| *** eguz has quit IRC | 00:02 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Adding oslo testing to improve debugging https://review.openstack.org/84212 | 00:03 |
|---|---|---|
| *** blamar has quit IRC | 00:06 | |
| *** blamar has joined #openstack-ironic | 00:09 | |
| *** blamar has quit IRC | 00:26 | |
| *** matsuhashi has joined #openstack-ironic | 00:28 | |
| *** blamar has joined #openstack-ironic | 00:32 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Adding oslotest to improve debugging https://review.openstack.org/84212 | 00:33 |
| *** Haomeng|2 has quit IRC | 00:36 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Adding oslotest to improve debugging https://review.openstack.org/84212 | 00:41 |
| *** harlowja is now known as harlowja_away | 01:01 | |
| *** matsuhashi has quit IRC | 01:05 | |
| *** matsuhashi has joined #openstack-ironic | 01:06 | |
| *** linggao has quit IRC | 01:16 | |
| *** nosnos has joined #openstack-ironic | 01:29 | |
| *** ilives has joined #openstack-ironic | 01:41 | |
| *** ilives has quit IRC | 01:48 | |
| *** ilives has joined #openstack-ironic | 01:49 | |
| *** rloo has quit IRC | 02:45 | |
| *** matsuhashi has quit IRC | 03:05 | |
| lifeless | dammit - devananda https://jenkins04.openstack.org/job/check-tripleo-ironic-undercloud-precise/152/console - thats my 30s delay patch | 03:07 |
| lifeless | devananda: haven't dug in yet | 03:07 |
| lifeless | http://logs.openstack.org/61/84361/1/check-tripleo/check-tripleo-ironic-undercloud-precise/81791d5/ | 03:08 |
| NobodyCam | lifeless: power state is NOSTATE is ssh working? | 03:12 |
| NobodyCam | peredioc task should set that to off | 03:13 |
| russell_h | lifeless: is there a long term fix for this? | 03:15 |
| lifeless | russell_h: for what ? | 03:19 |
| lifeless | NobodyCam: the seed booted, so we know the creds in teh testenv json are valid | 03:20 |
| NobodyCam | lifeless: I see this in the conductor log. http://paste.openstack.org/show/qpJlh1VMtwwiOv2RKlGu/ | 03:21 |
| *** nosnos has quit IRC | 03:23 | |
| *** rameshg87 has joined #openstack-ironic | 03:29 | |
| *** coolsvap has joined #openstack-ironic | 03:32 | |
| russell_h | lifeless: for this neutron issue | 03:33 |
| lifeless | NobodyCam: have a look in syslog, see if the node DHCPd | 03:33 |
| russell_h | lifeless: comments like | 03:35 |
| russell_h | "Horrible hack to mitigate bug 1300589" deserve some sort of long term plan IMO | 03:35 |
| NobodyCam | Apr 3 02:14:52 ubuntu dnsmasq-dhcp[5126]: DHCPDISCOVER(tap1e536302-66) 52:54:00:ab:d5:3e | 03:35 |
| NobodyCam | Apr 3 02:14:52 ubuntu dnsmasq-dhcp[5126]: DHCPOFFER(tap1e536302-66) 192.0.2.2 52:54:00:ab:d5:3e | 03:35 |
| lifeless | russell_h: oh thats easy | 03:36 |
| russell_h | I was given to understand that Nova has a similar bug and they're adding some sort of callback thing | 03:36 |
| lifeless | russell_h: generalise neutron's callback-to-nova to use a webhook or arbitrary url instead, so that you can say 'tell ironic at XXX when this complete' | 03:36 |
| russell_h | gotcha | 03:36 |
| lifeless | russell_h: they added it but its not generic. | 03:36 |
| russell_h | lifeless: is there an issue open to genericize it? | 03:37 |
| russell_h | basically, I'd like to see something like "TODO: refactor to block for neutron callback once bug xxxxx is resolved" | 03:38 |
| lifeless | russell_h: I think that should be a joint ironic / neutron blueprint. | 03:40 |
| lifeless | NobodyCam: that looks like it DHCPd | 03:51 |
| lifeless | NobodyCam: so either the ramdisk failed | 03:51 |
| NobodyCam | ya | 03:51 |
| lifeless | NobodyCam: or my fix wasn't aggressive enough | 03:51 |
| NobodyCam | and I would expect to see https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/ssh.py#L140-L142 | 03:51 |
| lifeless | NobodyCam: can you check the timestamps | 03:51 |
| NobodyCam | if ssh failed | 03:51 |
| lifeless | well it dhcpd so that means ssh power on worked | 03:52 |
| NobodyCam | yep... | 03:52 |
| lifeless | if we're not logging enough, perhaps ironics log default should be different :) | 03:52 |
| NobodyCam | we may want to bump up a couple of levels until things are working smoothly :) | 03:53 |
| lifeless | NobodyCam: well, think of it as a production exercise | 03:53 |
| lifeless | NobodyCam: this is exactly what debugging production problems is like; so if we're not logging enough, operators will find it problematic too. | 03:54 |
| *** eghobo has joined #openstack-ironic | 03:54 | |
| lifeless | by the time you realise you need logs, they've already been written and its too late | 03:54 |
| NobodyCam | lifeless: yes!!! | 03:54 |
| lifeless | so - I'm fairly sure tripleo is running ironics default | 03:54 |
| lifeless | perhaps ironic should change its default :) | 03:54 |
| russell_h | I took that to the extreme once | 03:54 |
| russell_h | and logged actually everything | 03:54 |
| lifeless | russell_h: so we did that with oopstools | 03:55 |
| lifeless | russell_h: we logged *everything* for each request, into a bson struct | 03:55 |
| lifeless | russell_h: and if the request failed, put that in a DB | 03:55 |
| lifeless | russell_h: if the request worked, we threw it away and just logged the top level HTTPD style line | 03:55 |
| lifeless | [by put ina DB, I mean tossed it at rabbit which tossed it at the DB] | 03:55 |
| russell_h | right | 03:55 |
| russell_h | thats basically what we were doing, but without the clever throwing away bit | 03:56 |
| russell_h | json log messages | 03:56 |
| NobodyCam | ahh you didn't pay for splunk | 03:56 |
| russell_h | hundreds per request, even for simple GETs | 03:56 |
| lifeless | we used bson because we handle binary data in launchpad | 03:56 |
| lifeless | and we wrote a display console to show the timelines | 03:56 |
| lifeless | debugging slow SQL etc was ---trivial--- with this | 03:57 |
| russell_h | NobodyCam: we could have bought a lot of splunk for what we spent on our graylog cluster | 03:57 |
| NobodyCam | :) | 03:57 |
| russell_h | actually probably not... | 03:57 |
| russell_h | you still have to buy your own gear for splunk right? | 03:57 |
| russell_h | at least the license I used once was that way | 03:57 |
| NobodyCam | you did when I last used them | 03:57 |
| NobodyCam | by the GB loged | 03:58 |
| russell_h | we were doing gigabits of logs | 03:58 |
| russell_h | at pretty trivial load | 03:58 |
| russell_h | then the ops guys made us turn down the logging :( | 03:58 |
| NobodyCam | lol | 03:58 |
| russell_h | they were spending more time scaling graylog than the actual software | 03:58 |
| russell_h | but debugging was glorious | 03:59 |
| lifeless | so the throwaway strat is important | 03:59 |
| russell_h | yeah, I like that | 03:59 |
| lifeless | e.g. AIUI google throwaway a large % of log messages | 03:59 |
| lifeless | just randomly | 03:59 |
| lifeless | the know enough will get through to show errors up | 03:59 |
| *** harlowja_away is now known as harlowja | 04:00 | |
| *** nosnos has joined #openstack-ironic | 04:09 | |
| *** matsuhashi has joined #openstack-ironic | 04:15 | |
| NobodyCam | lifeless: ssh.py 5 Loging lines 4 debug and one error. I'd like to add a few more at other levels, but it is late for me here | 04:18 |
| NobodyCam | I can do it first thing in the AM | 04:19 |
| lifeless | NobodyCam: so, I suggest taking over my patch | 04:25 |
| lifeless | and stabbing at it more and more until you can see a) whats wrong and b) fix it | 04:25 |
| lifeless | NobodyCam: then, when it passes, we can split that work out into small individual patches in a stack | 04:25 |
| *** BadCub01 has quit IRC | 04:28 | |
| *** killer_prince has joined #openstack-ironic | 04:39 | |
| *** matsuhashi has quit IRC | 04:40 | |
| *** matsuhashi has joined #openstack-ironic | 04:46 | |
| *** eghobo has quit IRC | 05:20 | |
| *** pradipta_away is now known as pradipta | 05:27 | |
| *** Mikhail_D_ltp has joined #openstack-ironic | 05:27 | |
| *** subah has joined #openstack-ironic | 05:29 | |
| *** harlowja is now known as harlowja_away | 05:32 | |
| *** radsy has quit IRC | 05:38 | |
| lifeless | NobodyCam: so | 05:49 |
| lifeless | R(tap1e536302-66) 52:54:00:ab:d5:3e | 05:49 |
| lifeless | Apr 3 02:14:51 ubuntu dnsmasq-dhcp[5126]: DHCPOFFER(tap1e536302-66) 192.0.2.2 52:54:00:ab:d5:3e | 05:49 |
| lifeless | Apr 3 02:14:52 ubuntu dnsmasq-dhcp[5126]: DHCPDISCOVER(tap1e536302-66) 52:54:00:ab:d5:3e | 05:49 |
| lifeless | Apr 3 02:14:52 ubuntu dnsmasq-dhcp[5126]: DHCPOFFER(tap1e536302-66) 192.0.2.2 52:54:00:ab:d5:3e | 05:49 |
| lifeless | Apr 3 02:14:54 ubuntu dnsmasq-dhcp[5126]: DHCPREQUEST(tap1e536302-66) 192.0.2.2 52:54:00:ab:d5:3e | 05:49 |
| lifeless | bah | 05:49 |
| lifeless | Apr 3 02:14:43 ubuntu dnsmasq-dhcp[5126]: read /var/run/neutron/dhcp/f0317e27-e6e8-4869-bff1-6b8e0ea93572/opts | 05:49 |
| lifeless | then the DHCP | 05:49 |
| lifeless | so that looks like the patch worked - just, only 8 seconds leeway | 05:50 |
| lifeless | the cleanup failure is interesting and needs fixing :) | 05:51 |
| lifeless | NobodyCam: what abbout this? | 05:51 |
| lifeless | (wsme.api): 2014-04-03 02:11:05,105 WARNING Client-side error: Couldn't apply patch '[{'path': '/extra/vif_port_id', 'op': 'remove'}]'. Reason: u'vif_port_id' | 05:51 |
| lifeless | from API log ? | 05:52 |
| lifeless | I can't see the vendor callback in the api log | 05:52 |
| lifeless | maybe we should be getting the console out of the VM ? | 05:53 |
| openstackgerrit | Adam Gandelman proposed a change to openstack/ironic: Add tox target + script for nova driver tests https://review.openstack.org/84033 | 05:55 |
| *** Mikhail_D_ltp has quit IRC | 06:02 | |
| openstackgerrit | Jenkins proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/83956 | 06:07 |
| *** ndipanov has joined #openstack-ironic | 06:32 | |
| *** ndipanov has quit IRC | 06:32 | |
| *** matsuhashi has quit IRC | 06:36 | |
| *** matsuhashi has joined #openstack-ironic | 06:39 | |
| *** ndipanov has joined #openstack-ironic | 06:43 | |
