*** achanda_ has joined #openstack-ironic | 00:03 | |
*** achanda has quit IRC | 00:07 | |
*** achanda_ has quit IRC | 00:08 | |
*** Marga_ has joined #openstack-ironic | 00:09 | |
NobodyCam | romcheg: :) cool i'll look it up | 00:12 |
---|---|---|
*** Marga_ has quit IRC | 00:27 | |
*** mtanino_ has joined #openstack-ironic | 00:27 | |
*** Marga_ has joined #openstack-ironic | 00:28 | |
*** mtanino has quit IRC | 00:29 | |
*** ijw has quit IRC | 00:31 | |
*** naohirot has joined #openstack-ironic | 00:33 | |
romcheg | Done, it's time to go to bed now | 00:36 |
romcheg | See you guys! | 00:36 |
*** romcheg has quit IRC | 00:36 | |
*** Marga_ has quit IRC | 00:49 | |
*** kkoski has joined #openstack-ironic | 00:53 | |
*** Marga_ has joined #openstack-ironic | 01:04 | |
*** kkoski has quit IRC | 01:07 | |
* BadCub updates the review day pad with new bugs | 01:07 | |
*** ChuckC has quit IRC | 01:08 | |
mrda | BadCub: This is a simple fix that improves our backward compat: https://review.openstack.org/#/c/169101 (might be worthwhile getting in for K) | 01:09 |
openstackgerrit | Zhenguo Niu proposed openstack/ironic: Use node UUID in logs instead of node ID https://review.openstack.org/167623 | 01:13 |
* BadCub looks | 01:17 | |
*** spandhe has quit IRC | 01:18 | |
BadCub | mrda: can you tag the commit for 169101 with the micro-version BP ? | 01:18 |
mrda | BadCub: Sure | 01:21 |
BadCub | ty :-) | 01:21 |
openstackgerrit | Tan Lin proposed openstack/ironic: Enable pxe_amt driver with localboot https://review.openstack.org/167975 | 01:23 |
*** mitz has quit IRC | 01:26 | |
openstackgerrit | Michael Davies proposed openstack/ironic: Improve backwards compat on HTTP error codes https://review.openstack.org/169101 | 01:26 |
BadCub | mrda: I added it to the review day pad | 01:27 |
mrda | thnx | 01:27 |
BadCub | yw :-) | 01:27 |
BadCub | I'm gonna head off to do dinner-y things. Should be back for the meeting tonight | 01:28 |
*** chenglch has joined #openstack-ironic | 01:32 | |
*** kkoski has joined #openstack-ironic | 01:32 | |
*** ChuckC has joined #openstack-ironic | 01:34 | |
*** mitz has joined #openstack-ironic | 01:36 | |
*** jerryz_ has joined #openstack-ironic | 01:39 | |
jroll | I have flappy conductors, wtf | 01:50 |
jroll | no clue as to why | 01:50 |
* jroll suddenly wonders if he has broken configs or if we broke something in the hash ring stuff | 01:52 | |
* jroll walks away | 01:53 | |
openstackgerrit | Haomeng,Wang proposed openstack/ironic: Fix UnicodeEncodeError issue when the language is not en_US https://review.openstack.org/168268 | 01:56 |
*** kkoski has quit IRC | 02:01 | |
openstackgerrit | Michael Davies proposed openstack/ironic: Improve backwards compat on API behaviour https://review.openstack.org/169101 | 02:16 |
*** rloo has quit IRC | 02:16 | |
* mrda notes the good timing of pushing the patch to address rloo's comments just as she leaves :) | 02:18 | |
mrda | BadCub: FWIW, I've raised a bug for review 163730 and referenced that instead of the bp | 02:19 |
mrda | BadCub: Sorry, mea culpa, I meant for 169101 | 02:19 |
*** achanda has joined #openstack-ironic | 02:34 | |
*** Haomeng has joined #openstack-ironic | 02:36 | |
*** Haomeng|2 has quit IRC | 02:37 | |
*** ramineni has joined #openstack-ironic | 02:40 | |
*** achanda has quit IRC | 02:44 | |
*** achanda has joined #openstack-ironic | 02:45 | |
*** mtanino_ has quit IRC | 02:50 | |
*** Marga_ has quit IRC | 02:58 | |
*** harlowja_ is now known as harlowja_away | 03:07 | |
*** Marga_ has joined #openstack-ironic | 03:15 | |
*** saripurigopi has joined #openstack-ironic | 03:45 | |
ramineni | JoshNang: hi | 04:09 |
ramineni | there? | 04:09 |
*** krtaylor has quit IRC | 04:11 | |
*** achanda has quit IRC | 04:12 | |
*** krtaylor has joined #openstack-ironic | 04:14 | |
*** rameshg87 has joined #openstack-ironic | 04:17 | |
rameshg87 | good morning ironic | 04:17 |
