuriyz has quit IRC | 12:47 | |
| *** overlayer has joined #openstack-ironic | 12:47 | |
| *** yuriyz has joined #openstack-ironic | 12:50 | |
| *** overlayer has quit IRC | 12:54 | |
| *** jbjohnso_ has quit IRC | 12:58 | |
| *** killer_prince2 has quit IRC | 12:58 | |
| *** jbjohnso_ has joined #openstack-ironic | 13:13 | |
| *** saju_m has quit IRC | 13:15 | |
| dtantsur | Any ideas what to do with error "RPC do_node_tear_down not allowed for node e8d9b219-4c7f-44b5-baa5-00d939b35180 in state None"? | 13:16 |
| dtantsur | This happened after unsuccessful deploy because my nodes were in maintenance mode (for unknown reason) | 13:17 |
| dtantsur | Now I cannot even nova delete this instance | 13:17 |
| jroll | dtantsur: I've had the same problem, I ended up dropping to a mysql shell on ironic to delete the instance, then nova can delete it ok | 13:19 |
| jroll | (if I remember correctly) | 13:19 |
| jroll | also, good morning ironic! | 13:19 |
| dtantsur | jroll, morning, thanks. So there should be something wrong in error handling path of either nova or ironic... | 13:20 |
| jroll | dtantsur: yeah | 13:22 |
| jroll | this is where I left off friday, and I didn't take a look yesterday | 13:22 |
| jroll | but will be working on it today or tomorrow sometime, I think | 13:23 |
| dtantsur | jroll, do we have bug # for it? | 13:23 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic-python-agent: Add kernel parameter support https://review.openstack.org/84264 | 13:24 |
| rloo | dtantsur: my guess if your node went to maintenance mode, is that the periodic task to sync the power state didn't work and after 3 times was put in maint mode :-( | 13:24 |
| jroll | dtantsur: not yet, I haven't investigated enough that I'm sure it's a bug | 13:24 |
| dtantsur | jroll, not being able to delete failed instance is definitely a bug :) | 13:25 |
| rloo | dtantsur: would be good if you would open a bug about it | 13:25 |
| jroll | touché :P | 13:25 |
| dtantsur | rloo, jroll ok, I'll open a bug | 13:25 |
| jroll | awesome, thanks | 13:26 |
| rloo | thx dtantsur! | 13:26 |
| jroll | dtantsur: my rationale may have been that it was 6:30 pm on a friday and I wasn't sure if I was doing something dumb :P | 13:26 |
| jroll | agordeev: quick review ftw :) | 13:27 |
| *** saju_m has joined #openstack-ironic | 13:28 | |
| lucasagomes | dtantsur, jroll I opnened a bug about it, I'm taking a look at it right now | 13:28 |
| lucasagomes | https://bugs.launchpad.net/ironic/+bug/1300288 | 13:28 |
| jroll | ah ha! | 13:29 |
| dtantsur | lucasagomes, cool! | 13:29 |
| jroll | I was wondering where teardown was being called, I was looking for some deploy failed case | 13:29 |
| lucasagomes | before we didn't see that cause the validation of the do_node_deploy() was async | 13:29 |
| lucasagomes | jroll, it's nova | 13:29 |
| jroll | thanks lucasagomes :) | 13:29 |
| jroll | ahhh | 13:29 |
| lucasagomes | rescheduling the instance | 13:30 |
| jroll | lucasagomes: right, saw that in the bug :) | 13:30 |
| Haomeng|2 | hello every one, who know how to generate mo file for our ironic po files to enable i18n? | 13:30 |
| dtantsur | now I have one more problem: after I start boot, I see Ironic probing 3 nodes in a row (they are virtual, so I see the console), tftp succedes, kernel boots, then machine shuts down, next one is selected... | 13:33 |
| dtantsur | After all, boot fails with NoValidHost | 13:33 |
| dtantsur | An ideas, what the cause is? | 13:34 |
| *** jgrimm has quit IRC | 13:34 | |
| Haomeng|2 | have to run "python msgfmt.py *.mo" to create? | 13:35 |
| rloo | Haomeng|2, what's a .mo file? (ie, I don't know.) | 13:36 |
| NobodyCam | good morning Ironic... | 13:38 |
| romcheg | Morning NobodyCam | 13:38 |
| NobodyCam | hey hey romcheg :) | 13:38 |
| rloo | mornin' NobodyCam | 13:38 |
| NobodyCam | morning rloo :) | 13:39 |
| agordeev | morning NobodyCam, also morning/evening Ironic! | 13:39 |
| NobodyCam | and mornign agordeev :) | 13:39 |
| NobodyCam | morning even :-p | 13:39 |
| agordeev | rloo: it's all about translation strings. I saw devanda had been asking the same question in openstack-dev last night | 13:40 |
| rloo | agordeev: do you know if he got an answer? maybe Haomeng|2 should look there | 13:41 |
| agordeev | ah, devananda, it meant | 13:41 |
| agordeev | rloo: nope, he asked clarkb, but he didn't know too | 13:41 |
| rloo | Haomeng|2: ^^ | 13:42 |
| agordeev | Haomeng|2: this is somehow related to babel. Are you familiar with it? | 13:43 |
| rloo | lucasagomes: question about docn for python-ironicclient. do you have any thoughts about it? | 13:46 |
| rloo | lucasagomes: I see keystone has a README, and wiki. | 13:47 |
| rloo | lucasagomes: https://github.com/openstack/python-keystoneclient/blob/master/README.rst?#id10 | 13:47 |
| rloo | lucasagomes: http://docs.openstack.org/user-guide/content/keystoneclient_commands.html | 13:47 |
| *** jbjohnso_ has quit IRC | 13:49 | |
| NobodyCam | wow I was just reading the story book that is our last night scroll back | 13:50 |
| NobodyCam | have not finished | 13:50 |
| NobodyCam | have we tested the translation stuff? | 13:51 |
| romcheg | Shame on me. I didn't do the Ukrainian translation. | 13:58 |
| lucasagomes | rloo, hmm, idk really, before I thought about a man for the cli | 13:59 |
| lucasagomes | but I had some push backs on that | 14:00 |
| * lucasagomes trying to remember the reasons | 14:00 | |
| rloo | lucasagomes: I remember that. I think it was because we thought it would be too much to maintain? | 14:00 |
| lucasagomes | the keystone readme looks good | 14:00 |
| lucasagomes | rloo, I think so | 14:00 |
| rloo | lucasagomes: I just saw this review: https://review.openstack.org/#/c/84346/2/doc/source/ironic-commands.rst | 14:01 |
| lucasagomes | https://review.openstack.org/#/c/52902/ | 14:01 |
| *** jbjohnso_ has joined #openstack-ironic | 14:01 | |
| NobodyCam | romcheg: do you know how to test the translations? | 14:01 |
| romcheg | NobodyCam: hmm, I can judge on Ukrainian and Russian: There's no way to test translations | 14:02 |
| rloo | lucasagomes: so do we have usages examples in the developer page? | 14:02 |
| romcheg | You have to be a human being to make translations | 14:02 |
| NobodyCam | romcheg: I've searched for how to set them up doc.. and been unable to find a good one | 14:03 |
| lucasagomes | rloo, hmm I don't think we have any cli examples there no | 14:03 |
| romcheg | We can only check the coverage | 14:03 |
| romcheg | Like how many text was translated | 14:04 |
| lucasagomes | rloo, so, according to devananda comment at https://review.openstack.org/#/c/52902/ | 14:04 |
| romcheg | Spellchecking is not possible because of a lot of context-specific things | 14:04 |
| lucasagomes | we should have some examples in the dev docs | 14:04 |
| lucasagomes | which I agree | 14:05 |
| lucasagomes | for both, CLI and LIB | 14:05 |
| *** max_lobur1 has joined #openstack-ironic | 14:05 | |
| rloo | lucasagomes: if https://review.openstack.org/#/c/84346/2 is good enough wrt the CLI commands, maybe examples can be added here? | 14:05 |
| *** max_lobur has quit IRC | 14:07 | |
| lucasagomes | rloo, I think so, I don't have a strong opinion where it should go tho, I'm fine having examples in the dev doc... But I also think that a wiki would be easier to maintain | 14:08 |
| rloo | lucasagomes: ok, let me snoop around at the other projects and see if there is some consistency or not. That review looks like it needs a bit of work too... | 14:09 |
| NobodyCam | romcheg: when you say theres no way to test translations. I'm unclear. I am look to do the reverse. I would like to set up a test system to see the translated text even if I don't understand it | 14:10 |
| lucasagomes | rloo, heh yeah, there's some Ironic (capitalized there) which will fail :P | 14:10 |
| devananda | good morning, all | 14:10 |
| NobodyCam | good morning devananda :) wow late night for you lastnight | 14:10 |
| rloo | lucasagomes: exactly. might be other stuff too, i'm not sure how much development this person has done. | 14:10 |
| rloo | mornin' devananda | 14:10 |
| devananda | NobodyCam: yea... trying to fix translation | 14:11 |
| lucasagomes | yeah | 14:11 |
| lucasagomes | morning devananda | 14:11 |
| NobodyCam | i have a mo file. I just dont know how to use it | 14:11 |
| NobodyCam | :-p | 14:11 |
| devananda | oh! | 14:11 |
| devananda | how'd you get that??? | 14:11 |
| NobodyCam | msgfmt -o locale/es/LC_MESSAGES/ironic.po locale/es/LC_MESSAGES/ironic.mo | 14:12 |
| NobodyCam | you may nneed to apt-get install gettext | 14:12 |
| NobodyCam | but what do I do with said .mo file | 14:12 |
| romcheg | NobodyCam: It's not a problem to get a translated text, but what's the next step? | 14:12 |
| NobodyCam | I'll happly convert my seed to a different lang for testing | 14:13 |
| *** saju_m has quit IRC | 14:14 | |
| devananda | heh | 14:14 |
| devananda | so there are 2 situations where i18n should happen | 14:14 |