| *** romcheg has joined #openstack-ironic | 06:52 | |
| *** romcheg has quit IRC | 06:54 | |
| *** mrda is now known as mrda_away | 06:57 | |
| *** Haomeng|2 has joined #openstack-ironic | 06:59 | |
| *** zigo has quit IRC | 07:04 | |
| *** zigo has joined #openstack-ironic | 07:08 | |
| *** mikal has quit IRC | 07:10 | |
| *** mikal has joined #openstack-ironic | 07:10 | |
| *** dshulyak has joined #openstack-ironic | 07:11 | |
| *** dshulyak has quit IRC | 07:13 | |
| *** ifarkas has joined #openstack-ironic | 07:22 | |
| *** romcheg has joined #openstack-ironic | 07:29 | |
| *** martyntaylor has quit IRC | 07:43 | |
| *** romcheg has left #openstack-ironic | 07:45 | |
| *** max_lobur has joined #openstack-ironic | 07:49 | |
| rameshg87 | Hi | 07:52 |
| rameshg87 | i had a question related to diskimage-builder | 07:52 |
| rameshg87 | anyone up there ? :-) | 07:52 |
| *** martyntaylor has joined #openstack-ironic | 07:52 | |
| Haomeng|2 | rameshg87: welcome:) | 07:57 |
| rameshg87 | hello Haomeng|2: | 07:57 |
| rameshg87 | question i have is this .. | 07:57 |
| Haomeng|2 | rameshg87: sure | 07:58 |
| rameshg87 | does diskimage-builder cache the apt packages ? | 07:58 |
| Haomeng|2 | rameshg87: not sure, do you have performance issue? | 07:58 |
| rameshg87 | because everytime i try to create an image (fs image or ramdisk), it downloads from the internet | 07:58 |
| rameshg87 | does it happen to you as well ? | 07:58 |
| rameshg87 | it takes a lot of time :-) | 07:58 |
| Haomeng|2 | rameshg87: depends on your apt source I think | 07:59 |
| Haomeng|2 | Haomeng|2: yes | 07:59 |
| Haomeng|2 | rameshg87: we can raise the question in tripleo irc also I think:) | 07:59 |
| rameshg87 | it caches the ubuntu base image and checks the ubuntu cloud server whether it has changed, and uses the cached copy if it hasn't changed | 07:59 |
| rameshg87 | Haomeng|2: oh, is there a separate irc for tripleo ? | 07:59 |
| rameshg87 | i didn't know that :-) | 08:00 |
| Haomeng|2 | rameshg87: yes | 08:00 |
| Haomeng|2 | https://wiki.openstack.org/wiki/TripleO | 08:00 |
| Haomeng|2 | diskimage-builder is in TripleO project | 08:00 |
| *** jistr has joined #openstack-ironic | 08:01 | |
| rameshg87 | Haomeng|2: is it #triple ? | 08:01 |
| Haomeng|2 | rameshg87: #tripleo | 08:01 |
| Haomeng|2 | rameshg87: https://wiki.openstack.org/wiki/IRC | 08:02 |
| Haomeng|2 | it is special, no openstack words:) | 08:02 |
| rameshg87 | Haomeng|2: ah okay..finally got it :-) | 08:02 |
| Haomeng|2 | rameshg87: :) | 08:02 |
| rameshg87 | thanks | 08:02 |
| Haomeng|2 | rameshg87: maybe it is in /var/cache/apt/archieve | 08:03 |
| Haomeng|2 | can you check | 08:03 |
| Haomeng|2 | it should be apt-get problem:) | 08:03 |
| rameshg87 | Haomeng|2: but i think the apt-get installs are run within the chroot environment | 08:03 |
| Haomeng|2 | rameshg87: maybe | 08:04 |
| Haomeng|2 | rameshg87: check with default apt-get cache dir /var/cache/apt/archives/ | 08:04 |
| dtantsur | Morning Ironic | 08:04 |
| Haomeng|2 | dtantsur: morning | 08:04 |
| *** derekh has joined #openstack-ironic | 08:05 | |
| Haomeng|2 | rameshg87: we can find a lot of deb files | 08:05 |
| rameshg87 | Haomeng|2: yeah, but i am not sure if it is using them | 08:05 |
| Haomeng|2 | rameshg87: default behavior should be:) | 08:05 |
| Haomeng|2 | rameshg87: we can double check with tripleo guys:) | 08:06 |
| Mikhail_D_wk | Morning folks! :) | 08:06 |
| rameshg87 | Haomeng|2: for my laptop it works fine. once i install a package on my laptop, it doesn't download it again if i uninstall and reinstall. so apt for my laptop i think is fine | 08:06 |
| Haomeng|2 | rameshg87: :) | 08:06 |
| rameshg87 | Haomeng|2: but since diskimage-builder does it within the chroot environment, it is not using my laptops /var/cache/apt/archives i think | 08:07 |
| Haomeng|2 | Mikhail_D_wk: morning:) | 08:07 |
| Haomeng|2 | rameshg87: yes, this is point | 08:07 |
| Haomeng|2 | rameshg87: not the default enviroment | 08:07 |
| rameshg87 | Haomeng|2: for yum i can see it has used ~/.cache/image-create/yum/x86_64/20/fedora | 08:07 |
| Haomeng|2 | rameshg87: :) | 08:08 |
| rameshg87 | but not for apt :-( | 08:08 |
| Haomeng|2 | rameshg87: :) | 08:08 |
| Haomeng|2 | rameshg87: raise this problem into tripleo irc, will get response I think:) | 08:08 |
| rameshg87 | Haomeng|2: just wondering if i am missing something here and somebody knows caching for apt works for sure .. | 08:08 |
| rameshg87 | Haomeng|2: yeah sure, will do :-) | 08:08 |
| rameshg87 | Haomeng|2: thanks | 08:08 |
| Haomeng|2 | rameshg87: :) | 08:09 |
| Haomeng|2 | rameshg87: welcome | 08:09 |
| *** lucasagomes has joined #openstack-ironic | 08:09 | |
| Haomeng|2 | rameshg87: this can help you - http://www.tecmint.com/useful-basic-commands-of-apt-get-and-apt-cache-for-package-management/ | 08:11 |
| *** Mikhail_D_wk has left #openstack-ironic | 08:29 | |
| *** martyntaylor has quit IRC | 08:29 | |
| agordeev | morning Ironic :) | 08:30 |
| lucasagomes | lifeless, ping re 84361, I left a comment there about fixing the sleep in the tests | 08:32 |
| lucasagomes | morning agordeev | 08:33 |
| *** derekh has quit IRC | 08:33 | |
| lucasagomes | lifeless, apart from that the patch lgtm | 08:34 |
| lifeless | lucasagomes: I don't have the bandwidth to push that further; care to just take it over? | 08:34 |
| lucasagomes | lifeless, ack | 08:35 |
| lucasagomes | will send another patch set fixing it then :) | 08:35 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Workaround race with Neutron port updates https://review.openstack.org/84361 | 08:38 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Workaround race with Neutron port updates https://review.openstack.org/84361 | 08:39 |
| *** athomas has joined #openstack-ironic | 08:40 | |
| *** lynxman has quit IRC | 08:41 | |
| *** lynxman has joined #openstack-ironic | 08:42 | |
| *** max_lobur has quit IRC | 08:43 | |
| agordeev | lucasagomes: morning! | 08:43 |
| *** martyntaylor has joined #openstack-ironic | 08:44 | |
| lucasagomes | :) | 08:45 |
| *** matsuhashi has quit IRC | 09:07 | |
| *** athomas has quit IRC | 09:08 | |
| *** matsuhashi has joined #openstack-ironic | 09:12 | |
| *** athomas has joined #openstack-ironic | 09:13 | |
| lucasagomes | comstud, hi, i'd love to have a way to pull neutron and check whether the update_port() task was already finished or not | 09:16 |
| lucasagomes | instead of having that sleep :/ | 09:16 |
| *** ilives has quit IRC | 09:18 | |
| *** ilives has joined #openstack-ironic | 09:18 | |
| *** Mikhail_D_wk has joined #openstack-ironic | 09:31 | |
| *** max_lobur has joined #openstack-ironic | 09:33 | |
| *** derekh has joined #openstack-ironic | 09:33 | |
| lifeless | lucasagomes: rather than poll | 09:35 |
| *** coolsvap has quit IRC | 09:35 | |
| lifeless | lucasagomes: end the thread and wait for a callback | 09:35 |
| lifeless | lucasagomes: same as we do for for the image deploy | 09:35 |
| *** max_lobur1 has joined #openstack-ironic | 09:36 | |
| lucasagomes | lifeless, hmm... right, well that would introduce a lot of complexity with the locks, transitional state etc | 09:37 |
| * lucasagomes gotta look better at the code to see how it's going to fit there | 09:37 | |
| *** max_lobur has quit IRC | 09:38 | |
| *** Mikhail_D_wk has quit IRC | 09:40 | |
| *** Mikhail_D_wk has joined #openstack-ironic | 09:41 | |
| *** matsuhashi has quit IRC | 09:59 | |
| *** matsuhashi has joined #openstack-ironic | 09:59 | |
| *** martyntaylor has quit IRC | 10:13 | |
| *** martyntaylor has joined #openstack-ironic | 10:15 | |
| openstackgerrit | Sandhya Balakrishnan proposed a change to openstack/ironic: Ironic User Guide https://review.openstack.org/83343 | 10:17 |
| *** coolsvap has joined #openstack-ironic | 10:25 | |
| openstackgerrit | Sandhya Balakrishnan proposed a change to openstack/ironic: Ironic User Guide https://review.openstack.org/83343 | 10:27 |
| *** vkozhukalov has joined #openstack-ironic | 10:38 | |
| *** coolsvap has quit IRC | 10:38 | |
| *** ndipanov has quit IRC | 10:44 | |
| *** matsuhashi has quit IRC | 10:48 | |
| *** matsuhashi has joined #openstack-ironic | 10:55 | |
| *** romcheg has joined #openstack-ironic | 11:10 | |
| *** romcheg1 has joined #openstack-ironic | 11:10 | |
| *** romcheg has quit IRC | 11:10 | |
| *** matsuhashi has quit IRC | 11:15 | |
| *** matsuhashi has joined #openstack-ironic | 11:15 | |
| *** matsuhashi has quit IRC | 11:19 | |
| *** matsuhashi has joined #openstack-ironic | 11:21 | |
| vkozhukalov | jroll: which commit of openstack/oslo-incubator did you sync into ironic-python-agent? I need to add processutils and config.generator oslo modules and, of course, I need to re-sync them but in order to keep them consistent I need to know commit. Or maybe we can sync stable/havana? | 11:34 |
| *** pradipta is now known as pradipta_away | 11:40 | |
| *** rameshg87 has left #openstack-ironic | 11:41 | |
| *** matsuhashi has quit IRC | 11:52 | |
| *** subah has quit IRC | 12:19 | |
| *** linggao has joined #openstack-ironic | 12:29 | |
| agordeev | adam_g: https://review.openstack.org/#/c/85013/ | 12:31 |
| *** ndipanov has joined #openstack-ironic | 12:33 | |
| *** jdob has joined #openstack-ironic | 12:37 | |
| *** killer_prince has quit IRC | 12:45 | |
| *** nosnos has quit IRC | 12:50 | |
| *** matty_dubs|gone is now known as matty_dubs | 12:51 | |
| openstackgerrit | Pablo Fernando Cargnelutti proposed a change to openstack/ironic: Overwriting node_is_available in IronicDriver https://review.openstack.org/84214 | 12:56 |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Added execute utility and rootwrap module https://review.openstack.org/85025 | 13:00 |
| *** romcheg1 has left #openstack-ironic | 13:01 | |
| *** romcheg1 has joined #openstack-ironic | 13:01 | |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Added execute utility and rootwrap module https://review.openstack.org/85025 | 13:03 |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Added some infrastructure tools https://review.openstack.org/85030 | 13:21 |