*** achanda has joined #openstack-ironic | 04:19 | |
*** spandhe has joined #openstack-ironic | 04:21 | |
*** spandhe_ has joined #openstack-ironic | 04:24 | |
*** spandhe has quit IRC | 04:26 | |
*** spandhe_ is now known as spandhe | 04:26 | |
*** achanda has quit IRC | 04:27 | |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Convert inspect_ports to string before string operations https://review.openstack.org/168776 | 04:27 |
*** achanda has joined #openstack-ironic | 04:30 | |
Haomeng | rameshg87: good morning ramesh:) | 04:30 |
mrda | hey rameshg87 and Haomeng | 04:31 |
Haomeng | mrda: morning:) | 04:31 |
rameshg87 | good afternoon Haomeng and mrda (i hope it's afternoon for Haomeng too) :) | 04:31 |
Haomeng | rameshg87: yes, my local time is 12pm:) | 04:31 |
Haomeng | mrda, rameshg87, help me take a look for my patch, welcome your comments - https://review.openstack.org/#/c/168268/ | 04:31 |
Haomeng | if you have time, thank you! | 04:32 |
mrda | ahh, gate failures!! | 04:32 |
Haomeng | mrda: yes, should not my code issue I think:) | 04:33 |
mrda | I've been seeing this intermittent problem today a lot | 04:33 |
*** oomichi_ has joined #openstack-ironic | 04:33 | |
Haomeng | mrda: found some devstack log error - Could not install requirement XStatic-Angular-Irdragndrop>=1.0.2.1 (from horizon==2015.1.dev106) because of error 404 Client Error: Not Found | 04:34 |
mrda | Haomeng: love the test name "def test____init__(self)" | 04:34 |
Haomeng | mrda: :) | 04:34 |
rameshg87 | Haomeng: sure | 04:34 |
rameshg87 | Haomeng: found same set of failures here too | 04:35 |
rameshg87 | https://review.openstack.org/#/c/167975/ | 04:35 |
mrda | and https://review.openstack.org/#/c/169101/ | 04:35 |
Haomeng | mrda: yes | 04:35 |
Haomeng | should be same issue | 04:36 |
rameshg87 | mrda: Haomeng: might be related to http://lists.openstack.org/pipermail/openstack-dev/2015-March/060174.html | 04:36 |
mrda | recheck didn't work though :/ | 04:37 |
Haomeng | maybe some one change the requirement file, then can not install XStatic-Angular-Irdragndrop>=1.0.2.1 (from horizon==2015.1.dev106) | 04:37 |
Haomeng | rameshg87: ok, let me have a look | 04:37 |
Haomeng | mrda: yes, it can be recreated each time | 04:37 |
Haomeng | mrda: not the testing env unstable issue | 04:37 |
Haomeng | looks like need tripleo guys helping | 04:38 |
Haomeng | and for my test code, I just create new file ironic/tests/test_common.py, because I can not find such existing test file for common code such as the exception testing | 04:39 |
Haomeng | :) | 04:39 |
*** achanda has quit IRC | 04:40 | |
*** achanda has joined #openstack-ironic | 04:40 | |
NobodyCam | lintan_: are you around? | 04:43 |
Haomeng | NobodyCam: good evening:) | 04:44 |
BadCub | heya folks | 04:45 |
NobodyCam | good afternoon Haomeng :) | 04:45 |
rameshg87 | Hello BadCub, NobodyCam | 04:45 |
Haomeng | NobodyCam: :) | 04:45 |
openstackgerrit | Anusha Ramineni proposed openstack/ironic: iLO driver documentation for node cleaning https://review.openstack.org/169162 | 04:45 |
BadCub | hey rameshg87 | 04:45 |
NobodyCam | morning rameshg87 | 04:45 |
mrda | Haomeng: It could have been test_exception.py but I am not fussed | 04:45 |
Haomeng | BadCub: hello:) | 04:45 |
mrda | Welcome back NobodyCam | 04:45 |
BadCub | hey Haomeng :-) | 04:46 |
Haomeng | mrda: ok, let me change in next patch:) | 04:46 |
Haomeng | BadCub: :) | 04:46 |
mrda | and DabDub | 04:46 |
mrda | BadCub | 04:46 |
NobodyCam | morning mrda | 04:46 |
mrda | SO MUCH TYPING FAIL BY ME | 04:46 |
Haomeng | BadCub: you have a cool name badcub:) | 04:46 |
BadCub | hey mrda LOL | 04:46 |
NobodyCam | :) | 04:46 |
Haomeng | mrda: :) | 04:46 |
BadCub | thnx Haomeng :-) | 04:46 |
Haomeng | BadCub: :) | 04:47 |
NobodyCam | rameshg87: your ok with the pxe_amt change? https://review.openstack.org/#/c/167975 ? | 04:47 |