| devananda | LOG messages when system language is different | 14:15 |
| devananda | API responses when header Accept-Language is different | 14:15 |
| devananda | Haomeng's patch should be addressign the API response -- it needs to be "lazy translated", iow, dynamic based on the user's request | 14:15 |
| devananda | and LOG translation should already have been working | 14:15 |
| devananda | there were a lot of problems | 14:16 |
| devananda | * our .po files contained none of the translated strings which were in transifex (FIXED) | 14:16 |
| devananda | * jenkins wasn't submitting the translations (pending patch in infra) | 14:16 |
| devananda | * openstac/common/gettextutils is looking in system default location, so ironic needs to be packaged. install from source doesn't put the langs in the right place | 14:17 |
| romcheg | devananda: Sometimes we format error messages which then go both to the log and to the API response. | 14:17 |
| devananda | (can work around with export IRONIC_LOCALEDIR=/path/to/locale) | 14:17 |
| devananda | there's also no logging when gettextutils fails to find a translation | 14:18 |
| lucasagomes | NobodyCam, hey there, when you get a time can u give it a go and see if it's working fine in ur devtest env? https://review.openstack.org/#/c/84396/ | 14:18 |
| devananda | it silently falls back to en_US | 14:18 |
| romcheg | devananda: If system language is different from the Accept-Language, that will cause a problem | 14:18 |
| lucasagomes | I tested here in devstack, looks fine | 14:18 |
| devananda | romcheg: yes | 14:18 |
| devananda | romcheg: so this is something we will need to address in Juno | 14:19 |
| romcheg | I also think this is sort of out of Ironic's scope | 14:19 |
| NobodyCam | lucasagomes: I tested the pated patches yesterday. they worked then.. have they changed | 14:20 |
| *** jgrimm has joined #openstack-ironic | 14:20 | |
| romcheg | Because the other projects also use localisation | 14:20 |
| lucasagomes | NobodyCam, right... yeah it does pretty much the same... that's the short-term fix for icehouse that we disucssed yesterday in the meeting | 14:21 |
| NobodyCam | ack | 14:21 |
| NobodyCam | give me a couple of min | 14:21 |
| devananda | romcheg: it's the way we are creating messages in conductor and passing back to API | 14:21 |
| devananda | romcheg: where we pass an exception it should work | 14:21 |
| lucasagomes | NobodyCam, np! thanks | 14:21 |
| devananda | romcheg: api can localize the exc | 14:22 |
| devananda | romcheg: but it can't localize what we save to node.last_error, for example -- that has to be done by conductor (if it should be done at all, which is questionable) | 14:22 |
| romcheg | devananda: Yes, that is the fix for this problem. But testing that is quite hard. | 14:22 |
| NobodyCam | brb | 14:23 |
| dtantsur | lucasagomes, back to https://bugs.launchpad.net/ironic/+bug/1300288 will it also fix inability to delete failed instances? | 14:23 |
| *** mdurnosvistov1 has quit IRC | 14:23 | |
| romcheg | And it requires more projects than just Ironic | 14:24 |
| romcheg | s/requires/is required/ | 14:24 |
| *** lazy_prince has joined #openstack-ironic | 14:27 | |
| *** lazy_prince is now known as killer_prince | 14:27 | |
| lucasagomes | dtantsur, yes | 14:30 |
| lucasagomes | dtantsur, well at least I expect it to fix that too | 14:31 |
| lucasagomes | dtantsur, I've a quick hacky fixing it and it allows me to delete it afterwards | 14:31 |
| *** killer_prince has quit IRC | 14:34 | |
| *** lazy_prince has joined #openstack-ironic | 14:34 | |
| *** lazy_prince is now known as killer_prince | 14:35 | |
| NobodyCam | lucasagomes: testing now | 14:36 |
| lucasagomes | NobodyCam, :D cheers | 14:37 |
| dtantsur | lucasagomes, any ideas on this: http://paste.openstack.org/show/74756/ ? is it partition-flavor size mismatch or something else? | 14:41 |
| *** ciranttech has joined #openstack-ironic | 14:41 | |
| devananda | NobodyCam: that cmd does not seem to create valid .mo files | 14:43 |
| devananda | IOError: [Errno 0] Bad magic number: '/opt/stack/ironic/ironic/locale/zh_CN/LC_MESSAGES/ironic.mo' | 14:43 |
| devananda | NobodyCam: in fact, it doesn't compile the .mo at all | 14:43 |
| lucasagomes | dtantsur, hmmm haven't seem that... bad input/no partitions found... this whole disk stuff is quite broke right now | 14:43 |
| lucasagomes | the size of the disk supports the size of the partitions? | 14:44 |
| NobodyCam | :( devananda finding info on this is tuff.. I'll jump back on it right after I test lucasagomes is patch | 14:44 |
| NobodyCam | it at least created a .mo file for me | 14:44 |
| devananda | we already have .po files | 14:44 |
| devananda | this is creating that format | 14:44 |
| dtantsur | lucasagomes, not sure. My "baremetals" are just libvirt VMs with 21G hard drive | 14:44 |
| lucasagomes | dtantsur, sounds like enough space for tests heh | 14:45 |
| lucasagomes | dtantsur, so, hmm you mind trying it with the new patch I put up? | 14:45 |
| dtantsur | lucasagomes, which one exactly? | 14:45 |
| lucasagomes | dtantsur, https://review.openstack.org/#/c/84396/ | 14:46 |
| dtantsur | will try | 14:46 |
| NobodyCam | msgfmt -o locale/es/LC_MESSAGES/ironic.mo locale/es/LC_MESSAGES/ironic.po | 14:46 |
| NobodyCam | hahhah I had the order wrong in my paste | 14:46 |
| NobodyCam | sorry :( | 14:46 |
| lucasagomes | dtantsur, the sfdisk + 1MB lead space is causing a lot of misaligned problems, I hope that ^ will fix it | 14:46 |
| *** ciranttech has left #openstack-ironic | 14:47 | |
| lucasagomes | I mean that ^ 84396 | 14:47 |
| devananda | got it! | 14:50 |
| NobodyCam | devananda: the mo file? | 14:50 |
| devananda | NobodyCam: Haomeng|2: pybabel compile -i <file.po> -o <file.mo> | 14:50 |
| devananda | or | 14:51 |
| dtantsur | lucasagomes, Command: sudo ironic-rootwrap /etc/ironic/rootwrap.conf parted -a optimal -s /dev/disk/by-path/ip-192.0.2.14:3260-iscsi-iqn-c1361bc2-b28e-461f-a5a7-1714cf2c2d4c-lun-1 -- mklabel msdos unit MiB mkpart primary 0 21504 mkpart primary linux-swap 21504 21505 | 14:51 |
| devananda | cd ironic/locale && for file in $(ls -1 */LC_MESSAGES/ironic.po); do pybabel compile -i $file -o ${file%%.po}.mo ; done | 14:51 |
| dtantsur | lucasagomes, 'Error: The location 21504 is outside of the device /dev/sda.\n' | 14:51 |
| lucasagomes | dtantsur, a-ha! so the disk is too small | 14:51 |
| dtantsur | lucasagomes, too small for what? | 14:51 |
| devananda | awesome. now when I change LC_ALL, conductor logs are translated | 14:51 |
| lucasagomes | dtantsur, the root partition | 14:52 |
| lucasagomes | check the properties local_gb | 14:52 |
| lucasagomes | 21504 = 21 G? | 14:52 |
| dtantsur | lucasagomes, u'local_gb': u'21' | 14:52 |
| lucasagomes | rigth yeah | 14:52 |
| lucasagomes | and what's the size of the disk of the machine ur trying to deploy to? | 14:54 |
| dtantsur | lucasagomes, mkpart primary linux-swap 21504 21505 <-- is it correct? | 14:54 |
| NobodyCam | devananda: the command I used did create a mo file.. I orginaly pasted the command with the options reversed | 14:54 |
| dtantsur | lucasagomes, virt-manager shows 21.00 GB | 14:54 |
| lucasagomes | dtantsur, unfortunately... in trunk, if no swap is specified we create a partition with 1MB | 14:54 |
| lucasagomes | dtantsur, which sucks I know | 14:54 |
| lucasagomes | dtantsur, https://review.openstack.org/#/c/83726/ fix that | 14:55 |
| lucasagomes | but it won't land until juno opens | 14:55 |
| dtantsur | lucasagomes, I think we try to create partition (N, N+1), maybe we should try (N-1, N) instead? N - size of disk | 14:55 |
| lucasagomes | dtantsur, +1 | 14:55 |
| lucasagomes | dtantsur, actually we need to detect it before | 14:55 |
| devananda | NobodyCam: it's not "just a .mo file" -- it's a specific format of file | 14:55 |
| lucasagomes | dtantsur, https://review.openstack.org/#/c/78001/ | 14:56 |
| devananda | NobodyCam: msgfmt doesn't create them, babel does | 14:56 |
| NobodyCam | ahh | 14:56 |
| lucasagomes | dtantsur, not only that, we should consider root+swap+ephemeral | 14:56 |
| NobodyCam | devananda: would you throw up ether pad with the comamnd you use to test translations | 14:56 |
| lucasagomes | that's why I want the disk partitioning refactor to land so we can start tackling all these related problems with disk | 14:56 |
| lucasagomes | dtantsur, right now in trunk, it's very messy | 14:57 |
| NobodyCam | so I can learn | 14:57 |
| NobodyCam | how to too | 14:57 |
| devananda | https://etherpad.openstack.org/p/TestingTranslations | 14:57 |
| dtantsur | lucasagomes, is it possible to fix size calculation within your patch? For me it's plainly wrong to try to write partition at (N,N+1) | 14:58 |
| *** dwalleck has joined #openstack-ironic | 14:59 | |
| lucasagomes | dtantsur, yeah, so that patch is a short-term one | 14:59 |
| lucasagomes | for the rc1 release only | 14:59 |
| lucasagomes | cause the refactor is too big | 14:59 |