| *** killer_prince has joined #openstack-ironic | 13:23 | |
| *** rloo has joined #openstack-ironic | 13:24 | |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Added execute utility and rootwrap module https://review.openstack.org/85025 | 13:35 |
| *** jbjohnso_ has joined #openstack-ironic | 13:36 | |
| *** jgrimm has quit IRC | 13:45 | |
| *** ilives has quit IRC | 13:48 | |
| *** sabah has joined #openstack-ironic | 13:49 | |
| *** ilives has joined #openstack-ironic | 13:52 | |
| NobodyCam | Good morning Ironic | 13:54 |
| romcheg1 | Morning NobodyCam lucasagomes and everyone else! | 13:54 |
| NobodyCam | morning romcheg1 :) | 13:55 |
| *** romcheg1 is now known as romcheg | 13:55 | |
| romcheg | fixed :) | 13:55 |
| NobodyCam | :) | 13:55 |
| agordeev | NobodyCam: romcheg morning! | 13:56 |
| lucasagomes | morning romcheg NobodyCam | 13:56 |
| NobodyCam | morning lucasagomes | 13:57 |
| NobodyCam | lucasagomes: I see in scroll back you pucked up lifeless' patch | 13:57 |
| NobodyCam | picked even | 13:58 |
| NobodyCam | ПӨBӨDΥCΛM need coffee | 13:58 |
| lucasagomes | NobodyCam, it was just a quick fix in the tests | 13:58 |
| lucasagomes | so that sleep doesn't affect the time to run the tests | 13:59 |
| NobodyCam | ahh ok. So I still need to dig to what up with the TripleO ci | 14:00 |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Added some oslo modules https://review.openstack.org/85036 | 14:01 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add DiskPartitioner https://review.openstack.org/83396 | 14:01 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Use DiskPartitioner https://review.openstack.org/83399 | 14:01 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Get rid of the swap partition https://review.openstack.org/83726 | 14:01 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Use GB instead of MB for swap https://review.openstack.org/83788 | 14:01 |
| jbjohnso_ | question, would people cry a lot if I changed pyghmi to use M2Crypto instead of PyCrypto? | 14:02 |
| lucasagomes | jbjohnso_, hmmm, PyCrypto is py3 compatible isnt it? | 14:03 |
| lucasagomes | where M2Crypto is not | 14:03 |
| lucasagomes | idk how py3 friendly is pyghmi right now | 14:03 |
| lucasagomes | but it sounds like a step back | 14:03 |
| lucasagomes | what r the reasons? | 14:03 |
| jbjohnso_ | lucasagomes, well, was going to see if there was a good reason, performance I was thinking of measuring | 14:05 |
| jbjohnso_ | lucasagomes, currently, 'while :; do dmesg; done' on a SOL console takes 3% of a 2.6 GHz CPU | 14:06 |
| jbjohnso_ | mostly in pycrypto stuff | 14:06 |
| jbjohnso_ | so was contemplating that | 14:06 |
| lucasagomes | jbjohnso_, and with the M2Crypto? how much improvement you get? | 14:06 |
| jbjohnso_ | dunno, was deciding whether it was worth checking | 14:06 |
| lucasagomes | if it's not something like WOW I would stick to PyCrypto | 14:07 |
| jbjohnso_ | if there was a 'no way', then not bother | 14:07 |
| jbjohnso_ | I'll probably at least change to use python hashlib instead | 14:07 |
| lucasagomes | I think py3 is an important variable in the game | 14:07 |
| jbjohnso_ | for hash | 14:07 |
| lucasagomes | right | 14:07 |
| lucasagomes | yeah sounds grand | 14:07 |
| jbjohnso_ | the cipher stuff though | 14:07 |
| jbjohnso_ | well, guess I'll change the hmac for now, see how much closer that gets me | 14:08 |
| jbjohnso_ | and maybe consider opportunistic use of M2Crypto if it looks appealing | 14:09 |
| *** sabah has quit IRC | 14:09 | |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Added some infrastructure tools https://review.openstack.org/85030 | 14:10 |
| jbjohnso_ | hmm, realizing why I see HMAC so high in the profiling results.. | 14:11 |
| jbjohnso_ | the AES in pycrypto is C code, the HMAC is pure python, guess that might help a great deal... | 14:11 |
| lucasagomes | heh yeah | 14:11 |
| lucasagomes | plus the problem with HMAC only accepting byte strings | 14:12 |
| jbjohnso_ | well, python core and PyCrypto share that attribute | 14:12 |
| lucasagomes | yeah :( I remember u were testing | 14:12 |
| jbjohnso_ | no big deal, doing an '.encode' is trivial | 14:12 |
| jbjohnso_ | and that info is only processed once during login | 14:13 |
| jbjohnso_ | the rest of the hmacs are things that are naturally bytearrays (packet data) | 14:13 |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Added execute utility and rootwrap module https://review.openstack.org/85025 | 14:15 |
| lucasagomes | yeah I remember the code you put on pyghmi to convert/tolerate unicode strings | 14:15 |
| lucasagomes | good stuff | 14:15 |
| jbjohnso_ | utf8 should be good enough for anyone | 14:16 |
| lucasagomes | +1 | 14:16 |
| jbjohnso_ | some non-ascii passphrases may not be able to be 20 characters long though | 14:16 |
| jbjohnso_ | and I have no idea if ipmitool can actually set non-ascii passphrases | 14:16 |
| lucasagomes | me neither | 14:17 |
| lucasagomes | worth testing tho | 14:17 |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Fix max retries message for _sync_power_states https://review.openstack.org/85044 | 14:18 |
| jbjohnso_ | yeah, to have a password like 葵 | 14:18 |
| NobodyCam | 刀o乃odﾘcﾑﾶ | 14:18 |
| romcheg | NobodyCam: шо то в дідька таке? :) | 14:19 |
| lucasagomes | heh | 14:20 |
| NobodyCam | romcheg: http://allfont.wapka.mobi/index.xhtml | 14:20 |
| lucasagomes | ééããççç | 14:20 |
| lucasagomes | :D | 14:20 |
| NobodyCam | ﾚucﾑ丂ﾑgoﾶ乇丂 | 14:21 |
| romcheg | г๏๓ςђєﻮ | 14:21 |
| dtantsur | wow Oo | 14:21 |
| NobodyCam | :) | 14:21 |
| NobodyCam | dtantsur: see link I just posted | 14:21 |
| dtantsur | ๔tคภtรยг | 14:22 |
| NobodyCam | :) | 14:22 |
| lucasagomes | gosh heh | 14:22 |
| dtantsur | I have one great IRC nick now :) | 14:22 |
| dtantsur | does IRC support non-latin nicks? | 14:22 |
| NobodyCam | thou irrsi wont let me set a utf8 nick | 14:22 |
| jbjohnso_ | let's find out... | 14:22 |
| lucasagomes | is that thai? | 14:22 |
| romcheg | dtantsur: No | 14:22 |
| jbjohnso_ | nope... | 14:22 |
| romcheg | I've just tried | 14:22 |
| NobodyCam | hehehe | 14:22 |
| dtantsur | romcheg, that's a pity :( | 14:22 |
| jbjohnso_ | 'erroneous nickname' | 14:22 |
| NobodyCam | yep | 14:23 |
| jbjohnso_ | now that would have been an evil thing to do to tab completion of nicks | 14:23 |
| NobodyCam | heheheh | 14:23 |
| NobodyCam | brb | 14:23 |
| jroll | morning ironic | 14:24 |
| *** jgrimm has joined #openstack-ironic | 14:24 | |
| *** sseago_ has joined #openstack-ironic | 14:24 | |
| jroll | vkozhukalov: afaik I synced master? I just ran their update.py script with the config in the agent repo... | 14:25 |
| matty_dubs | NobodyCam: ⓞⓗ, ⓣⓗⓘⓢ ⓘⓢ ⓒⓞⓞⓛ. | 14:25 |
| jroll | vkozhukalov: I'm not sure why it matters with commit | 14:25 |
| jroll | vkozhukalov: I would prefer master though | 14:25 |
| vkozhukalov | jroll: i'd prefer stable, afaik it is likely to get not working state when you sync unstable oslo | 14:27 |
| NobodyCam | morning jroll | 14:28 |
| NobodyCam | matty_dubs: hehehe | 14:28 |
| jroll | vkozhukalov: if something doesn't work, then we fix it upstream :) | 14:28 |
| jroll | vkozhukalov: maybe later on we can sync some stable release, but the agent doesn't even work right now, so... | 14:28 |
| vkozhukalov | ok, will sync master then | 14:29 |
| jroll | thanks | 14:30 |
| *** _LXXIII_ has joined #openstack-ironic | 14:30 | |
| jroll | vkozhukalov: why do we need config.generator? | 14:31 |
| agordeev | jroll: morning | 14:31 |
| NobodyCam | morning agordeev :) | 14:31 |
| jroll | the reason I ask, is because operationally, config files will be hard to manage in a ramdisk | 14:32 |
| NobodyCam | and morning matty_dubs | 14:32 |
| jroll | I'd prefer kernel or command line args | 14:32 |
| *** _LXXIII has joined #openstack-ironic | 14:33 | |
| *** _LXXIII has quit IRC | 14:33 | |
| vkozhukalov | jroll: i want agent to use oslo.config and config.generator is nothing than just a convenient way to generate config sample | 14:33 |
| *** sabah has joined #openstack-ironic | 14:34 | |
| Mikhail_D_wk | Morning/afternoon all! :) | 14:34 |
| agordeev | Mikhail_D_wk: morning | 14:34 |
| vkozhukalov | jroll: but it is quite convenient for testing | 14:34 |
| NobodyCam | Morning Mikhail_D_wk | 14:35 |
| jroll | vkozhukalov: ok, let's back up then. I'm skeptical on using oslo.config at all. what kind of options are going in there? | 14:35 |
| *** _LXXIII_ has quit IRC | 14:36 | |
| dtantsur | Mikhail_D_wk, мσяηιηg | 14:37 |
| devananda | morning, all | 14:38 |
| agordeev | jroll: doesn't oslo.config allow to use config file and command line args simultaneously? IMBW | 14:38 |
| agordeev | devananda: morning! | 14:38 |
| NobodyCam | good mornig devananda | 14:38 |
| vkozhukalov | jroll: plenty of them: logging options, authentication options, extensions activation options, path options, etc | 14:38 |
| Mikhail_D_wk | morning devananda :) | 14:38 |
| lucasagomes | morning devananda vkozhukalov | 14:38 |
| vkozhukalov | lucasagomes: morning | 14:38 |
| lucasagomes | morning Mikhail_D_wk | 14:39 |
| jroll | agordeev: does it support /proc/cmdline? :P | 14:39 |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Added oslo modules https://review.openstack.org/85052 | 14:39 |
| jroll | vkozhukalov: hmm | 14:39 |
| jroll | vkozhukalov: I guess let's get the reviews up and go from there | 14:39 |
| devananda | jroll: what about a kernel param hook for it to pull in a config file? | 14:39 |
| jroll | devananda: as in, pull in a remote config file? | 14:40 |
| devananda | jroll: if the initial ramdisk is served from a static DHCP/TFTP server, even if taht chains up to iPXE, you wont have control over unique kernel params | 14:40 |
| devananda | jroll: so if the driver needs to determine start-up options for the agent itself, you'll need a config file or comnandline params, ie, oslo.config | 14:41 |