openstackgerrit | Anusha Ramineni proposed openstack/ironic: iLO driver documentation for node cleaning https://review.openstack.org/169162 | 04:47 |
BadCub | so lookin over this new bug. Seems kinda ouchy | 04:48 |
NobodyCam | ramineni: w00 h00 for docs :) | 04:48 |
mrda | BadCub: mine? | 04:48 |
rameshg87 | NobodyCam: yes | 04:48 |
BadCub | https://bugs.launchpad.net/ironic/+bug/1436824 | 04:49 |
openstack | Launchpad bug 1436824 in Ironic "Lost control of nodes with amt driver after user reboot " [Undecided,In progress] - Assigned to Tan Lin (tan-lin-good) | 04:49 |
rameshg87 | NobodyCam: we need to do the same thing for pxe_amt + ipa ramdisk too, but may be better if it was a separate patch in my opinion | 04:49 |
NobodyCam | the linked bug basiclly will exclude partition images | 04:49 |
ramineni | NobodyCam: :) | 04:50 |
NobodyCam | we are going to need a driver matrix as to which drivers suport for capabilities | 04:50 |
NobodyCam | oh fyi meeting i about nine minutes | 04:51 |
NobodyCam | :-p | 04:51 |
*** achanda has quit IRC | 04:51 | |
NobodyCam | s/i/in/ | 04:51 |
openstackgerrit | Haomeng,Wang proposed openstack/ironic: Fix UnicodeEncodeError issue when the language is not en_US https://review.openstack.org/168268 | 04:55 |
Haomeng | mrda: yes, the test_common.py is too *common*:), so accept your good idea, changed to ironic/tests/test_exception.py, done | 04:56 |
mrda | Haomeng: \o/ | 04:56 |
Haomeng | mrda: :) | 04:56 |
jroll | \o | 04:58 |
lintan_ | hi NobodyCam | 05:00 |
jroll | lintan_: meeting is starting :) | 05:01 |
*** kalpase has joined #openstack-ironic | 05:09 | |
NobodyCam | hi lintan_ yep we're in the meeting | 05:11 |
NobodyCam | atm | 05:11 |
lintan_ | NobodyCam , we can talk after meeting :) | 05:12 |
NobodyCam | ya | 05:12 |
*** Nisha has joined #openstack-ironic | 05:15 | |
*** achanda has joined #openstack-ironic | 05:22 | |
*** oomichi_ has quit IRC | 05:29 | |
firl | So I went and migrated my “flat” network to the same management network. Now the DHCP server is giving out the proper IP address, however the TFTP server doesn’t seem to be listening on the interface. “IP 10.0.8.101 > 10.0.8.99: ICMP 10.0.8.101 udp port tftp unreachable, length 91” | 05:36 |
firl | It is joined via a ovs bridge | 05:36 |
firl | Is there a way to have the DHCP for ironic serve up an alternative IP? | 05:37 |
rameshg87 | NobodyCam: i think https://bugs.launchpad.net/ironic/+bug/1436824 says to introduce localboot for pxe_amt driver | 05:49 |
openstack | Launchpad bug 1436824 in Ironic "Lost control of nodes with amt driver after user reboot " [Medium,In progress] - Assigned to Tan Lin (tan-lin-good) | 05:49 |
rameshg87 | NobodyCam: not a very clear description of bug infact :) | 05:50 |
*** pradipta has joined #openstack-ironic | 05:52 | |
*** ukalifon1 has joined #openstack-ironic | 05:52 | |
*** pcrews_ has joined #openstack-ironic | 05:59 | |
*** pcrews has quit IRC | 05:59 | |
*** pcrews_ has quit IRC | 05:59 | |
jroll | rameshg87: so what sort of order does this thing have? e.g. if I put 1,2 | 06:00 |
jroll | is that order guaranteed etc | 06:00 |
rameshg87 | jroll: ilo out-of-band inspection cannot detect the link status | 06:01 |
lintan_ | Hi NobodayCam, | 06:01 |
devananda | rameshg87: ugh. API parameter validation should be in the API, not inthe driver :( | 06:01 |
NobodyCam | lintan_: hi | 06:01 |
devananda | rameshg87: I mean, this needs to be standardized across all drivers | 06:01 |
Haomeng | firl: see some invalid chars with your words | 06:01 |
rameshg87 | jroll: so we request the operator to say what all ports they wanted ironic to enroll as ironic port | 06:01 |
NobodyCam | lintan_: I think rameshg87 asnwered my question | 06:01 |
NobodyCam | above ^^ | 06:01 |
jroll | rameshg87: that doesn't tell me if it's random or not. I don't think an integer is a very good identifier | 06:02 |