| *** sphoorti_ has joined #openstack-ironic | 15:00 | |
| *** sphoorti has quit IRC | 15:00 | |
| dtantsur | lucasagomes, that's ok, but I think we need to fix deployment for rc1 ;) | 15:00 |
| dtantsur | in the meanwhile, how can I work around? | 15:01 |
| dtantsur | make flavor smaller? | 15:01 |
| lucasagomes | heh yeah :( | 15:01 |
| devananda | NobodyCam: I think that's it | 15:01 |
| devananda | Haomeng|2: https://etherpad.openstack.org/p/TestingTranslations | 15:01 |
| NobodyCam | woo hoo.. and you get different output | 15:02 |
| devananda | yep | 15:02 |
| devananda | NobodyCam: so that worked? was typing from last night's memory :) | 15:02 |
| *** matty_dubs|gone is now known as matty_dubs | 15:02 | |
| NobodyCam | mind if I share in #openstack-translation ? | 15:02 |
| agordeev | dtantsur: it's partially work-arounded in devstack. https://github.com/openstack-dev/devstack/blob/master/tools/ironic/scripts/create-nodes#L14-L16 HTH | 15:02 |
| devananda | NobodyCam: no -- please do | 15:02 |
| NobodyCam | :) | 15:03 |
| devananda | I would like this to become automated | 15:03 |
| devananda | so it doesn't break again | 15:03 |
| * lucasagomes wants juno to opens | 15:03 | |
| lucasagomes | open* | 15:03 |
| dtantsur | agordeev, I see, thank you. So smaller flavor + smaller ironic node size. | 15:03 |
| dtantsur | that's why things don't break for guys who use devstack | 15:03 |
| devananda | lucasagomes: me2 | 15:04 |
| devananda | lucasagomes: i'm confident now that we have translations at least for operator log files | 15:05 |
| devananda | even if the API isn't translated yet | 15:05 |
| devananda | haomeng and i were looking into that ... but we're missing a middleware :( | 15:05 |
| devananda | and my late-night attempts to hack one together failed | 15:05 |
| lucasagomes | devananda, ah ouch! | 15:06 |
| lucasagomes | but yeah it's very needed | 15:06 |
| devananda | yea | 15:06 |
| lucasagomes | and nobody seems to have noticed it was broken | 15:06 |
| lucasagomes | until u guys started looking into it | 15:06 |
| lucasagomes | so good work :) | 15:06 |
| devananda | cause testing of i18n is not done prior to packaging | 15:06 |
| devananda | which happens AFTER i tag the release | 15:06 |
| devananda | :( | 15:06 |
| devananda | anyhow, i think the fix for API i18n will be small | 15:07 |
| devananda | and contained in just a few files | 15:07 |
| devananda | easy backport if we/someone want it | 15:07 |
| lucasagomes | yay! tag me on the review when u get it up, I'll be happy to review that | 15:07 |
| devananda | lucasagomes: 74626 is mostly there | 15:07 |
| devananda | just missing amiddleware | 15:08 |
| lucasagomes | and test (ironic in portuguese :D) | 15:08 |
| devananda | woops | 15:08 |
| lucasagomes | devananda, will take a look | 15:08 |
| devananda | wrong link | 15:08 |
| devananda | lucasagomes: https://review.openstack.org/#/c/84362/ | 15:08 |
| devananda | that's got all the pieces | 15:08 |
| devananda | just the broken middleware | 15:08 |
| devananda | small enough I'm OK with backporting it | 15:09 |
| devananda | so | 15:09 |
| devananda | I'm goign to tag the release unless anyone stops me in the next 15 minutes :) | 15:09 |
| NobodyCam | I am still testing lucasagomes sfdidk fix | 15:09 |
| NobodyCam | sfdisk even | 15:10 |
| lucasagomes | devananda, ack | 15:10 |
| lucasagomes | devananda, btw, the fix that NobodyCam is testing is the quick fix for the swap >= 1G | 15:10 |
| lucasagomes | NobodyCam, :) thanks | 15:10 |
| lucasagomes | <=* | 15:11 |
| lucasagomes | urgh | 15:11 |
| lucasagomes | >= | 15:11 |
| NobodyCam | hehehe | 15:11 |
| * lucasagomes confused | 15:11 | |
| openstackgerrit | Tom Fifield proposed a change to openstack/ironic: Fix grammar in error string in pxe driver https://review.openstack.org/84471 | 15:11 |
| *** sphoorti__ has joined #openstack-ironic | 15:11 | |
| * NobodyCam hands lucasagomes more coffee | 15:11 | |
| lucasagomes | NobodyCam, cheers buddy hah | 15:11 |
| lucasagomes | I need it | 15:11 |
| *** sphoorti_ has quit IRC | 15:12 | |
| NobodyCam | woo hoo patches from the i18n / L10n team | 15:12 |
| openstackgerrit | Tom Fifield proposed a change to openstack/ironic: Fix spelling error in conductor/manager https://review.openstack.org/84472 | 15:14 |
| NobodyCam | lucasagomes: undecloud deploying in process... sorry have to rebuild everything | 15:16 |
| * NobodyCam makes coffee | 15:16 | |
| lucasagomes | :D no worries | 15:17 |
| openstackgerrit | A change was merged to openstack/ironic: Check that all po/pot files are valid https://review.openstack.org/84210 | 15:17 |
| *** dwalleck has quit IRC | 15:20 | |
| devananda | lucasagomes: link? | 15:22 |
| NobodyCam | lucasagomes: undercloud deployed and configured! | 15:23 |
| mordred | NobodyCam: w00t | 15:23 |
| lucasagomes | devananda, https://review.openstack.org/#/c/84396/ | 15:23 |
| lucasagomes | NobodyCam, yay! | 15:23 |
| lucasagomes | NobodyCam, thanks for testing :) | 15:23 |
| NobodyCam | morning mordred :) | 15:24 |
| mordred | morning NobodyCam ! | 15:24 |
| BadCub01 | Morning mordred :-) | 15:24 |
| BadCub01 | And Mornng Deva and NobodyCam | 15:25 |
| *** ifarkas has quit IRC | 15:26 | |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Open Juno development https://review.openstack.org/84474 | 15:27 |
| * lucasagomes wants to +2 ^ :P | 15:29 | |
| romcheg | lucasagomes: -2 is already there :) | 15:29 |
| lucasagomes | heh romcheg yeah | 15:29 |
| NobodyCam | lol | 15:29 |
| lucasagomes | damn I was too slow | 15:29 |
| lucasagomes | heh | 15:29 |
| devananda | lol | 15:29 |
| rloo | how does python-ironicclient relate to the icehouse version of ironic? | 15:30 |
| rloo | the other clients seem to be released on a diff schedule. | 15:31 |
| devananda | rloo: yep, independent schedules | 15:31 |
| devananda | I haven't seen any changes recently that requried a new release | 15:31 |
| devananda | but imbw. let me know :) | 15:32 |
| devananda | lucasagomes: looking at https://review.openstack.org/#/c/84396/1/ironic/drivers/modules/deploy_utils.py | 15:32 |
| devananda | lucasagomes: I'd like more eyes on that | 15:32 |
| devananda | as it's a pretty important change, and we have lots of history with sfdisk and none with parted | 15:32 |
| rloo | looks like python-ironicclient-0.1.2 was uploaded on 2014-02-25. | 15:33 |
| lucasagomes | devananda, right, yeah I tested it locally here with the ephemeral, w/o ephemeral, 1mb swap, 2G swap, etc.. situations | 15:33 |
| lucasagomes | devananda, yeah | 15:33 |
| lucasagomes | devananda, NobodyCam also tested it in his last run of devtest | 15:33 |
| lucasagomes | I know it's risky, but currently sfdisk + 1MB lead partition is broken | 15:33 |
| devananda | lucasagomes: ack | 15:33 |
| devananda | but it works with < 1GB swap | 15:33 |
| devananda | so | 15:33 |
| lucasagomes | yeah | 15:34 |
| devananda | i'd also really like lifeless' input on it | 15:34 |
| devananda | as he had a lot of opinions on the sfdisk patches originally | 15:35 |
| NobodyCam | I good with cutting the RC with current... that is was a release candidate is all about | 15:35 |
| lucasagomes | devananda, +1 | 15:35 |
| devananda | ok | 15:36 |
| devananda | so we have two items with backport potential | 15:36 |
| devananda | API i18n and sfdisk->parted | 15:36 |
| devananda | it's good practice for me :) | 15:36 |
| romcheg | +1 | 15:36 |
| NobodyCam | :) | 15:36 |
| rloo | wrt python-ironicclient, there are some requirements changes. (version updates for pbr, httplib2, keystoneclient, six) | 15:41 |
| rloo | and code changes related to console (which we don't care), --maintenance option for node-list, driver-show cmd | 15:43 |
| NobodyCam | bbt... brb | 15:52 |
| devananda | lucasagomes: so having set-console-mode, get-console in the CLI when they're not in the API is a bit odd | 15:52 |
| lucasagomes | devananda, they r in the api yes, no? /me checks | 15:53 |
| devananda | ah. they are. | 15:53 |
| devananda | nvm | 15:53 |
| devananda | just no driver implements it | 15:53 |
| lucasagomes | https://github.com/openstack/ironic/blob/master/ironic/api/controllers/v1/node.py#L73 | 15:54 |
| lucasagomes | yeah | 15:54 |
| devananda | yea | 15:54 |
| devananda | sorry :) | 15:54 |
| lucasagomes | np | 15:54 |
| *** mdurnosvistov1 has joined #openstack-ironic | 15:54 | |
| devananda | rloo: lucasagomes: new 0.1.3 release of the client coming soon | 15:54 |
| devananda | just pushed the tag to gerrit | 15:54 |
| lucasagomes | yay! | 15:55 |
| lucasagomes | thanks | 15:55 |
| *** matty_dubs is now known as matty_dubs|lunch | 15:57 | |
| rloo | devananda: thx (and lucasagomes too) | 15:58 |
| *** romcheg has quit IRC | 16:00 | |
| rloo | devananda: http://docs.openstack.org/developer/ironic/. It is still alpha quality right? | 16:02 |
| devananda | rloo: should probably be updated to say beta, heh | 16:03 |
| devananda | but that's just my opinion | 16:03 |