| devananda | i'm not sure that it does. just saying "if" :) | 14:41 |
| jroll | devananda: we already have dynamic kernel params today, not from the driver though | 14:42 |
| devananda | eg, for plugins. I think two approaches are both valid: config specifies which plugins to load || plugins are loaded by stevedore when requested via the agent's API | 14:42 |
| devananda | jroll: from where? | 14:42 |
| jroll | ipxe.txt is served over http | 14:42 |
| jroll | so we have a little webapp sitting on our dhcp server | 14:42 |
| jroll | not ideal by any means | 14:43 |
| devananda | nope | 14:43 |
| jroll | but it's a quick hack we did to work around some other things | 14:43 |
| devananda | driver needs to be able to programatically set things ilke kernel params | 14:43 |
| jroll | it's something I want to think more about, it's kind of a hard problem no matter which way we go | 14:43 |
| devananda | serving ipxe.txt from ironic-api is an interesting idea ... | 14:43 |
| jroll | but wdyt about yes that | 14:43 |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Synced oslo modules https://review.openstack.org/85036 | 14:43 |
| devananda | seems logical to me | 14:44 |
| devananda | well | 14:44 |
| devananda | network layer seapration | 14:44 |
| jroll | the hard problem we solved with a dynamic ipxe.txt was, "how does the agent know what IP it's accessible at" | 14:44 |
| jroll | mmm | 14:45 |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Get rid object model `dict` methods part 3 https://review.openstack.org/64108 | 14:45 |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Get rid object model `dict` methods part 1 https://review.openstack.org/60025 | 14:45 |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Get rid object model `dict` methods part 2 https://review.openstack.org/62331 | 14:45 |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Get rid object model `dict` methods part 5 https://review.openstack.org/64278 | 14:45 |
| devananda | DC/sec/ops folks want to keep the IPMI network separate, only bridging ironic-conductor to it, and then also lock down the node's early network access to prevent injection from other running instances | 14:45 |
| jroll | yeah, in decom mode, we won't want the agents to be able to access ironic-api | 14:45 |
| devananda | right | 14:45 |
| devananda | well | 14:46 |
| devananda | we need the agent to access *an* ironic-api service for enrollment and just to talk to the driver | 14:46 |
| russell_h | right | 14:46 |
| devananda | so pulling ipxe.txt from the same IP is not any change to networking | 14:46 |
| russell_h | should that be "the ironic API"? | 14:46 |
| devananda | there may be a public IP and a private IP for reaching "the ironic API" endpoint | 14:47 |
| devananda | depends on deloyer's needs | 14:47 |
| jroll | russell_h: well, we could have a "decom" ironic-api service that only allows lookup/heartbeat | 14:47 |
| russell_h | right, thats how I saw it | 14:47 |
| russell_h | devananda: we're going to have an internal address that only allows whitelisted access to a few paths | 14:47 |
| devananda | russell_h: have you looked at the auth stuff for implementing that? | 14:48 |
| devananda | (i haven't) | 14:48 |
| jroll | I haven't either | 14:48 |
| russell_h | nope, I was under the impression that something like this exists | 14:48 |
| russell_h | but if it doesn't, I'll make it or something :) | 14:49 |
| devananda | i'm not sure. haven't looked enough at keystone v3's capabilities | 14:49 |
| russell_h | plus, nginx | 14:49 |
| devananda | mmmm, more dependencies -- | 14:49 |
| russell_h | right, if this is going to be the prescribed method the prescription should not say "deploy nginx with this configuration" | 14:50 |
| devananda | right | 14:50 |
| devananda | we should investigate existing methods before going down that route, though | 14:50 |
| russell_h | at any rate, conceptually, you feel we should stick to extending the existing API instead of building some new one for the agent to hit? | 14:51 |
| devananda | yes | 14:51 |
| russell_h | devananda: what do you want to do about this: https://review.openstack.org/#/c/81919/ | 14:52 |
| russell_h | email never generated much traffic | 14:52 |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Added some infrastructure tools https://review.openstack.org/85030 | 14:53 |
| devananda | russell_h: " I don't believe that this is a useful way for vendors to implement new consumer-facing functionality." | 14:55 |
| devananda | russell_h: is that in reference to #1 or #2 in your email? | 14:55 |
| russell_h | I don't think its a useful way to accomplish #1 | 14:56 |
| devananda | russell_h: in re-reading your email, I think it wasn't a clear distinction between the functionality of "node-specific" and "node-non-specific" passthru | 14:57 |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Added execute utility and rootwrap module https://review.openstack.org/85025 | 14:57 |
| russell_h | at least I can't envision what sort of consumer-facing API a vendor would want to add which isn't specific to a node | 14:57 |
| devananda | I also dropped the ball in replying to it during the release preparation period | 14:57 |
| russell_h | or at least a chassis or something if they're getting wild | 14:58 |
| devananda | russell_h: discover | 14:58 |
| russell_h | discover what? unregistered nodes? | 14:58 |
| devananda | yes | 14:58 |
| russell_h | interesting | 14:58 |
| russell_h | hadn't thought of it that way | 14:58 |
| devananda | chassis-level and IPMI-net-range have both been suggested | 14:58 |
| russell_h | right, so those make sense | 14:59 |
| devananda | eg, go discover 10.1.255.255 | 14:59 |
| jbjohnso_ | devananda, btw, I was going to add detection of snooped service processor and firmware identifiers | 14:59 |
| russell_h | and are probably where discover _should_ be implemented (if it should be implemented) | 14:59 |
| jbjohnso_ | devananda, to my hardware management daemon | 14:59 |
| jbjohnso_ | devananda, it would be able to passively detect a lot of things unobtrusively | 14:59 |
| devananda | jbjohnso_: hmm... hardware mgmt daemon? I'm sure you mentioned it and I'm just blanking on the reference | 14:59 |
| jbjohnso_ | devananda, yeah, I should demo it... | 15:00 |
| NobodyCam | we need better logging in ssh driver :-p | 15:00 |
| jbjohnso_ | devananda, it can do pretty nifty console stuff now... | 15:00 |
| devananda | NobodyCam: yes!!!! | 15:00 |
| devananda | NobodyCam: i beleive adam_g put up a patch to squelch a lot of the debug noise | 15:00 |
| jbjohnso_ | I was thinking of having a show and tell of the updated state of my daemon tomorrow using a crappy web conference | 15:00 |
| jbjohnso_ | (anything that calls for a java plugin is crappy) | 15:01 |
| NobodyCam | devananda: I am looking into the ci issue with triple O | 15:01 |
| jbjohnso_ | even crappier because I have to use windows to share my screen | 15:01 |
| jbjohnso_ | it's very weird to RDP into a windows vm to Xpra over to a linux box | 15:02 |
| *** martyntaylor has quit IRC | 15:02 | |
| devananda | jbjohnso_: you should, you know, just present it at OSCON or something ;) | 15:02 |
| jbjohnso_ | but a sneak example | 15:02 |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Unified object model https://review.openstack.org/85065 | 15:03 |
| jbjohnso_ | for power state, one url to use could be http://localhost:4005/nodes/n1/power/state.json | 15:03 |
| *** ndipanov_ has joined #openstack-ironic | 15:03 | |
| *** ndipanov__ has joined #openstack-ironic | 15:03 | |
| jbjohnso_ | the console throught http is not restful though | 15:04 |
| *** ndipanov__ has quit IRC | 15:04 | |
| jbjohnso_ | it takes about 9-10 milliseconds to get power state for one node | 15:04 |
| jbjohnso_ | actually, that's including starting wget | 15:04 |
| *** ndipanov has quit IRC | 15:04 | |
| jbjohnso_ | the console part is the cool part though, well that and the configuration and the filesystem-like interactive cli | 15:05 |
| devananda | comstud: if you have time, could you look at mikhail's patch series ^^ | 15:05 |
| devananda | comstud: it seems possibly overlapping with yours? | 15:05 |
| NobodyCam | devananda: when you have a second can you un -2 https://review.openstack.org/#/c/64100 | 15:05 |
| devananda | NobodyCam: done | 15:06 |
| NobodyCam | ty | 15:06 |
| *** _LXXIII_ has joined #openstack-ironic | 15:06 | |
| devananda | hah | 15:07 |
| devananda | our queue went from < 1page to > 2 pages overnight :) | 15:07 |
| NobodyCam | :) yay juno | 15:09 |
| *** dguerri_ is now known as _dguerri | 15:11 | |
| *** _dguerri is now known as dguerri_ | 15:12 | |
| dtantsur | As to me, we still have things that I would call critical for I, like https://review.openstack.org/#/c/84497/ or https://review.openstack.org/#/c/84396/ :( | 15:13 |
| vkozhukalov | guys, I have a question. lucasagomes suggested to use parted python binding https://fedorahosted.org/pyparted/ for partitioner. I believe it is quite stable and it is written in C. But I think it is much more convenient to use subprocess + parted instead, because many people know how to use it, it is much easier to review, it is much easier to fix bugs. On the other hand there are lvm python bindings as well. Some of them pure python, othe | 15:13 |
| *** martyntaylor has joined #openstack-ironic | 15:13 | |
| jbjohnso_ | devananda, I need a conference near my lazy self | 15:14 |
| NobodyCam | jbjohnso_: do like I do ... get an RV and call it home. then move the house to the conf :-p | 15:15 |
| dtantsur | vkozhukalov, at least they need to do this: https://fedorahosted.org/pyparted/ticket/52 | 15:17 |
| jbjohnso_ | so 4 consoles going full tilt in a vm on a 2.6 ghz core runs consistently between 8 and 11 percent cpu... | 15:17 |
| devananda | dtantsur: i agree that both of those are backport potential | 15:18 |