rameshg87 | devananda: but we have only one out-of-band driver using this no ? | 06:02 |
devananda | rameshg87: what if another driver accepts the value of "zebra"? now the API is different between drivers and our users are very, very sad | 06:02 |
jroll | idk what link status has to do with it. | 06:02 |
rameshg87 | jroll: in proliant hardware, ports are numbered in the physical hardware in the back of the server | 06:02 |
devananda | rameshg87: what does that have to do with building an API that supports LOTS of drivers doing it? | 06:02 |
rameshg87 | devananda: yeah :( | 06:03 |
rameshg87 | Nisha: ^^^ | 06:03 |
devananda | perhaps I should be a whole lot clearer perhaps that we shouldn't accept features when only one driver implements them ... because omg the pain if another driver implements the API differently :( | 06:03 |
lintan_ | OK, thanks NobodayCam and rameshg87 | 06:03 |
devananda | rameshg87: wait, the integer number here indicates which physical port to inspect in the chassis???!?! | 06:04 |
jroll | yes. | 06:04 |
jroll | my servers don't have numbers on the back, just mac address | 06:04 |
jroll | RIP | 06:04 |
* devananda tries to inspect port number 999999 | 06:04 | |
rameshg87 | devananda: which all physical ports in the server to enroll as ironic port | 06:05 |
devananda | rameshg87: True or False are the only values i can imagine having any meaning here | 06:05 |
devananda | either "find and update the ports" or "do not..." | 06:06 |
jroll | devananda: I could see mac addresses being almost sane | 06:06 |
rameshg87 | devananda: jroll: the context of the problem was that - we cannot detect whether a physical cable is plugged into a nic or not | 06:06 |
jroll | or I guess that's what is being discovered. womp. | 06:06 |
devananda | jroll: presumably ... right. | 06:06 |
rameshg87 | devananda: jroll: from ilo | 06:06 |
jroll | rameshg87: I fail to see what link status has to do with this :/ | 06:06 |
*** pcrews has joined #openstack-ironic | 06:06 | |
devananda | jroll: me too | 06:06 |
rameshg87 | devananda: jroll: if we enroll all the nics, that might lead to deploy failures - because nova maps to neutron networks randomly | 06:07 |
devananda | oh! | 06:07 |
devananda | so you're working around the multi-nic bug agove | 06:07 |
devananda | *above | 06:07 |
rameshg87 | devananda: unfortunately yes :( | 06:07 |
devananda | by implementing the work around in the API and requiring users to know which port (#1, #2, #3...) is plugged in? | 06:07 |
jroll | so someone is physically checking what NICs are plugged in | 06:07 |
jroll | blahhhhhh | 06:07 |
devananda | rameshg87: rather than having tried to fix the prolem? | 06:07 |
*** achanda has quit IRC | 06:08 | |
*** achanda has joined #openstack-ironic | 06:08 | |
* BadCub heads off_to_bed. | 06:08 | |
*** BadCub is now known as BadCub_Hiding | 06:08 | |
* devananda has a sad | 06:08 | |
rameshg87 | devananda: jroll: if they still want to enroll all the nic, they can do with "inspect_ports=all" | 06:08 |
jroll | sure. | 06:09 |
* rameshg87 wonders if inspect_ports is not conveying the meaning at all | 06:09 | |
jroll | I guess this isn't horribly far off of how we register hardware | 06:09 |
openstackgerrit | Tan Lin proposed openstack/ironic: Enable pxe_amt driver with localboot https://review.openstack.org/167975 | 06:10 |
rameshg87 | :( | 06:10 |
jroll | rameshg87: what I mean is, this is similar to how we register hardware in our environment | 06:10 |
jroll | just someone writes down numbers rather than mac addresses. | 06:11 |
NobodyCam | lintan_: can you update teh bug with the info in your reply to me on that patch :) | 06:11 |
rameshg87 | jroll: that was to work around the link-status issue and nova-ironic-neutron-port-mapping issue | 06:12 |
NobodyCam | and I need to head to sleep land | 06:12 |