| rloo | devananda: the only other thing is that it mentions baremetal in grizzly and havana, should probably say icehouse too. but seems minor. | 16:03 |
| devananda | yea | 16:03 |
| rloo | devananda: it is too late to change for icehouse rc, right? | 16:04 |
| devananda | no | 16:04 |
| rloo | devananda. well, we're going to be adding more docn anyway... | 16:04 |
| devananda | but it's an online doc | 16:04 |
| devananda | right | 16:04 |
| *** eghobo has joined #openstack-ironic | 16:04 | |
| *** BadCub01 has quit IRC | 16:05 | |
| rloo | yeah, on-line, and if someone actually wants to build the doc themselves, but let's assume not ;) | 16:05 |
| NobodyCam | post bbt walkies... bbiafm | 16:05 |
| lucasagomes | someone knows any exception that can be returned in the spawn() method of nova | 16:12 |
| lucasagomes | to tell nova to _not_ try to re-scheduler the instance?! | 16:13 |
| devananda | nope | 16:13 |
| lucasagomes | I've tried DeployInstanceFailure, NovaException, InvalidParameterError | 16:13 |
| lucasagomes | :( | 16:13 |
| devananda | lucasagomes: but. look at nova/filters/retry? | 16:13 |
| lucasagomes | all then try to reshceduler it | 16:13 |
| lucasagomes | 2014-04-01 17:09:21.635 TRACE nova.compute.utils [instance: 8ebc65de-45af-4b20-a074-6547246db027] RescheduledException: Build of instance 8ebc65de-45af-4b20-a074-6547246db027 was re-scheduled: test internal server error | 16:13 |
| devananda | it's the retry filter causing that | 16:13 |
| lucasagomes | ah! | 16:13 |
| devananda | so if there is a wa yto signal "dont retry" it'd be there | 16:13 |
| lucasagomes | devananda, cheers will see | 16:13 |
| *** jistr has quit IRC | 16:17 | |
| devananda | yurg | 16:18 |
| devananda | can't "nova delete" when the power is off | 16:18 |
| devananda | *when the provision state is none | 16:18 |
| devananda | so if provision fails (as it does sometimes) then nova is stuck | 16:19 |
| devananda | ironic won't let it delete teh instance | 16:19 |
| lucasagomes | well seems there's none | 16:19 |
| lucasagomes | devananda, heh that's what I'm fixing | 16:19 |
| devananda | great | 16:19 |
| lucasagomes | devananda, https://bugs.launchpad.net/ironic/+bug/1300288 | 16:19 |
| devananda | i was trying to test your parted patch | 16:20 |
| lucasagomes | devananda, right, any errors on the ir-cond related to the parted stuff? | 16:20 |
| NobodyCam | and back :) | 16:20 |
| jroll | lucasagomes: if you can't figure it out, probably worth asking comstud about it | 16:22 |
| jroll | wrt nova stuff | 16:22 |
| lucasagomes | jroll, yeah | 16:23 |
| lucasagomes | jroll, I took a look at the filter it seems to retry anyway, doesn't even check for a particular exception | 16:23 |
| lucasagomes | but I will ask in #openstack-nova | 16:23 |
| jroll | yeah, that works too | 16:23 |
| devananda | lucasagomes: none yet, since it failed before then | 16:24 |
| lucasagomes | jroll, thanks, asked there and pinged comstud :) | 16:24 |
| jroll | why do we not want nova to reschedule it? because ironic will retry? | 16:24 |
| jroll | or because we should assume the node is broken somehow? | 16:24 |
| lucasagomes | jroll, if the deploy() fails in things like validation of the driver parameters | 16:24 |
| lucasagomes | there's no reason to retry it | 16:25 |
| *** mdurnosvistov1 has quit IRC | 16:25 | |
| jroll | yeah | 16:25 |
| devananda | well | 16:25 |
| devananda | maybe that node's driver_info is bad | 16:25 |
| devananda | and another node will work | 16:25 |
| devananda | there are cases where retry is good | 16:25 |
| *** dwalleck has joined #openstack-ironic | 16:25 | |
| devananda | we can allow the driver to make some decisions there | 16:25 |
| jroll | I'm wondering if we should put the provision state in something other than None | 16:25 |
| *** dwalleck_ has joined #openstack-ironic | 16:25 | |
| lucasagomes | devananda, hmm yeah, but driver_info is populated by the nova ironic driver no?, depends on the driver | 16:25 |
| lucasagomes | yeah there's many corner cases | 16:26 |
| *** mdurnosvistov1 has joined #openstack-ironic | 16:26 | |
| devananda | lucasagomes: so, not entirely, and not for long | 16:26 |
| devananda | lucasagomes: we're going to move all the stuff which nova is writing to driver_info today, into instance_info | 16:26 |
| lucasagomes | yeah | 16:26 |
| devananda | jroll has a patch up for that, iirc | 16:26 |
| * lucasagomes wants the deploy_info stuff! | 16:26 | |
| jroll | devananda: I have a patch up to create the instance_info field | 16:27 |
| jroll | nothing up to move the data yet | 16:27 |
| devananda | right. that'll be in the nova driver | 16:27 |
| jroll | right | 16:27 |
| *** dwalleck has quit IRC | 16:29 | |
| *** dwalleck has joined #openstack-ironic | 16:31 | |
| *** dwalleck_ has quit IRC | 16:33 | |
| devananda | lucasagomes: oh, also, looks like nova delete is keying on the stored node_uuid | 16:35 |
| devananda | lucasagomes: rather than the instance_uuid | 16:35 |
| *** dwalleck_ has joined #openstack-ironic | 16:35 | |
| devananda | lucasagomes: when I manually unassociated it in ironic, nova's still calling do_node_teardown on the node! | 16:35 |
| lucasagomes | devananda, yeah, I have to change the tear_down to check which state the provision_state of the node is | 16:36 |
| lucasagomes | so based on that it then issue the request to tear down the node | 16:36 |
| devananda | lucasagomes: and to qery by instance_uuid | 16:36 |
| lucasagomes | devananda, yup | 16:36 |
| lucasagomes | node = validate_instance_and_node(icli, instance) | 16:37 |
| lucasagomes | it tries to get the node using the instance_uuid | 16:37 |
| *** dwalleck has quit IRC | 16:37 | |
| lucasagomes | if it's not there we just return | 16:37 |
| lucasagomes | cause it's a non-existent instance | 16:37 |
| lucasagomes | I will try to push my patch soon | 16:37 |
| openstackgerrit | Pablo Fernando Cargnelutti proposed a change to openstack/ironic: Overwriting node_is_available in IronicDriver https://review.openstack.org/84214 | 16:37 |
| devananda | right now any attempt i'm making to boot | 16:39 |
| devananda | yields this | 16:39 |
| devananda | http://paste.openstack.org/show/74762/ <-- NobodyCam, lucasagomes | 16:39 |
| * NobodyCam looks | 16:39 | |
| * lucasagomes checks | 16:39 | |
| NobodyCam | devananda: did you move the driver location? | 16:40 |
| NobodyCam | /opt/stack/nova/nova/virt/ironic/driver.py | 16:40 |
| NobodyCam | should that not be /opt/stack/ironic/nova/virt/ironic/driver.py | 16:41 |
| *** ndipanov has quit IRC | 16:41 | |
| *** athomas has quit IRC | 16:41 | |
| lucasagomes | looks like an old error | 16:41 |
| lucasagomes | devananda, https://review.openstack.org/#/c/78686/ | 16:42 |
| devananda | hrmm | 16:43 |
| *** derekh has quit IRC | 16:44 | |
| devananda | ah. old .pyc files in nova.virt.ironic | 16:48 |
| NobodyCam | doh | 16:49 |
| * lucasagomes hate when it happens | 16:49 | |
| *** hemna_ has joined #openstack-ironic | 16:51 | |
| devananda | HTTPBadRequest: Couldn't apply patch '[{'path': '/extra/vif_port_id', 'op': 'remove'}]'. Reason: u'vif_port_id' | 16:52 |
| devananda | ok, other than that, deploy seems to be working | 16:52 |
| NobodyCam | devananda: vif_port_id not set on that node? | 16:52 |
| devananda | NobodyCam: ive seen that error in all the recent virtual-ironic jenkins test runs | 16:54 |
| *** BadCub01 has joined #openstack-ironic | 16:57 | |
| *** matty_dubs|lunch is now known as matty_dubs | 16:57 | |
| *** stevehuang has joined #openstack-ironic | 16:59 | |
| *** dwalleck_ has quit IRC | 17:01 | |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Reduce logging output from non-Ironic libraries https://review.openstack.org/84496 | 17:04 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Fix Nova rescheduling tear down problem https://review.openstack.org/84497 | 17:04 |
| Shrews | soooo much easier to grok the logs with that ^^^^, but open to adding some back if others find the noise useful | 17:04 |
| lucasagomes | dtantsur, ^ | 17:05 |
| devananda | Shrews: some of those might be better left at INFO | 17:06 |
| devananda | Shrews: but I agree that DEBUG is too noisy | 17:06 |
| Shrews | devananda: i didn't see any cases where they output anything at INFO level | 17:06 |
| Shrews | devananda: but it's easy to miss it as it is | 17:06 |
| lucasagomes | Shrews, devananda +1 very noise yeah | 17:06 |
| devananda | lucasagomes: ok. so you have 2 patches up that are both pretty important fixes | 17:07 |
| lucasagomes | devananda, right | 17:07 |
| devananda | well. honestly. we have a lot of fixes ... | 17:07 |
| lucasagomes | +1 | 17:07 |
| NobodyCam | review jam tomorrow morning? | 17:07 |
| lucasagomes | the i18n stuff also very important for the release | 17:08 |
| lucasagomes | NobodyCam, we can do that, I mean, I would like to have a test jam! | 17:08 |
| lucasagomes | heh | 17:08 |
| devananda | lucasagomes: API or LOG i18n? or both | 17:08 |
| lucasagomes | I guess both | 17:08 |