| lucasagomes | vkozhukalov, the reason why i didn't use the parted binds is becuase the process needs to run as root | 15:18 |
| lucasagomes | but they are very useful | 15:18 |
| dtantsur | lucasagomes, the fact it's not on pypi may be a problem | 15:19 |
| lucasagomes | I used it in another project, I liked... but I wouldn't oppose to have the parted + subprocess tho | 15:19 |
| lucasagomes | dtantsur, yeah indeed | 15:19 |
| * NobodyCam gets mo joe | 15:22 | |
| vkozhukalov | lucasagomes: dtantsur: devananda: ok, thank you for your opinions | 15:24 |
| vkozhukalov | will use commands + subprocess + rootwrap | 15:25 |
| lucasagomes | vkozhukalov, np :) | 15:25 |
| lucasagomes | vkozhukalov, ack | 15:25 |
| devananda | dtantsur: i've tagged both bugs with icehouse-backport-potential | 15:25 |
| lucasagomes | dtantsur, ops I forgot to add the 1MiB offset to that patch | 15:26 |
| dtantsur | lucasagomes, yes, thus -1 ;) | 15:27 |
| lucasagomes | dtantsur, heh, I fixed on the other one (long-term) https://review.openstack.org/#/c/83396/ | 15:27 |
| lucasagomes | and forgot that one | 15:27 |
| lucasagomes | will change it | 15:27 |
| dtantsur | lucasagomes, thanks! | 15:28 |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Switch from PyCrypto HMAC to standard python hmac https://review.openstack.org/85075 | 15:32 |
| NobodyCam | watching a jenins test can be much like watching paint dry | 15:34 |
| NobodyCam | jenkins even | 15:34 |
| *** eghobo has joined #openstack-ironic | 15:47 | |
| *** dwalleck has joined #openstack-ironic | 15:49 | |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Synced oslo modules https://review.openstack.org/85052 | 15:49 |
| NobodyCam | where is the check-tripleo/check-tripleo-ironic-undercloud-precise job | 15:51 |
| NobodyCam | devananda: ^^ do you know? | 15:51 |
| *** dguerri_ is now known as _dguerri | 15:53 | |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Synced oslo modules https://review.openstack.org/85052 | 15:53 |
| *** ndipanov_ has quit IRC | 15:54 | |
| *** _dguerri is now known as dguerri_ | 15:56 | |
| NobodyCam | bouncing bubbies tell me it is bbt time... brb | 15:57 |
| *** matty_dubs is now known as matty_dubs|lunch | 15:58 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: JsonPatch add operation on existing property https://review.openstack.org/85086 | 15:58 |
| lucasagomes | devananda, ^ it wasn't fixed no, the 'add' as replace only worked for multi-level attributes (like extra, driver_info, prorperties) | 15:59 |
| devananda | NobodyCam: last i heard, triple-ironic-seed is trustworthy, but undercloud is not. but IMBW | 16:02 |
| devananda | lucasagomes: ahhh. right. thanks | 16:02 |
| NobodyCam | yes looking in to that | 16:02 |
| devananda | i'm heading into the office today, will be on intermittently | 16:05 |
| NobodyCam | ack :) say hi to folks from me | 16:05 |
| *** ndipanov_ has joined #openstack-ironic | 16:08 | |
| lucasagomes | NobodyCam, root_gb in nova is the "disk size" from flavor? | 16:12 |
| *** jistr has quit IRC | 16:13 | |
| *** krtaylor has quit IRC | 16:13 | |
| NobodyCam | lucasagomes: yes.. but I think it should be root_gb = disk_size - swap - eff | 16:14 |
| lucasagomes | hmmm yeah idk if they r doing that calculation tho | 16:15 |
| * NobodyCam wander if we should have also had a reserved | 16:15 | |
| NobodyCam | lucasagomes: i do not think we are | 16:17 |
| devananda | theyre not. last i looked | 16:17 |
| devananda | root = disk. swap = swap. eph = eph | 16:17 |
| lucasagomes | yeah they are not | | u'pxe_root_gb': u'10', | 16:18 |
| lucasagomes | NobodyCam, in Ironic its correct we have 1 field for each | 16:18 |
| lucasagomes | NobodyCam, properties/local_gb == disk size | 16:18 |
| lucasagomes | root_gb, swap_gb, ephemeral_gb | 16:18 |
| lucasagomes | whcih is grand | 16:18 |
| devananda | another impedence mismatch between virt (just allocate what you ned) and physical (you have what you have, except for when you have a SAN) | 16:18 |
| lucasagomes | problem is in nova | 16:18 |
| lucasagomes | devananda, yeah, could be workarounded in the driver tho | 16:19 |
| lucasagomes | I mean Ironic driver | 16:19 |
| NobodyCam | so to debug this is issue are there objections to me setting debug true for a bit? | 16:19 |
| lucasagomes | NobodyCam, ? | 16:19 |
| NobodyCam | tripleo undercloud is having issues atm | 16:20 |
| lucasagomes | devananda, ephemeral and swap is only a baremetal thing for nova? | 16:20 |
| devananda | ok, no | 16:20 |
| devananda | s/ok// | 16:20 |
| devananda | lucasagomes: those are for VMs too | 16:21 |
| lucasagomes | devananda, so nova should do the calculation | 16:21 |
| lucasagomes | disk == root_gb - swap_mb - ephemeral_gb | 16:21 |
| devananda | if ironic exposes "disk size", then the nova ExactMatchDiskScheduer needs to do something | 16:21 |
| devananda | ya | 16:21 |
| devananda | or | 16:21 |
| devananda | if (disk != (root + swap + eph)) and not using_san(): | 16:22 |
| devananda | # don't match | 16:22 |
| lucasagomes | devananda, yeah | 16:22 |
| lucasagomes | +1 | 16:22 |
| devananda | or something | 16:22 |
| devananda | ok, i gotta run, bbiah | 16:22 |
| lucasagomes | I had one more thing heh | 16:22 |
| lucasagomes | we can talk later about it then | 16:22 |
| NobodyCam | stay warm devananda :) | 16:22 |
| devananda | lucasagomes: pnig me on gtalk | 16:22 |
| lucasagomes | devananda, ack | 16:22 |
| openstackgerrit | Ruby Loo proposed a change to openstack/python-ironicclient: Documentation for contributors https://review.openstack.org/85096 | 16:22 |
| lucasagomes | it's not critical anyway | 16:22 |
| *** max_lobur1 has quit IRC | 16:23 | |
| *** sphoorti has joined #openstack-ironic | 16:25 | |
| *** sphoorti has quit IRC | 16:30 | |
| comstud | devananda: will do | 16:32 |
| NobodyCam | post bbt walkies...bbiafm | 16:36 |
| comstud | devananda: just a 2 line patch right now that doesn't conflict | 16:36 |
| *** stevehuang has quit IRC | 16:36 | |
| comstud | but leaving a msg | 16:36 |
| *** ifarkas has quit IRC | 16:36 | |
| comstud | ah | 16:36 |
| *** romcheg has quit IRC | 16:39 | |
| *** killer_prince has quit IRC | 16:42 | |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Switch from PyCrypto HMAC to standard python hmac https://review.openstack.org/85075 | 16:48 |
| *** krtaylor has joined #openstack-ironic | 16:50 | |
| openstackgerrit | Ruby Loo proposed a change to openstack/python-ironicclient: Adds documentation for ironicclient API https://review.openstack.org/85107 | 16:53 |
| lucasagomes | aight, I came to the office today so I'm heading home now | 16:55 |
| lucasagomes | have a gnight devananda NobodyCam everyone | 16:55 |
| NobodyCam | have a good night lucasagomes | 16:55 |
| *** lucasagomes has quit IRC | 16:56 | |
| russell_h | agordeev: do you all have some sort of JSON payload that describes the hardware in a box? | 17:06 |
| russell_h | agordeev: in Fuel or something related | 17:06 |
| russell_h | ^ or anyone else in the know | 17:06 |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Added some oslo modules https://review.openstack.org/85036 | 17:07 |
| *** derekh has quit IRC | 17:08 | |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Added some infrastructure tools https://review.openstack.org/85030 | 17:09 |
| openstackgerrit | Vladimir Kozhukalov proposed a change to openstack/ironic-python-agent: Added execute utility and rootwrap module https://review.openstack.org/85025 | 17:12 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding a reference driver for the agent https://review.openstack.org/84795 | 17:13 |
| vkozhukalov | russell_h: in Fuel discovery agent parses output of ohai and reduces it significally, and yes it reports on master node hardware info as JSON | 17:14 |
| *** harlowja_away is now known as harlowja | 17:14 | |
| russell_h | vkozhukalov: do you have a link to what that JSON looks like? | 17:14 |
| russell_h | vkozhukalov: also, any thoughts on whether you'd change it any | 17:14 |
| russell_h | vkozhukalov: I'm thinking about what the agent's discovery payload should look like | 17:15 |
| russell_h | vkozhukalov: and I'm not interested in re-inventing this particular wheel if you all already have something | 17:15 |
| russell_h | hopefully would make things easier for you all too | 17:15 |
| pquerna | ohai. | 17:16 |
| pquerna | just a tiny dependency | 17:16 |
| JayF | ohai is a pretty good tool, but it's ruby and pretty heavy for a python agent | 17:17 |
| russell_h | we don't need to _use_ ohai | 17:17 |
| JayF | just mimic the structures it uses | 17:17 |
| russell_h | if it makes sense | 17:17 |
| russell_h | we don't _need_ to do that either, but if its good lets use it | 17:17 |
| vkozhukalov | russell_h: in Fuel we recognize node by its MACs | 17:17 |
| *** ilives has quit IRC | 17:17 | |
| *** rloo has quit IRC | 17:18 | |
| vkozhukalov | russell_h: and it looks like that http://paste.openstack.org/show/74992/ | 17:18 |
| *** ilives has joined #openstack-ironic | 17:18 | |
| vkozhukalov | pquerna: salt grains has pretty much the same functionality | 17:20 |
| *** max_lobur has joined #openstack-ironic | 17:20 | |
| russell_h | vkozhukalov: nice, thanks | 17:20 |
| *** lazy_prince has joined #openstack-ironic | 17:20 | |
| *** lazy_prince is now known as killer_prince | 17:20 | |
| devananda | back | 17:20 |
| NobodyCam | wb | 17:21 |
| *** athomas has quit IRC | 17:21 | |
| openstackgerrit | A change was merged to stackforge/pyghmi: Switch from PyCrypto HMAC to standard python hmac https://review.openstack.org/85075 | 17:23 |
| devananda | JoshNang: how close is IPA's driver to the reference PXE driver? (i haven't looked yet) | 17:24 |
| vkozhukalov | russell_h: but actually, we are not limited by this particular format, it is just important for us that new format contains at least that amount of data about node or even more | 17:25 |
| *** romcheg has joined #openstack-ironic | 17:25 | |
| JayF | vkozhukalov: I think it's more of a matter of, we need a format, and if one already exists might as well use it as a base unless there's a good reason not to | 17:25 |
| JayF | vkozhukalov: and I don't see a reason we can't format the data in a similar + possibly compatible way | 17:25 |
| JayF | Although we'll certainly want to include more. | 17:26 |
| JoshNang | devananda: it does deploy, tear down, and validate, uses vendor passthru. | 17:26 |
| vkozhukalov | russell_h: JayF: great | 17:26 |
| JoshNang | devananda: i'm not totally sure what you mean. | 17:26 |
| devananda | JoshNang: to be a DeployDriver, it must do those | 17:26 |
| devananda | hm, who's pcsforeducation ? | 17:27 |
| russell_h | thats JoshNang | 17:27 |
| JayF | JoshNang is pcsforeducation | 17:27 |
| JoshNang | yeah mea | 17:27 |
| JoshNang | *me | 17:27 |
| devananda | oh | 17:27 |
| JoshNang | where did it pop up? | 17:28 |
| devananda | # TODO(pcsforeducation) | 17:28 |
| devananda | is in a few place | 17:28 |
| JoshNang | oh :) i'm pcsforeducation on github | 17:28 |
| devananda | heh | 17:28 |
| russell_h | what name is that supposed to be? | 17:28 |
| devananda | i was about to say, please use your irc nic or gerrit user name | 17:28 |
| NobodyCam | woo hoo I can read the WhiteBoard again :-p | 17:29 |
| Shrews | For the record, I find it odd that a nova API method called get_host_stats() is only used to get a single attribute called 'hypervisor_hostname' | 17:29 |
| JoshNang | ack | 17:29 |
| Shrews | devananda: ^^^^ go figure | 17:30 |
| devananda | Shrews: ... right ... | 17:30 |
| devananda | the nova driver API seems to have grown a bit since I stopped working in there a year ago | 17:31 |
| Shrews | devananda: so, i think i can simplify your patch a great deal. running it through devstack now | 17:31 |
| devananda | Shrews: ++ | 17:32 |
| *** matty_dubs|lunch is now known as matty_dubs | 17:35 | |
| *** vkozhukalov has quit IRC | 17:36 | |
| *** rainya has quit IRC | 17:43 | |
| *** sabah has quit IRC | 17:49 | |
| devananda | adam_g: have you run the virtual-ironic job on https://review.openstack.org/#/c/81958/7 ? | 17:53 |
| *** pcargnel has joined #openstack-ironic | 17:54 | |
| pcargnel | Hi! Can someone help to figure out which is the error in this one? https://review.openstack.org/#/c/84214/ | 17:57 |
| pcargnel | Is check-tripleo-ironic-undercloud-precise a new job in the builds? I didn't notice until yesterday | 17:57 |
| openstackgerrit | Chris Krelle proposed a change to openstack/ironic: Add INFO level logging to ssh.py https://review.openstack.org/85124 | 17:58 |
| *** derekh has joined #openstack-ironic | 17:58 | |
| NobodyCam | pcargnel: I am looking into that now | 17:58 |
| devananda | pcargnel: I don't believe that check-tripleo-ironic-undercloud-precise is passing for anything yet | 17:59 |
| NobodyCam | I have a review up that should give us debug logges .. so we can figure out whats going on. | 17:59 |
| adam_g | devananda, hmm no i haven't. i didnt know that it would be triggered there | 17:59 |
| devananda | NobodyCam: last i heard from lifeless, check-tripleo-ironic-seed should be passing (and it does) but -undercloud is not | 17:59 |
| devananda | adam_g: post a comment sayign "check experimental" | 17:59 |
| adam_g | devananda, oh, but once that merges theres one minor change to devstack that would need to land to enable the driver testing | 17:59 |
| NobodyCam | yes | 17:59 |
| devananda | adam_g: so the virtual-ironic job is in the experimental queues for devstack, devstack-gate, and tempest | 18:00 |
| NobodyCam | devananda: I put up this https://review.openstack.org/85101 | 18:00 |
| devananda | adam_g: and we should run it for any patch we're making to those projects | 18:00 |
| NobodyCam | so I can get debug logs | 18:00 |
| devananda | adam_g: as an experimental job, it only runs when someone triggers it with a comment | 18:00 |
| adam_g | devananda, cool, but as i said that exercise wont run until devstack has a patch to set 'driver_enabled=True' in [baremetal] of tempest.conf | 18:01 |
| devananda | adam_g: ahhh. ok. can we land that in devstack first? :) | 18:02 |
| adam_g | devananda, yeah--may need to relax that crazy regex too to include a single scenario test | 18:03 |
| adam_g | also two more things CI related | 18:03 |
| adam_g | http://logs.openstack.org/99/83399/11/check/check-tempest-dsvm-virtual-ironic/dc5733d/logs/testr_results.html.gz | 18:03 |
| adam_g | is looking *much* better.. the two failures are actually bugs in tempest | 18:03 |
| *** rwsu has quit IRC | 18:03 | |
| dwalleck | wow. nice | 18:04 |
| adam_g | https://bugs.launchpad.net/tempest/+bug/1301782 | 18:04 |
| adam_g | https://bugs.launchpad.net/neutron/+bug/1265498 | 18:04 |
| devananda | awesome | 18:05 |
| NobodyCam | nice... and now its Bagel time | 18:05 |
| adam_g | devananda, and re: the nova unit tests not showing up in the results of https://review.openstack.org/#/c/84033/ -- its separate tox command that need to be run in addition to py27/py27 | 18:06 |
| adam_g | so we will likely need additional gate-ironic-nova-python26 + 27 check jobs | 18:06 |
| *** overlayer has joined #openstack-ironic | 18:06 | |
| adam_g | unless there is some magic we can do to get the nova tests to actually run as part of the ironic suite, but i havne't been able to | 18:07 |
| devananda | i think there should be | 18:09 |
| devananda | lemme see what i can do | 18:09 |
| adam_g | devananda, there are issues having testr import both set of tests (and ironic + nova together) during test discovery. i think comments in that review describe the issue | 18:10 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Add agent fields to virt driver https://review.openstack.org/85131 | 18:10 |
| dwalleck | adam_g: So it looks like the libvirt issue is fixed? | 18:13 |
| adam_g | dwalleck, yeah--i just bumped the job to use the icehouse UCA pocket instead of havana, where a newer libvirt package has all thebug fixes we're waiting on | 18:13 |
| openstackgerrit | Pablo Fernando Cargnelutti proposed a change to openstack/ironic: Change admin_url help in ironic driver https://review.openstack.org/84148 | 18:20 |
| *** k4n0 has joined #openstack-ironic | 18:21 | |
| *** rloo has joined #openstack-ironic | 18:23 | |
| Shrews | NobodyCam: for https://review.openstack.org/85124, you might want to first check out my review: https://review.openstack.org/84496 | 18:24 |
| * NobodyCam looks | 18:24 | |
| *** rloo has quit IRC | 18:25 | |
| *** rloo has joined #openstack-ironic | 18:25 | |
| openstackgerrit | Andreas Jaeger proposed a change to openstack/ironic: Use xargs -0 instead of --null https://review.openstack.org/85135 | 18:26 |
| *** rloo has quit IRC | 18:26 | |
| *** rloo has joined #openstack-ironic | 18:26 | |
| NobodyCam | Shrews: you mean set 84496 as Dependencies | 18:27 |
| devananda | adam_g: jsut talked with clarkb about your proposed unit test stuff, he has a slightly different idea i'm giong to try | 18:27 |
| Shrews | NobodyCam: i mean it might negate the need for your change. | 18:27 |
| adam_g | devananda, was that on IRC somewhere or face-to-face? im interested to know what it is | 18:28 |
| Shrews | NobodyCam: logs with DEBUG are much more readable with my proposal | 18:28 |
| NobodyCam | Shrews: I added because we had no info level logging in ssh at all... | 18:28 |
| NobodyCam | but yes I can see how 84496 cleans up the debug level | 18:29 |
| adam_g | devananda, if there were no issues importing nova/ironic together (oslo.config) it would be a simple as making nova available (either via git clone or test-requirements.txt) and adding the missing __init__.py to ironic/nova/tests/ so it is discovered and loaded with the others by testr | 18:29 |
| NobodyCam | Shrews: we support verbose= in the conf file .. .but don't actually log much at info level | 18:29 |
| NobodyCam | nobodycamAir:ironic NobodyCam$ grep -r LOG.info *|wc -l | 18:32 |
| NobodyCam | 36 | 18:32 |
| NobodyCam | if you take common out we have only 4 lines of info logging | 18:34 |
| NobodyCam | so I think that we need more info level logging over all | 18:35 |
| *** derekh has quit IRC | 18:37 | |
| *** dshulyak has joined #openstack-ironic | 18:38 | |
| *** dshulyak has quit IRC | 18:38 | |
| *** derekh has joined #openstack-ironic | 18:39 | |
| *** killer_prince has quit IRC | 18:42 | |
| devananda | adam_g: f2f, i'm in the office | 18:43 |
| Shrews | ugh... that moment when you realize that the LOG.debug() call you put in to help find a bug was actually causing another bug | 18:47 |
| NobodyCam | ugggh :-p | 18:47 |
| NobodyCam | log caused a error? | 18:47 |
| Shrews | yeah | 18:48 |
| JoshNang | heh i was just debugging one of those last night. a bad mock was messing with log.exception but not log.error in some tests. | 18:49 |
| NobodyCam | nice | 18:50 |
| NobodyCam | brb | 18:50 |
| openstackgerrit | A change was merged to openstack/ironic: Fix grammar in error string in pxe driver https://review.openstack.org/84471 | 18:51 |
| devananda | lifeless: having testr not barf on bad imports would, you know, be helpful. apparently that's fixed now? | 18:51 |
| NobodyCam | and back | 18:53 |
| *** derekh has quit IRC | 18:54 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding a reference driver for the agent https://review.openstack.org/84795 | 19:05 |
| dwalleck | adam_g: Once you fixed the libvirt issue with RS, did you have any issues launching servers? I got everything working but I've had this server in the building state, trying to figure out where it's hung up | 19:05 |
| adam_g | dwalleck, no its been working okay for me , let me give you the localrc i've been using | 19:06 |
| *** vkozhukalov has joined #openstack-ironic | 19:06 | |
| *** stevehuang has joined #openstack-ironic | 19:06 | |
| adam_g | devananda, you get better output now letting you know what failed to import in a human readable format | 19:07 |