| devananda | cause we have LOG i18n now | 17:08 |
| lucasagomes | that patch u pointed me to? /me have to test | 17:08 |
| devananda | that patch is for APi i18n | 17:09 |
| lucasagomes | ah right, I haven't tested i18n at all, so... :( | 17:09 |
| lucasagomes | ok I will test the api i18n | 17:09 |
| devananda | it's gonna break. the middleware is busted in that WIP | 17:09 |
| lucasagomes | :/ | 17:10 |
| devananda | so when i said i was gonna tag it unless anyone objects, you uploaded more patches and I see you targeted teh parted to RC1 | 17:11 |
| devananda | (maybe you did that before) | 17:11 |
| devananda | :) | 17:11 |
| devananda | lucasagomes: should I take that as an answer, you want me to hold the tag? | 17:11 |
| devananda | *hold off on | 17:11 |
| lucasagomes | heh I did that before | 17:11 |
| devananda | ah | 17:12 |
| lucasagomes | but, hmmmm idk it sounds like an important fix | 17:12 |
| devananda | yea | 17:12 |
| devananda | i agree | 17:12 |
| lucasagomes | partitioning broke really sucks | 17:12 |
| devananda | after fixing up my env issues, parted worked for me with 2GB swap | 17:12 |
| lucasagomes | but I understand the risk as well | 17:12 |
| lucasagomes | :) | 17:12 |
| lucasagomes | yeah | 17:12 |
| devananda | example of my concer - has anyone tested parted on 12.04 | 17:12 |
| lucasagomes | so I tested here locally many combinations, but you never know | 17:12 |
| lucasagomes | I haven't :( | 17:13 |
| devananda | since that's waht's in the gate | 17:13 |
| lucasagomes | maybe the devstack guys | 17:13 |
| NobodyCam | brb | 17:13 |
| lucasagomes | I've a ubuntu vm here but I think it's not 12.04, lemme check | 17:14 |
| *** harlowja_away is now known as harlowja | 17:14 | |
| lucasagomes | 13.04 :/ | 17:14 |
| devananda | i hesitate to tag when we already have a fix up. why do a backport if we can just wait a few more hours to land this? | 17:14 |
| lucasagomes | yeah | 17:15 |
| lucasagomes | makes sense to hold a bit more | 17:15 |
| devananda | lifeless: when you're up, I would like your input on https://review.openstack.org/#/c/84396/1 -- this (new patch) is holding up the RC, and I think you had expressed opinios on parted/sfdisk/etc things in the past. | 17:16 |
| lucasagomes | devananda, better wait until tomorrow tops? then we can see lifeless opnions about it | 17:17 |
| lucasagomes | otherwise we can tag it and if it's _really_ needed we backport | 17:17 |
| lucasagomes | is good that there's a plan b | 17:17 |
| devananda | yea | 17:17 |
| devananda | ok. things work, just with limitation right now | 17:18 |
| lucasagomes | yeah | 17:18 |
| devananda | lucasagomes: if you agree, would you mind commenting / untargeting the bug? | 17:18 |
| lucasagomes | devananda, sure, no objections | 17:18 |
| devananda | great | 17:18 |
| JayF | FWIW parted as old as lucid iirc supports MiB | 17:19 |
| JayF | I've used it since 10.04 with that flag | 17:20 |
| lucasagomes | devananda, done | 17:20 |
| lucasagomes | JayF, nice it's a good feedback, thanks | 17:22 |
| JayF | Parted is great. And it's good to move to it (even if in Juno) because it supports GPT whereas sfdisk doesn't, and GPT is required for extra-large boot devices among other things | 17:22 |
| lucasagomes | exactly | 17:22 |
| lucasagomes | >2TB partitions will need gpt | 17:23 |
| NobodyCam | yes | 17:23 |
| lucasagomes | the idea of the refactoring was to enable that as well | 17:23 |
| JayF | I used parted here http://git.openstack.org/cgit/openstack/ironic-python-agent/tree/ironic_python_agent/shell/copy_configdrive_to_disk.sh#n25 because it lets you set a negative offset to put something at the end vs the beginning of a disk, was super useful for that | 17:23 |
| lucasagomes | nice! | 17:24 |
| lucasagomes | JayF, for the agent you guys might want to use the python-parted libs as well | 17:25 |
| lucasagomes | I've worked with it in another project (https://github.com/umago/carbono) | 17:25 |
| lucasagomes | it's quite useful | 17:25 |
| JayF | I'm sure we'll work all that shell stuff out of the agent over time | 17:25 |
| JayF | was just a quick and easy way to interact with the disks for getting something working fast | 17:25 |
| lucasagomes | yeah I understand | 17:26 |
| devananda | lucasagomes: approved. we should see it merge soon | 17:30 |
| devananda | lucasagomes: at which point ttx will branch, and we should retarget your fix to that branch | 17:31 |
| devananda | lucasagomes: and land all the diskpartitioning things on master | 17:31 |
| devananda | and al lthe other things | 17:31 |
| openstackgerrit | A change was merged to openstack/ironic-python-agent: Add kernel parameter support https://review.openstack.org/84264 | 17:31 |
| *** dwalleck has joined #openstack-ironic | 17:34 | |
| NobodyCam | brb | 17:38 |
| lucasagomes | devananda, yay! ack | 17:42 |
| lucasagomes | will retarget | 17:42 |
| lucasagomes | devananda, all the other things you mean, get rid of swap, etc? | 17:42 |
| lucasagomes | whole series? | 17:42 |
| lucasagomes | serie* | 17:43 |
| devananda | yes | 17:43 |
| lucasagomes | devananda, ah I see, approved the tag u mean? | 17:43 |
| lucasagomes | nice ok | 17:43 |
| devananda | rc1 status page was green :) | 17:44 |
| NobodyCam | I have tested them.. (not with > 1 gb of awap thou | 17:44 |
| NobodyCam | ) | 17:44 |
| NobodyCam | was | 17:44 |
| devananda | for the i18n stuff, i added some more notes to https://etherpad.openstack.org/p/TestingTranslations just now | 17:44 |
| devananda | for testing the API | 17:44 |
| lucasagomes | so if we are going to approve the series that short-term can be abandoned | 17:44 |
| devananda | lucasagomes: no... | 17:44 |
| devananda | lucasagomes: well. it can be. yes. | 17:45 |
| NobodyCam | there are questions out to lifeless on the parted patches no? | 17:45 |
| lucasagomes | yeah, that waas a quick fix for the rc1 release only | 17:45 |
| devananda | but I would prefer you leave it up as a WIP or target it to rc2 | 17:45 |
| lucasagomes | NobodyCam, yeah | 17:45 |
| devananda | we can have more RCs, and folks may need things back ported | 17:45 |
| lucasagomes | devananda, ack will WIP it out | 17:45 |
| NobodyCam | lol :) | 17:46 |
| devananda | i'm not sure exactly how it will go. first time we've done a release with folks using it :) | 17:46 |
| * NobodyCam heard devo in his head | 17:46 | |
| lucasagomes | ack :) | 17:46 |
| lucasagomes | wipped | 17:46 |
| devananda | thanks! | 17:46 |
| Shrews | NobodyCam: ugh | 17:46 |
| devananda | lifeless: pls disregard my earlier ping. unless you just want to talk about parted vs sfdisk in general :) | 17:47 |
| Shrews | NobodyCam: that one song makes me wish i had chosen a different term than "work in progress" when i implemented that | 17:47 |
| NobodyCam | https://www.youtube.com/watch?v=j_QLzthSkfM | 17:47 |
| NobodyCam | lol | 17:48 |
| lucasagomes | lifeless yeah if there's objects about parted please put at https://review.openstack.org/#/c/83396 | 17:48 |
| lucasagomes | objections* | 17:48 |
| lucasagomes | ack I'm done for today then | 17:48 |
| * lucasagomes tired | 17:48 | |
| NobodyCam | have a good night lucasagomes :) | 17:48 |
| lucasagomes | have a g'night devananda NobodyCam everyone :) | 17:48 |
| NobodyCam | enjoy a pint for me | 17:49 |
| *** lucasagomes is now known as lucas-afk | 17:49 | |
| lucas-afk | NobodyCam, heh will do :) | 17:49 |
| devananda | g'ngiht lucas-afk ! | 17:50 |
| Shrews | adam_g: fyi, giving your nova driver tests change a test drive | 17:56 |
| adam_g | Shrews, cool :) | 17:58 |
| *** sphoorti__ has left #openstack-ironic | 17:58 | |
| openstackgerrit | A change was merged to openstack/ironic: Open Juno development https://review.openstack.org/84474 | 17:59 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Add BackOffLoopingCall with jitter https://review.openstack.org/84303 | 18:00 |
| *** dwalleck_ has joined #openstack-ironic | 18:02 | |
| *** dwalleck_ has quit IRC | 18:03 | |
| JoshNang | woo Juno! | 18:04 |
| NobodyCam | w00t!!!1 | 18:04 |
| *** dwalleck_ has joined #openstack-ironic | 18:04 | |
| *** romcheg has joined #openstack-ironic | 18:04 | |
| devananda | let the flood gates open, heh | 18:05 |
| *** dwalleck has quit IRC | 18:06 | |
| Shrews | I'm bored with Juno already. Can't we just move on to the Kardashian release already? | 18:07 |
| *** derekh has joined #openstack-ironic | 18:07 | |
| devananda | :p | 18:08 |
| devananda | Shrews: just kidding. really i want you to work on backporting all of ironic to essex. | 18:08 |
| devananda | Shrews: can you do that by friday? kthx :) | 18:09 |
| devananda | </kidding> | 18:09 |
| Shrews | oh, sure. no problem. | 18:09 |
| Shrews | in related news, mordred, i quit | 18:09 |
| Shrews | :) | 18:09 |
| * NobodyCam looks at the 4/1 date on his cal and just shakes his head :-p | 18:10 | |
| jroll | is there any known reason that one out of every 2-3 requests would fail with "No valid host was found. Reason: No conductor service registered which supports driver teeth"? | 18:11 |