| dwalleck | thanks. I believe it's jacked on waiting for a neutron port to come up. No errors though. Good times | 19:07 |
| adam_g | dwalleck, http://paste.ubuntu.com/7200113/ <- this is fairly close to what gets used in the devstack gate checks | 19:08 |
| dwalleck | Thanks. It looks pretty close to what I'm using (I'm using what's in the dev docs). Let me try this and see if I have more luck | 19:10 |
| adam_g | dwalleck, if you watch syslog wihile the node is booting you can confirm its coming up with correct networking by watching it DHCP + tftp | 19:11 |
| *** derekh has joined #openstack-ironic | 19:11 | |
| dwalleck | thanks, I'll give it a go | 19:14 |
| jroll | NobodyCam: I think it was you I was asking about the fields thing in the virt driver - does this make sense to you? https://review.openstack.org/#/c/85131/ | 19:17 |
| NobodyCam | jroll: if I am recalling what we chatted about .. ya... | 19:20 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Add agent fields to virt driver https://review.openstack.org/85131 | 19:21 |
| jroll | ok, cool | 19:22 |
| jroll | thanks | 19:22 |
| JayF | check-tripleo-ironic-seed-precise are the failures that are supposed to be voting, and the undercloud is reliably broken, right? | 19:23 |
| NobodyCam | JayF: just waiting for debug log to dive in | 19:23 |
| JayF | NobodyCam: I'm asking this more in the context of reviewing | 19:24 |
| NobodyCam | been in the queue for 3 hours now | 19:24 |
| NobodyCam | ahh yes then it be broken | 19:24 |
| JayF | i.e. I probably shouldn't +1 things that are failing check-tripleo-ironic-seed-preicse, right? | 19:24 |
| NobodyCam | :-p | 19:24 |
| NobodyCam | yes seed is working | 19:24 |
| JayF | while you're looking, it's 'recheck experimental' to get it to rerun, correct? | 19:25 |
| NobodyCam | I use recheck nobug or recheck bug# most offten | 19:26 |
| NobodyCam | experimental is for the tempest jobs ... I think.... | 19:26 |
| NobodyCam | devananda: ^^^ | 19:26 |
| JayF | Yeah I just didn't know if that retriggered the tripleo ones as wel | 19:26 |
| JayF | *well | 19:27 |
| NobodyCam | ok be back in a few min... | 19:29 |
| *** overlayer has quit IRC | 19:29 | |
| lifeless | devananda: it is, pending some pbr changes to release. you can pip install from bzr | 19:33 |
| linggao | matty_dubs, ping | 19:36 |
| *** mrda_away is now known as mrda | 19:42 | |
| matty_dubs | linggao: pong | 19:44 |
| matty_dubs | (sorry, was AFK) | 19:44 |
| linggao | matty_dubs, np. I am playing with the console code that sjing has put in. And found some problems. Just want to check with you to see what were the probmems when you were testing it. | 19:46 |
| matty_dubs | linggao: I had problem with shellinaboxd starting/shutting down | 19:47 |
| matty_dubs | And I think it was causing the node to stay locked | 19:47 |
| linggao | matty_dubs I saw that problem. | 19:48 |
| linggao | the pid file was not created. | 19:48 |
| matty_dubs | My memory is a little hazy; I think I was getting a .pid file, but the service couldn't be stopped. | 19:49 |
| matty_dubs | But a lot could have changed since. | 19:49 |
| linggao | I have a question for a different problem, should each node have different ipmi_terminal_port? | 19:50 |
| linggao | otherwise the console will run into each other if we open console for more than one nodes. | 19:50 |
| matty_dubs | linggao: I didn't consider that at first, and I don't think the cond considers it. | 19:50 |
| matty_dubs | *code | 19:50 |
| matty_dubs | But yes, we will. | 19:51 |
| matty_dubs | Since only the first one can be on 4200 | 19:51 |
| matty_dubs | (Unless I am missing something fundamental.) | 19:51 |
| linggao | then the admin assign a port for each node when registering a node? | 19:52 |
| linggao | or should there be a terminal port pool that the code will assign the terminal port at run time when user request a console? | 19:53 |
| *** ilives has quit IRC | 19:55 | |
| *** ilives has joined #openstack-ironic | 19:56 | |
| *** rwsu has joined #openstack-ironic | 19:56 | |
| matty_dubs | linggao: You know, I honestly hadn't thought this through previously. The way it works now is that the port is assigned when the user registers the node. | 19:57 |
| matty_dubs | I don't know if that's ideal, though. | 19:57 |
| matty_dubs | I suppose it works. | 19:57 |
| NobodyCam | and back | 19:57 |
| linggao | matty_dubs, thanks. I'll do some research in this area. I just worry about the port numbers will be all used if we have thousands of bm nodes. | 19:59 |
| matty_dubs | linggao: Yes, good point. It could also be a pain to have to manually assign. | 19:59 |
| *** dwalleck has quit IRC | 19:59 | |
| linggao | Did you see other problems? | 19:59 |
| matty_dubs | linggao: I don't _remember_ any other problems, but it's possible I forgot. Are you hitting more? | 20:00 |
| linggao | yes, like it does not pick up the path for shellinaboxd in the conf. | 20:01 |
| linggao | ironic node-get-console returns 'unicode' object has no attribute 'iteritems' | 20:01 |
| *** harlowja has quit IRC | 20:02 | |
| linggao | I'll go a head fix them. | 20:02 |
| linggao | matty_dubs, thanks a lot for your time. | 20:03 |
| *** harlowja has joined #openstack-ironic | 20:03 | |
| *** ilives has quit IRC | 20:03 | |
| matty_dubs | Ah, ironic node-get-console didn't exist when I was doing this, so I wouldn't have hit that. | 20:04 |
| matty_dubs | linggao: No problem, thanks for picking this up! I'll be happy to help test/review as well. | 20:04 |
| linggao | matty_dubs, that will be great! | 20:04 |
| *** ilives has joined #openstack-ironic | 20:05 | |
| *** rwsu has quit IRC | 20:07 | |
| russell_h | NobodyCam: thoughts on this? https://review.openstack.org/#/c/84814/ | 20:08 |
| *** rwsu has joined #openstack-ironic | 20:09 | |
| NobodyCam | rloo: love the usually on line 75 of api_v1.rst (https://review.openstack.org/#/c/85107/1/doc/source/api_v1.rst) | 20:09 |
| rloo | NobodyCam: well, I am not familiar with all that code, so couldn't say for sure that it *always* :-) | 20:10 |
| NobodyCam | :) | 20:10 |
| rloo | also, ClientException is deprecated -- I want to ask lucas about that... | 20:10 |
| NobodyCam | heheh | 20:10 |
| *** rwsu has quit IRC | 20:11 | |
| *** rwsu has joined #openstack-ironic | 20:11 | |
| NobodyCam | russell_h: I like it.. Just gave it a quick look through.. | 20:14 |
| russell_h | NobodyCam: enough to give it a +2? | 20:19 |
| NobodyCam | russell_h: that one will need testing... | 20:20 |
| NobodyCam | my ci tests just started running so I'm watching the jenkins logs atm | 20:20 |
| NobodyCam | but I agree that hash ring is a big enough pice to have its own class :) so in theroy yes I do !!! | 20:21 |
| russell_h | NobodyCam: how do you feel about use of dbapi in ironic/common/ | 20:22 |
| russell_h | I sort of feel like it oughtn't be there, but figured I'd do it the easy way and see how people feel | 20:23 |
| NobodyCam | thats one of the things i'd like to look closer at.. but a TODO: remove dbapi would cover my initial conserns | 20:24 |
| devananda | back | 20:27 |
| *** rwsu has quit IRC | 20:27 | |
| NobodyCam | wb devananda | 20:27 |
| devananda | adam_g: you mentioned a DuplicateOptError? | 20:28 |
| *** rwsu has joined #openstack-ironic | 20:31 | |
| openstackgerrit | A change was merged to openstack/ironic-python-agent: Synced oslo modules https://review.openstack.org/85052 | 20:32 |
| *** derekh has quit IRC | 20:35 | |
| *** rainya- has joined #openstack-ironic | 20:40 | |
| openstackgerrit | A change was merged to openstack/ironic-python-agent: Added some oslo modules https://review.openstack.org/85036 | 20:42 |
| openstackgerrit | A change was merged to openstack/ironic-python-agent: Added some infrastructure tools https://review.openstack.org/85030 | 20:43 |
| adam_g | devananda, yeah i believe that was the exception. use_syslog was the option, IIRC | 20:49 |
| *** dkehn__ has joined #openstack-ironic | 20:50 | |
| *** dkehn_ has quit IRC | 20:52 | |
| *** dkehn__ is now known as dkehn_ | 20:54 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding swift temp url support https://review.openstack.org/81391 | 20:57 |
| *** romcheg has quit IRC | 20:58 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Adding a reference driver for the agent https://review.openstack.org/84795 | 20:58 |
| *** k4n0 has left #openstack-ironic | 20:59 | |
| *** matty_dubs is now known as matty_dubs|gone | 21:00 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Adding oslotest to improve debugging https://review.openstack.org/84212 | 21:03 |
| openstackgerrit | A change was merged to openstack/ironic-python-agent: Adding oslotest to improve debugging https://review.openstack.org/84212 | 21:09 |
| *** jdob has quit IRC | 21:10 | |
| *** linggao has quit IRC | 21:11 | |
| *** dwalleck has joined #openstack-ironic | 21:11 | |
| *** jbjohnso_ has quit IRC | 21:11 | |
| *** dkehn__ has joined #openstack-ironic | 21:15 | |
| *** dkehn__ has quit IRC | 21:16 | |
| *** dkehn__ has joined #openstack-ironic | 21:16 | |
| *** dkehn_ has quit IRC | 21:16 | |
| *** dwalleck has quit IRC | 21:20 | |
| *** rwsu has quit IRC | 21:22 | |
| *** martyntaylor has left #openstack-ironic | 21:22 | |
| *** dkehn_ has joined #openstack-ironic | 21:23 | |
| *** dkehn__ has quit IRC | 21:25 | |
| *** vkozhukalov has quit IRC | 21:30 | |
| NobodyCam | wt... FATAL: Unable to delete script file /tmp/hudson5356793296994292431.sh | 21:32 |
| NobodyCam | 2014-04-03 21:30:37.361 | hudson.util.IOException2: remote file operation failed: /tmp/hudson5356793296994292431.sh at hudson.remoting.Channel@179cda28:tripleo-precise-tripleo-test-cloud-342761 | 21:32 |
| *** dkehn__ has joined #openstack-ironic | 21:32 | |
| *** dkehn_ has quit IRC | 21:34 | |
| *** dkehn__ is now known as dkehn-x | 21:37 | |
| *** dwalleck has joined #openstack-ironic | 21:38 | |
| *** pcargnel has quit IRC | 21:40 | |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Refactor nova.virt.ironic.driver get_host_stats https://review.openstack.org/83853 | 21:52 |