| jroll | (given that I have a conductor running with the teeth driver loaded) | 18:11 |
| jroll | maybe a conductor did not unregister properly? | 18:12 |
| devananda | jroll: neat. um | 18:12 |
| * mordred throws an angry wet cat at Shrews head | 18:12 | |
| devananda | jroll: look at the log for said conductor | 18:12 |
| devananda | jroll: also run "ironic driver-list" to see what drivers are being advertised | 18:13 |
| jroll | devananda: I don't see anything in the conductor logs around the time of failure | 18:13 |
| devananda | jroll: it could be before that | 18:13 |
| devananda | jroll: that error comes from API side when it tries to map a requested node to the hash ring, and can't find a conductor that handles that driver | 18:14 |
| devananda | jroll: eg, the conductorw ith 'teeth' driver isn't updating its heartbeat | 18:14 |
| jroll | devananda: right | 18:15 |
| jroll | devananda: I have to run off for now, but that points me in the right direction. thanks | 18:15 |
| * NobodyCam makes a bagel | 18:16 | |
| *** dwalleck_ has quit IRC | 18:24 | |
| NobodyCam | devananda: +2 the parted patches. | 18:32 |
| devananda | mmmm, macaroni .... | 18:45 |
| NobodyCam | :) | 18:46 |
| Shrews | Did I see someone post something about a 'flavor_get' attribute error earlier? Could have swore I did, but not finding it in any srollback | 18:49 |
| devananda | Shrews: yea, i posted it. | 18:49 |
| devananda | Shrews: problem was old .pyc files in nova/nova/virt/ironic/ | 18:49 |
| Shrews | devananda: i shouldn't be seeing that in a freshly cloned nova branch | 18:51 |
| Shrews | which is what adam_g's change is doing to run the unit tests | 18:52 |
| adam_g | Shrews, hows that working for you? | 18:53 |
| Shrews | adam_g: it's not | 18:53 |
| Shrews | AttributeError: <nova.virt.fake.FakeVirtAPI object at 0x7be64d0> does not have the attribute 'flavor_get' | 18:53 |
| adam_g | Shrews, did you pull both patches? | 18:53 |
| Shrews | adam_g: oh, there are 2? | 18:54 |
| * Shrews looks | 18:54 | |
| adam_g | https://review.openstack.org/#/c/84033/ | 18:54 |
| adam_g | https://review.openstack.org/#/c/84043/ | 18:54 |
| adam_g | the unittests needed updating to account for that flavor_get change | 18:54 |
| adam_g | as well as other things we've changed in driver | 18:54 |
| Shrews | adam_g: ok thx. retrying | 18:54 |
| Shrews | adam_g: perhaps those changes should be combined | 18:55 |
| NobodyCam | devananda: added TestingTranslations to the whiteBoard | 18:55 |
| NobodyCam | (etherpads) | 18:55 |
| adam_g | Shrews, they could. i was expecting to make some tweaks to the nova-tests.sh script separate from the tests tho | 18:56 |
| adam_g | may not be needed tho | 18:56 |
| Shrews | adam_g: ok, so that works | 18:59 |
| adam_g | Shrews, cool | 18:59 |
| devananda | i'm gonna be semi-afk for the rest of the day | 19:00 |
| devananda | around for the release and meetings | 19:00 |
| devananda | but have some offline meetings as well | 19:00 |
| Shrews | adam_g: any reason you don't invoke the tests through tox? | 19:00 |
| *** jistr has joined #openstack-ironic | 19:01 | |
| NobodyCam | have a good rest of the day devananda :) | 19:01 |
| adam_g | Shrews, AIUI tox commands will execute from the top level of ironic tree. we need to execute nova's setup.py from within its tree so testr only loads nova code + tests. | 19:02 |
| adam_g | Shrews, also i'm expecting we will need to adjust as that script a bit for it to fit into the unit test gate jobs, in terms of generating and storing its storing its subunit results, etc.in a place that the jenkins job expect | 19:03 |
| NobodyCam | adam_g: would tox -e py27 -- ironic.nova.tests work? | 19:03 |
| Shrews | adam_g: or 'cd $NOVA_REPO && tox -epy27' | 19:04 |
| Shrews | tox reads the tox.ini from current dir, iirc | 19:04 |
| adam_g | NobodyCam, i dont think it would. thats what i originally wanted--but testr will go and load *all* of the tests in ironic, which cause an import of both ironic.* and nova.*, and there are issues there WRT oslo config resources being stomped on | 19:04 |
| NobodyCam | ahh | 19:05 |
| adam_g | Shrews, oh--your original question was about why nova-tests.sh invokes setup.py direclty instead of tox? | 19:05 |
| Shrews | adam_g: correct | 19:06 |
| adam_g | Shrews, yeah-i think that would probably work too | 19:06 |
| Shrews | it would abstract the test runner away | 19:07 |
| adam_g | Shrews, are we still able to pass positional arguments down to setup.py/testr? | 19:07 |
| Shrews | adam_g: yes | 19:07 |
| adam_g | im not sure we want to run the entire nova suite or just the ironic tests | 19:07 |
| Shrews | adam_g: tox -epy27 -- my_arguments | 19:07 |
| * adam_g gives it a shot | 19:08 | |
| *** blamar has joined #openstack-ironic | 19:08 | |
| Shrews | adam_g: examples in the quick start: http://docs.openstack.org/developer/ironic/dev/dev-quickstart.html# | 19:08 |
| *** romcheg has quit IRC | 19:08 | |
| Shrews | adam_g: you can select individual tests, or pass testr args | 19:09 |
| adam_g | Shrews, right | 19:09 |
| *** jbjohnso_ has quit IRC | 19:10 | |
| *** eguz has joined #openstack-ironic | 19:11 | |
| adam_g | Shrews, so that ends up creating a nested venv | 19:11 |
| Shrews | adam_g: it creates a venv that tox will use to run the tests. that should be ok | 19:11 |
| Shrews | i'd rather see that than have my personal dev env get mucked around with | 19:12 |
| adam_g | Shrews, how so? everything is contained in the .tox/nova-tests/ venv | 19:12 |
| adam_g | as is now, at least | 19:12 |
| Shrews | adam_g: i think you could eliminate the whole in_env stuff | 19:12 |
| Shrews | adam_g: sorry, yeah, it's not doing it now (mucking around with your in_venv environment). | 19:14 |
| Shrews | adam_g: hrm, i see your point about the nesting though | 19:15 |
| *** eghobo has quit IRC | 19:15 | |
| adam_g | Shrews, also, relying on nova's tox means the entire execution happens outside of ironic's tox--you'd run tools/nova-tests.sh only. | 19:15 |
| NobodyCam | brb... | 19:15 |
| Shrews | adam_g: i was thinking that nova-tests.sh would call tox instead of testr, but now that i'm thinking about it, i'm seeing more disadvantage than advantage to doing that | 19:16 |
| adam_g | Shrews, this'll require a bit of tweaking here and in infra to have this running as part of the gate pipeline. making it executable from ironic as 'tox -e nova-tests' will probably cause the least ammount of friction getting it in there | 19:17 |
| Shrews | yeah, never intended to change that part | 19:17 |
| Shrews | adam_g: leave it as-is for now, is my suggestion | 19:17 |
| * Shrews needs to stimulate his brain with chai tea | 19:18 | |
| JayF | devananda: +1 GF Macaroni and Cheese :P | 19:19 |
| JayF | devananda: if you're ever back in the Bay Area (idk where you live), make it down to Jack's in San Bruno for GF Bacon Mac. Mmmm | 19:19 |
| NobodyCam | JayF: ummmmmm.... yummy | 19:26 |
| JayF | they have regular Bacon Mac. So don't worry. You can have wheat :P | 19:27 |
| NobodyCam | heheheh :) | 19:27 |
| comstud | argh | 19:29 |
| NobodyCam | :( ? | 19:29 |
| comstud | Can someone answer a question for me regarding TaskManager() __init__() ? | 19:29 |
| comstud | I guess I can search the code, but.. | 19:29 |
| comstud | 137 :param node_ids: A list of ids or uuids of nodes to lock. | 19:29 |
| devananda | JayF: seattle. I get t othe bay fairly often tho | 19:29 |
| comstud | Why can that be either? | 19:29 |
| comstud | in here is: | 19:29 |
| comstud | 167 node = self.dbapi.get_node(id) | 19:29 |
| comstud | which will only search by uuid | 19:30 |
| devananda | comstud: there's some magic in the dbapi | 19:30 |
| devananda | comstud: it actually takes either one | 19:30 |
| devananda | comstud: optimization for databases. if you already know the 'id', it's faster to look up that way | 19:30 |
| NobodyCam | comstud: id are never expoded to users | 19:30 |
| NobodyCam | but are used internally | 19:30 |
| comstud | yeah I see | 19:30 |
| comstud | but that's very inconsistent | 19:30 |
| devananda | comstud: internally or w/ othe rprojects? | 19:31 |
| comstud | and the object version of Node only has a 'get_by_uuid' method | 19:31 |
| comstud | which I guess should just be 'get' | 19:31 |
| comstud | since it can take either. | 19:31 |
| *** jbjohnso_ has joined #openstack-ironic | 19:31 | |
| devananda | hmm. yes. | 19:31 |
| NobodyCam | :-p | 19:31 |
| comstud | I'm just speaking in general | 19:31 |
| devananda | name is inconsistent there | 19:31 |
| comstud | I'm fixing up a bunch of object stuff and ran into this | 19:31 |
| *** max_lobur1 has quit IRC | 19:31 | |
| NobodyCam | ++ | 19:31 |
| *** max_lobur has joined #openstack-ironic | 19:31 | |
| comstud | and I generally hate "well this variable can be a list of x or a list of y" | 19:31 |
| comstud | but I can deal with it, I guess | 19:32 |
| NobodyCam | :) | 19:32 |