| *** dwalleck has quit IRC | 21:55 | |
| NobodyCam | oh :( jenkins just happens to have a break down running my test... hummm.... coincidence... :-p | 21:56 |
| *** matty_dubs|gone is now known as matty_dubs | 21:56 | |
| Shrews | NobodyCam: you broke it! shame | 21:59 |
| *** max_lobur has quit IRC | 22:00 | |
| NobodyCam | no no really it just broke for other reasons | 22:00 |
| Shrews | Also, nova driver code makes me want to curl up in a corner and cry | 22:00 |
| NobodyCam | :-( | 22:00 |
| NobodyCam | Shrews: we could write a google glasses app that would just block it from view.. you know put cute kitten pics or something like that over top of the code | 22:02 |
| Shrews | ++ | 22:03 |
| Shrews | NobodyCam: So for 83853, it's ready for you to test again. I just -1'd it b/c I just realized that the driver unit test needs updated, but that shouldn't stop your testing. | 22:05 |
| NobodyCam | humm /me now wounders if he could write a Gglasses app that would convert python code on the fly.. and display it in what ever lang the users wants | 22:05 |
| Shrews | ugh, and i missed the import | 22:07 |
| Shrews | oh, no i didn't | 22:07 |
| * Shrews needs beer | 22:07 | |
| NobodyCam | Shrews: ack... but so you know.. I know I have landed patches that I know broke the current nova driver test.. | 22:08 |
| JoshNang | NobodyCam: you wouldn't want to do it for Glass. staring at that screen for more than about 10 seconds isn't very comfortable. But they do have an awesome translation app for non-programming languages that does in place text replacement. its amazing | 22:08 |
| Shrews | NobodyCam: oh. with adam_g's recent work, i don't think we want to continue doing that. or he might get all stabby on us | 22:09 |
| NobodyCam | JoshNang: :) | 22:09 |
| NobodyCam | Shrews: it was awhile ago | 22:09 |
| NobodyCam | *cough* | 22:09 |
| adam_g | Shrews, we'll ideally have those running as checks in the test pipeline in one way or another | 22:10 |
| NobodyCam | Shrews: http://www.rozmon.com/t_shirt_rozmon_canadian_bullet_proof_clothing.php | 22:13 |
| NobodyCam | also good vs. atabbings | 22:13 |
| NobodyCam | s/atabbings/stabbings/ | 22:14 |
| Shrews | lol | 22:14 |
| NobodyCam | heheh | 22:14 |
| devananda | adam_g: did you find a solution to that exception? | 22:15 |
| adam_g | devananda, haven't, finsihing up this 2013.2.3 release. AFAIK oslo.config options are global and both nova and ironic use a use_syslog (among others, probably) | 22:16 |
| adam_g | devananda, what did clark suggestion? | 22:17 |
| devananda | adam_g: add http://tarballs.openstack.org/nova/nova-master.tar.gz#egg=nova | 22:19 |
| devananda | to test-requirements.txt | 22:19 |
| devananda | adam_g: and put nova's unit tests in ironic/tests/nova/ | 22:19 |
| devananda | but this runs into that DuplicateOpt exc | 22:19 |
| adam_g | devananda, right, that would be the easiest from a setup POV. i had tried something similar originally but hit those oslo issues | 22:19 |
| *** dwalleck has joined #openstack-ironic | 22:20 | |
| adam_g | may be solvable in oslo.confg. i'm guessing it uses some dict thats key'd by option name. illl take a look at that in a bit | 22:20 |
| devananda | LOL | 22:20 |
| devananda | so | 22:20 |
| devananda | clarkb apapretly fileda bug about this issue in oslo | 22:21 |
| devananda | which was promptly marked notabug | 22:21 |
| adam_g | ha | 22:21 |
| adam_g | #? | 22:21 |
| *** krtaylor has quit IRC | 22:22 | |
| devananda | was yeeeears ago, lol | 22:23 |
| devananda | *shrug* | 22:23 |
| adam_g | oh ha | 22:24 |
| adam_g | ill take a look at oslo.confg in a bit and get another bug filed that describes our use case. something tells me we are blazing trails here | 22:24 |
| *** rwsu has joined #openstack-ironic | 22:26 | |
| *** dkehn-x has quit IRC | 22:36 | |
| devananda | i'm not having any good luck here | 22:37 |
| devananda | yea | 22:37 |
| devananda | adam_g: looks like the test discovery imports each module, which is triggering oslo.config to load the conf options into its global state | 22:52 |
| devananda | adam_g: so putting nova unit tests under the path where ironic unit tests also live causes the test discovery to fail | 22:53 |
| devananda | unless they're exlucded | 22:53 |
| devananda | at least, that's our theory at the moment | 22:53 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Add BackOffLoopingCall with jitter https://review.openstack.org/84303 | 23:00 |
| *** dwalleck has quit IRC | 23:00 | |
| adam_g | devananda, yeah | 23:05 |
| adam_g | test_ironic_foo - > from ironic import foo -> oslo config opts registered | 23:05 |
| adam_g | test_nova_foo - > from nova import foo -> oslo config opts registered | 23:05 |
| *** matsuhashi has joined #openstack-ironic | 23:05 | |
| adam_g | gonna start poking at that now.. had to eat something. | 23:06 |
| adam_g | horrible portland pizza | 23:06 |
| *** dkehn has joined #openstack-ironic | 23:06 | |
| dhellmann | what's the issue with the config options? | 23:06 |
| *** matty_dubs is now known as matty_dubs|gone | 23:07 | |
| *** dkehn is now known as dkehnx | 23:07 | |
| adam_g | dhellmann, testr importing all tests during discovery ends up importing both nova and ironic, causing conflicts on options registered by both (eg, use_syslog) | 23:08 |
| dhellmann | the ironic tests are loading nova code? | 23:09 |
| * NobodyCam knew that was comming | 23:09 | |
| NobodyCam | dhellmann: we have the ironic nova driver in our tree | 23:09 |
| dhellmann | ceilometer has something similar, we had to put them in a separate test directory | 23:10 |
| dhellmann | http://git.openstack.org/cgit/openstack/ceilometer/tree/nova_tests/test_notifier.py | 23:10 |
| dhellmann | NobodyCam: I'm not judging, just checking that I understand :-) | 23:10 |
| NobodyCam | hehe ya :) | 23:10 |
| NobodyCam | its late in hte day | 23:10 |
| adam_g | http://paste.ubuntu.com/7200995/ | 23:10 |
| NobodyCam | the even | 23:10 |
| adam_g | dhellmann, they currently live in separate directories, but testr is going to import everything it discovers | 23:11 |
| dhellmann | adam_g: run 2 testr jobs, one for each directory | 23:11 |
| dhellmann | hmm, it appears ceilometer is no longer testing that nova notifier plugin at all, probably for a similar reason :-( | 23:12 |
| adam_g | dhellmann, tests get run via setup.py testr | 23:12 |
| dhellmann | adam_g: tests are run by "tox -e something" so you have control over that | 23:13 |
| dhellmann | and testr can be told which directory to look at -- it doesn't have to start with "." | 23:13 |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Handling validation in conductor consistently https://review.openstack.org/84374 | 23:18 |
| *** jgrimm has quit IRC | 23:18 | |
| NobodyCam | afternoon walkies...bbiafm | 23:27 |
| devananda | dhellmann: adam_g: sounds like you two are thinking along similar lines | 23:28 |
| devananda | adam_g: the diff here is that I think we can add nova-latest-tarball (and mox) as test-requirements, then do something like what we've been discussing | 23:29 |
| adam_g | devananda, maybe. banging on testr/subunit now | 23:29 |
| devananda | yea. i banged on it for a while between meetings too, didn't get past the DuplciateOpt error though | 23:30 |
| adam_g | devananda, yeah--it would be ideal if we didn't have to change anything in gate and can just rely on jenkins calling tox -e py27/py26 and having the nova tests run | 23:30 |
| devananda | so if we can't, then your current patch is good | 23:30 |
| devananda | it's fine if we need to have jenkins call a separate test | 23:30 |
| dhellmann | tox can run more than one command in the same env | 23:30 |
| devananda | but when i showed teh bash script to clark, he was not happy | 23:30 |
| devananda | I gotta run again .... back later | 23:31 |
| adam_g | dhellmann, AFAICS the starting dir for test discovery is defined in .testr.conf's test_command | 23:31 |
| NobodyCam | enjoy devananda | 23:31 |
| dhellmann | adam_g: you can pass it on the command line, too | 23:31 |
| adam_g | dhellmann, cool. will take a closer look when venv is done rebuilding | 23:31 |
| dhellmann | adam_g: oslo-incubator calls testr more than once, but not with a different directory | 23:32 |
| dhellmann | adam_g: hmm, I might be wrong about the test dir thing | 23:34 |
| adam_g | dhellmann, yeah, looking as olso its using regex | 23:34 |
| dhellmann | yeah | 23:34 |
| adam_g | dhellmann, which doenst help here--testr is going to import everythign before applying that regex | 23:34 |
| dhellmann | try looking through ceilometer's history to see where they removed that test job -- jd__ or gordc may be able to help with that | 23:34 |
| adam_g | dhellmann, cool. | 23:35 |
| dhellmann | why are the config options clashing? are the defaults different? | 23:35 |
| adam_g | dhellmann, https://github.com/openstack/oslo.config/blob/master/oslo/config/cfg.py#L502 | 23:38 |
| adam_g | the ironic code was originally a copy from nova, so it makes sense they both use 'use-syslog' | 23:39 |
| dhellmann | adam_g: that actually checks more than the name | 23:39 |
| dhellmann | it compares the 2 option objects, so if their settings are the same you don't get an error | 23:39 |
| dhellmann | the error should be updated to complain about the difference | 23:39 |
| adam_g | right | 23:39 |
| dhellmann | so having 2 options with the same name is fine, as long as all of their other settings are the same | 23:40 |
| dhellmann | what is different between the nova and ironic settings? | 23:40 |
| dhellmann | could it be the help string, for example? | 23:40 |
| adam_g | not sure off the top of my head | 23:40 |
| dhellmann | adam_g: figuring that out might be the shorter path to make it work | 23:41 |
| *** eghobo has quit IRC | 23:42 | |
| adam_g | dhellmann, hmph looks like they're actually clashing on help messages | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!