| devananda | comstud: you'll hate that TaskManagers can lock "a node" or "a list of nodes" then | 19:32 |
| NobodyCam | :-p shhhhh... | 19:33 |
| comstud | I can deal better with a singleton vs a list | 19:33 |
| devananda | heh | 19:33 |
| devananda | k | 19:33 |
| comstud | assuming they are the same type | 19:33 |
| comstud | i mean, list member is same type as singleton | 19:33 |
| comstud | heh | 19:33 |
| comstud | but anyway :) | 19:33 |
| comstud | Ok. | 19:33 |
| comstud | also interesting is.. | 19:34 |
| comstud | Node.get_by_uuid calls dbapi.get_node, which returns an object | 19:34 |
| comstud | and Node.get_by_uuid re-turns it into an object itself | 19:34 |
| comstud | So, here's what I want to do.. but I don't know if you agree.. | 19:34 |
| comstud | I'd like objects.Node.get() be the interface you use.. | 19:35 |
| comstud | internally it'll use the dbapi, but the dbapi itself doesn't need to form the object then | 19:35 |
| Shrews | adam_g: so, i left my observations on that review. In case I wasn't clear, only point #3 is a stickler for me. | 19:36 |
| comstud | The related problem is your objectify() decorator actually should be passing a context to _from_db_object() | 19:36 |
| Shrews | s/wasn't clear/wasn't clear in my comments/ | 19:36 |
| comstud | yet noone of the dbapi calls have context on them | 19:36 |
| comstud | none | 19:36 |
| comstud | so to clean this up, I either: add context to all of the dbapi methods, or remove the objectify decorator | 19:37 |
| comstud | I'd rather do the latter.. and if the objects code is the interface you use, it is what makes sure to return the object version itself | 19:37 |
| NobodyCam | comstud: what I'm tring to think about is would having context in the dbapi be useful downthe road | 19:39 |
| comstud | I find it kind of pointless and at the wrong layer | 19:39 |
| comstud | the object calls will all have context and can do things with it there | 19:40 |
| comstud | whether it be telling the db layer to apply certain filters or what | 19:40 |
| NobodyCam | very true | 19:40 |
| comstud | We have them in nova on the dbapi calls, but a lot of it is kinda dumb... we require admin context for certain dbapi calls, yet we have some policy checks for them also | 19:41 |
| comstud | (up higher) | 19:41 |
| devananda | comstud: latter | 19:42 |
| NobodyCam | comstud: while at this point I am not 100% there I am also NOT opposed to your suggestion | 19:42 |
| openstackgerrit | A change was merged to openstack/ironic: Fix spelling error in conductor/manager https://review.openstack.org/84472 | 19:42 |
| comstud | well, I'll be throwing up some patches... it'll be easier to see there | 19:42 |
| devananda | comstud: afaict we shouldnt ever need context on the dbapi calls | 19:43 |
| comstud | and we can debate in the reviews | 19:43 |
| NobodyCam | oh devananda while I have you. | 19:43 |
| devananda | comstud: noav uses that for permanent delete, view deleted stuff, etc. | 19:43 |
| comstud | devananda: that's what I would hope | 19:43 |
| devananda | comstud: and ironic doesn't keep history like that | 19:43 |
| devananda | comstud: access should be checked at the API tier only, IMO | 19:43 |
| comstud | nova also filters by project for getting lists of instances | 19:43 |
| comstud | and so forth | 19:43 |
| comstud | but | 19:43 |
| devananda | ah | 19:43 |
| NobodyCam | is htere a list of patches that I can ping the owners of to re-submitt now that J is offically open? | 19:43 |
| comstud | some of that now is actually passed in to dbapi as 'filter this project' | 19:43 |
| comstud | which I prefer | 19:43 |
| comstud | at a higher layer | 19:44 |
| comstud | keep the db layer a little more stupid | 19:44 |
| devananda | we could eventually have projects in ironic. but yes ^ | 19:44 |
| comstud | cools | 19:44 |
| devananda | NobodyCam: no. but go down this page https://review.openstack.org/#/q/status:abandoned+project:openstack/ironic,n,z | 19:44 |
| NobodyCam | devananda: quick after noon walkies then I'll ping the -2'd patch owners :) | 19:46 |
| *** florentflament has quit IRC | 19:51 | |
| comstud | devananda: hm, yeah, I think interestingly... | 19:56 |
| comstud | you could currently pass an id like '1' to ironic API... | 19:56 |
| comstud | and it'll happily fetch that node | 19:56 |
| NobodyCam | comstud: it should... | 19:56 |
| NobodyCam | it used to | 19:56 |
| NobodyCam | lol | 19:56 |
| NobodyCam | just dont tell anyone | 19:57 |
| comstud | You think it *should*? | 19:57 |
| comstud | :) | 19:57 |
| comstud | Hehe | 19:57 |
| devananda | no | 19:57 |
| NobodyCam | I have not tested that in quite a while | 19:57 |
| devananda | it should not | 19:57 |
| comstud | ok | 19:57 |
| devananda | i believe we fixed that | 19:57 |
| NobodyCam | oh | 19:57 |
| devananda | GET /v1/nodes/1 | 19:57 |
| comstud | I don't see how it can be fixed | 19:57 |
| NobodyCam | we may have | 19:57 |
| devananda | this ^ should fail | 19:57 |
| comstud | since you call Node.get_by_uuid() which calsl dbapi.get_node() | 19:57 |
| devananda | comstud: enforcing in teh API layer that a UUID is received | 19:57 |
| devananda | type cehcking in wsme | 19:57 |
| NobodyCam | ahh that is right... | 19:58 |
| comstud | yeah | 19:59 |
| comstud | ok. | 19:59 |
| devananda | https://launchpad.net/ironic/icehouse/icehouse-rc1 | 20:00 |
| devananda | Released: 1 minute ago | 20:00 |
| devananda | heh | 20:00 |
| *** romcheg has joined #openstack-ironic | 20:00 | |
| JayF | \o/ | 20:01 |
| NobodyCam | w00t \o/ | 20:01 |
| NobodyCam | ┏(-_-)┛┗(-_- )┓┗(-_-)┛┏(-_-)┓ | 20:02 |
| *** derekh has quit IRC | 20:05 | |
| *** jbjohnso_ has quit IRC | 20:07 | |
| *** mdurnosvistov1 has left #openstack-ironic | 20:07 | |
| uberj | devananda: does that mean ironic will be in icehouse? | 20:11 |
| dividehex | I want to know also ^^ | 20:12 |
| JayF | AIUI, Ironic is still incubating, it's not graduating in Icehouse. We | 20:13 |
| JayF | *We're just following the release process in good faith and so if someone chooses to use Ironic with Icehouse, it's stablized for that purpose. | 20:13 |
| uberj | thanks, JayF | 20:14 |
| devananda | JayF: that's correct | 20:18 |
| *** romcheg has quit IRC | 20:19 | |
| *** florentflament has joined #openstack-ironic | 20:24 | |
| *** eguz has quit IRC | 20:27 | |
| *** mrda_away is now known as mrda | 20:27 | |
| mrda | morning all | 20:27 |
| *** eghobo has joined #openstack-ironic | 20:28 | |
| NobodyCam | morning mrda | 20:29 |
| mrda | Hey NobodyCam | 20:29 |
| adam_g | devananda, https://review.openstack.org/#/c/84543/ | 20:33 |
| *** romcheg has joined #openstack-ironic | 20:35 | |
| *** jistr has quit IRC | 20:37 | |
| openstackgerrit | Adam Gandelman proposed a change to openstack/ironic: Add tox target + script for nova driver tests https://review.openstack.org/84033 | 20:39 |
| NobodyCam | devananda: there is a list on the bottom of : https://etherpad.openstack.org/p/IronicWhiteBoard (Patch requested for re-submission now that Juno is Open) could you remove your -2's when you have a free minute | 20:40 |
| openstackgerrit | Ruby Loo proposed a change to openstack/python-ironicclient: Add main developer page https://review.openstack.org/84545 | 20:44 |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Handling validation in conductor consistently https://review.openstack.org/84374 | 20:46 |
| NobodyCam | lol fyi: Atlanta has very few rv parks :-p | 20:48 |
| devananda | NobodyCam: ah. later today or tonight. no time right now | 20:49 |
| NobodyCam | np list is there for you :) | 20:50 |
| NobodyCam | that sould be all of them | 20:50 |
| *** florentflament has quit IRC | 20:55 | |
| *** jdob has quit IRC | 20:57 | |
| *** blamar has quit IRC | 20:58 | |
| *** blamar has joined #openstack-ironic | 21:00 | |
| openstackgerrit | Ruby Loo proposed a change to openstack/python-ironicclient: Add main developer page https://review.openstack.org/84545 | 21:04 |
| *** blamar has quit IRC | 21:04 | |
| *** blamar has joined #openstack-ironic | 21:06 | |
| *** linggao has quit IRC | 21:11 | |
| *** blamar has quit IRC | 21:13 | |
| *** blamar has joined #openstack-ironic | 21:14 | |
| comstud | man | 21:14 |
| comstud | a number of modules are doing node.get() | 21:14 |
| comstud | when it should be the fricken node object | 21:14 |
| comstud | the tests are testing passing dicts | 21:16 |
| comstud | and not the objects. | 21:16 |
| *** blamar has quit IRC | 21:18 | |
| *** blamar has joined #openstack-ironic | 21:18 | |
| mrda | comstud: sounds like an opportunity :) | 21:20 |
| *** blamar_ has joined #openstack-ironic | 21:20 | |
| comstud | yeah, well, it has to be fixed as a part of what i'm doing | 21:21 |
| NobodyCam | comstud: how large is the patch getting? | 21:21 |
| comstud | well, there's going to be quite a number of patches | 21:21 |
| comstud | I'm only fixing get_node* and related things right now | 21:21 |
| NobodyCam | :) | 21:21 |
| comstud | it's not that bad | 21:22 |
| comstud | so far, +218, -138 | 21:22 |
| comstud | but i still ahve 90 test failures | 21:22 |
| comstud | :) | 21:22 |
| NobodyCam | oh not bad at all,, | 21:22 |
| NobodyCam | :) | 21:22 |
| comstud | i think there's something bad going on with the ports API tests also | 21:23 |
| *** blamar has quit IRC | 21:23 | |
| *** blamar has joined #openstack-ironic | 21:25 | |
| *** blamar_ has quit IRC | 21:25 | |
| NobodyCam | :-/ | 21:25 |
| devananda | hmm | 21:25 |
| comstud | I only happened to uncover these issues because I created a 'get' method on Node | 21:25 |
| comstud | heh | 21:25 |
| devananda | so we'll need an oslo liason -- https://wiki.openstack.org/wiki/Oslo/ProjectLiaisons | 21:25 |
| comstud | which overrides the dict.get | 21:25 |
| comstud | syntax | 21:26 |
| comstud | which no one should do with objects | 21:26 |
| comstud | just access the field, mann | 21:26 |
| comstud | File "ironic/drivers/modules/pxe.py", line 94, in _parse_driver_info | 21:26 |
| comstud | info = node.get('driver_info', {}) | 21:26 |
| comstud | as an example | 21:26 |
| comstud | and of course, when I fix that... | 21:27 |
| comstud | I find the tests are passing a dict. | 21:27 |
| comstud | and not the Node object | 21:27 |
| comstud | :) | 21:27 |
| *** dwalleck has joined #openstack-ironic | 21:27 | |
| comstud | actually, turns out the pxe module tests are correct.. but the seamicro and ssh ones were not | 21:30 |
| comstud | well anyway, fun | 21:31 |
| NobodyCam | rloo: good catch with bug 13000 | 21:32 |
| NobodyCam | 1300990 even | 21:32 |
| NobodyCam | :-p | 21:32 |
| rloo | NobodyCam: Once that gets in, wait til you see the docn I'm going to write (or not) ;) | 21:34 |
| comstud | is ironic/common/context used? | 21:38 |
| comstud | the tests, of course, mostly use ironic/opestack/common/context | 21:38 |
| NobodyCam | rloo: looking at conf.py on 84545 | 21:41 |
| NobodyCam | using ironic as reff. i see this: https://github.com/openstack/ironic/blob/master/doc/source/conf.py#L17 | 21:42 |
| NobodyCam | see any reason to add that? | 21:42 |
| rloo | NobodyCam: the wsme_protocols? Do you know what that is used for? | 21:42 |
| rloo | NobodyCam: I don't think ironicclient uses wsme anything. | 21:43 |
| rloo | NobodyCam: so I don't think it is needed. I figure we'll find out if there are errors, but I didn't see any related to wsme ;) | 21:44 |
| NobodyCam | I saw it on this review: https://github.com/openstack/ironic/commit/30206795b71d9b53c586385fe235d588bfe70a44 | 21:44 |
| NobodyCam | just need to test the patch here :) | 21:45 |
| rloo | NobodyCam: the client doesn't use pecan or wsme. I was wondering what httpdomain was for but it looks like used with pecanwsme. | 21:47 |
| NobodyCam | devananda: do you happen to recall adding this https://github.com/openstack/ironic/blob/master/doc/source/conf.py#L17 | 21:47 |
| devananda | nope | 21:48 |
| NobodyCam | :-p | 21:48 |
| rloo | NobodyCam: wrt sphinxcontrib.httpdomain, ironicclient doesn't need it cuz no RESTful api there: http://pythonhosted.org/sphinxcontrib-httpdomain/ | 21:49 |
| *** matty_dubs is now known as matty_dubs|gone | 21:51 | |
| *** dguerri_ has joined #openstack-ironic | 21:53 | |
| *** mtaylor has joined #openstack-ironic | 21:53 | |
| *** dguerri has quit IRC | 21:54 | |
| *** openstackgerrit has quit IRC | 21:54 | |
| *** mordred has quit IRC | 21:54 | |
| *** romcheg has quit IRC | 21:56 | |
| NobodyCam | rloo: lgtm | 22:00 |
| NobodyCam | brb | 22:02 |
| comstud | Ok | 22:03 |
| comstud | I found a case where the API will take a node ID | 22:03 |
| comstud | the API Port declaration has: | 22:04 |
| comstud | 80 node_uuid = wsme.wsproperty(types.uuid, _get_node_uuid, _set_node_uuid, 81 mandatory=True) | 22:04 |
| comstud | and _get_node_uuid doesn't care if it's an ID or a UUID | 22:04 |
| comstud | and there seems to be some tests passing an integer somehow | 22:04 |
| rloo | NobodyCam: thx! | 22:06 |
| * NobodyCam thinks comstud is digging up all our skeletons | 22:06 | |
| NobodyCam | hehehe | 22:06 |
| NobodyCam | :-p | 22:06 |
| comstud | yeah | 22:06 |
| comstud | hm. | 22:07 |
| comstud | POST: /v1/ports {'node_uuid': '1be26c0b-03f2-4d2e-ae87-c02d7f33c123', 'uuid': '1be26c0b-03f2-4d2e-ae87-c02d7f33c781', 'address': '52:54:00:cf:2d:31', 'extra': {'foo': 123}} | 22:07 |
| comstud | that is somehow causing a node lookup of 123 | 22:07 |
| rloo | comstud: thx for finding that; i think it is a bug ;) | 22:07 |
| NobodyCam | that iindeed sounds like a bug | 22:08 |
| comstud | http://paste.openstack.org/show/74776/ | 22:09 |
| comstud | that last line | 22:09 |
| * NobodyCam shutters but clicks any way | 22:10 | |
| comstud | i'm not sure where 'node_id' is coming from | 22:10 |
| comstud | is extra supposed to be a dict of key to node ids? | 22:11 |
| NobodyCam | no | 22:11 |
| mrda | comstud: better to find those bugs now :) | 22:13 |
| comstud | ah i see | 22:14 |
| comstud | create db record, node_id is set | 22:14 |
| comstud | and that's used to create Port() | 22:15 |
| comstud | i wonder if: setattr(self, 'node_uuid', kwargs.get('node_id')) | 22:15 |
| comstud | is supposed to be node_uuid | 22:15 |
| comstud | not node_id | 22:15 |
| comstud | although there's no node_uuid i guess in the DB | 22:15 |
| comstud | or is there | 22:15 |
| comstud | hm | 22:15 |
| comstud | no | 22:16 |
| comstud | not sure how to fix this, going to leave it supporting both in my patch | 22:16 |
| comstud | and file a bug | 22:16 |
| devananda | comstud: i love that you're digging up all our skeletons :) | 22:17 |
| NobodyCam | :) | 22:17 |
| devananda | gotta run... bbl | 22:17 |
| comstud | hehe | 22:17 |
| comstud | I'm not sure I'm loving it | 22:18 |
| comstud | but at least someone is | 22:18 |
| comstud | :) | 22:18 |
| NobodyCam | comstud: but we appreciate so much :) | 22:18 |
| *** jgrimm has quit IRC | 22:23 | |
| comstud | https://bugs.launchpad.net/ironic/+bug/1301046 | 22:25 |
| *** radsy has joined #openstack-ironic | 22:29 | |
| NobodyCam | Woo Hoo get to spend the night in Roswell New Mexico on the way to Atlanta... :-P | 22:32 |
| *** lucas-afk has quit IRC | 22:41 | |
| NobodyCam | anteaya: wow that was quick... | 22:41 |
| comstud | we'll see how this looks in gerrit.. | 22:43 |
| comstud | bot die? | 22:43 |
| comstud | https://review.openstack.org/#/c/84573/ | 22:44 |
| NobodyCam | may just be slow | 22:44 |
| NobodyCam | openstack user still in channel | 22:44 |
| anteaya | hello | 22:45 |
| NobodyCam | hi anteaya :) | 22:45 |
| anteaya | I dislike spending the night in roswell new mexico | 22:45 |
| anteaya | I spent one night there and I wouldn't do so again voluntarily | 22:45 |
| NobodyCam | anteaya: you've done it before? | 22:46 |
| NobodyCam | lol :) | 22:46 |
| anteaya | dreampt that my father was being coorerced by aliens and was trying to coorce me | 22:46 |
| NobodyCam | may I or SHOULD I ask why | 22:46 |
| anteaya | so I got up and cleaned the energy of the room for the next few hours | 22:46 |
| anteaya | that is why | 22:46 |
| anteaya | never had that kind of dream before or since | 22:46 |
| NobodyCam | ahh | 22:47 |
| NobodyCam | :) | 22:47 |
| anteaya | it was when I drove with them down to tucson | 22:47 |
| anteaya | so what was I quick about? | 22:47 |
| anteaya | I like new mexico | 22:48 |
| anteaya | the white sands are nice | 22:48 |
| NobodyCam | the confirmed replay to the Ironic PTL email | 22:48 |
| anteaya | ah | 22:48 |
| anteaya | I was watching | 22:48 |
| NobodyCam | hehehh :) | 22:48 |
| NobodyCam | when is voting? | 22:48 |
| anteaya | starts friday | 22:48 |
| NobodyCam | :) | 22:48 |
| anteaya | when I open the elections | 22:48 |
| * JayF looks for said ironic ptl email | 22:48 | |
| anteaya | http://lists.openstack.org/pipermail/openstack-dev/2014-April/031599.html | 22:49 |
| anteaya | JayF: ^ | 22:49 |
| NobodyCam | :) | 22:49 |
| JayF | yeah my mail client is just a little slow | 22:50 |
| JayF | I think the IMAP server my host uses doesn't always properly notify unread count when server side filtering happeng | 22:50 |
| JayF | *happens | 22:50 |
| *** max_lobur has quit IRC | 22:50 | |
| comstud | probably helps if I git add the file I added | 23:02 |
| comstud | hrm | 23:02 |
| NobodyCam | anteaya: thank you :) | 23:03 |
| * NobodyCam need to do a quick walkies.. bbiafm | 23:05 | |
| NobodyCam | comstud: ya bot must have died :-p | 23:10 |
| comstud | yeah | 23:11 |
| comstud | i think i'm done for now | 23:11 |
| NobodyCam | ya getting to EOD | 23:11 |
| comstud | i need a break at least | 23:11 |
| comstud | :) | 23:11 |
| comstud | bbl | 23:11 |
| comstud | take care | 23:11 |
| * comstud & | 23:12 | |
| NobodyCam | TY comstud :) | 23:12 |
| anteaya | np | 23:15 |
| *** derekh has joined #openstack-ironic | 23:23 | |
| *** stevehuang has quit IRC | 23:36 | |
| * NobodyCam calls it a day.. I'll be back in a while to check on things.. Night all | 23:42 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!