NobodyCam | so good night all | 06:12 |
rameshg87 | jroll: the issue was the ilo doesn't support fetching the link status | 06:12 |
jroll | rameshg87: no, I get it. it's just... not ideal. | 06:12 |
rameshg87 | jroll: and without knowing where physical cables are plugged, it's even more dangerous to enroll all the ports as that might silently lead to deploy failures :( | 06:13 |
openstackgerrit | OpenStack Proposal Bot proposed openstack/ironic: Imported Translations from Transifex https://review.openstack.org/169184 | 06:13 |
jroll | rameshg87: right, we should make that part better | 06:13 |
devananda | rameshg87: there is nothing about this port-number-thing in the spec http://specs.openstack.org/openstack/ironic-specs/specs/kilo/ironic-node-properties-discovery.html | 06:13 |
rameshg87 | jroll: yes | 06:13 |
*** ijw has joined #openstack-ironic | 06:13 | |
devananda | g'night, NobodyCam | 06:13 |
rameshg87 | Nisha: are you around ? | 06:14 |
rameshg87 | devananda: i think this was lately identified | 06:14 |
devananda | rameshg87: what would be the impact if we remove it now (ie, before RC1) | 06:14 |
rameshg87 | devananda: and hence spec wasn't updated | 06:14 |
rameshg87 | devananda: might start leading to deploy failures | 06:15 |
*** Marga_ has quit IRC | 06:15 | |
rameshg87 | devananda: because we would end up enrolling all the ports | 06:15 |
devananda | rameshg87: well. if the down'd ports are removed before we deploy, then it's fine | 06:15 |
openstackgerrit | Michael Davies proposed openstack/ironic: Fix is_hostname_safe for RFC compliance https://review.openstack.org/165666 | 06:15 |
devananda | rameshg87: you can discover htem just fine. the problem is having them active at deploy time | 06:15 |
devananda | because nova picks them up and then fails, modulo some randomness from that bug | 06:16 |
rameshg87 | devananda: what do you mean by down'd the port ? | 06:16 |
lintan_ | done, NobodyCam :) | 06:16 |
rameshg87 | devananda: can we down an ironic port ? | 06:16 |
devananda | i mean the ports that dont have link status UP | 06:16 |
devananda | no. but we can DELETE them : | 06:16 |
devananda | :) | 06:16 |
rameshg87 | devananda: oh okay | 06:16 |
rameshg87 | devananda: so delete them after inspection ? | 06:16 |
devananda | I mean, that's not great either | 06:17 |
*** Nisha_away has joined #openstack-ironic | 06:17 | |
*** Nisha has quit IRC | 06:17 | |
rameshg87 | do you mean let operator delete them manually ? | 06:17 |
devananda | yes | 06:17 |
devananda | find all the ports then remove ones that are not connected to the network | 06:17 |
rameshg87 | devananda: but that might lead to the same work no ? know which ports are connected after inspection | 06:18 |
devananda | neither of these are good solutions, but this doesn't introduce a driver-specific non-discoverable non-conventional non-documented thing in the API | 06:18 |
devananda | clearly the right thing to do is to fix nova | 06:18 |
rameshg87 | devananda: but i idk how we would fix up nova for this | 06:19 |
rameshg87 | devananda: i think we are supposed to enroll only the nics that have physical connectivity | 06:19 |
rameshg87 | devananda: the base problem was that ilo cannot detect which nics have physical connectivity :( | 06:19 |
rameshg87 | devananda: so in my mind - either we enroll only the ports which operator requests OR enroll all ports and let operator decide which ones to remove | 06:21 |
* jroll goes off to bed, good luck with this one | 06:22 | |
rameshg87 | good night jroll | 06:22 |
devananda | the problem is that nova is picking one at random to set up the networking. see prior discussion of https://bugs.launchpad.net/ironic/+bug/1405131 | 06:23 |
openstack | Launchpad bug 1405131 in OpenStack Compute (nova) "Ports cannot be mapped to networks" [Low,In progress] - Assigned to Mark Goddard (mgoddard) | 06:23 |
rameshg87 | devananda: yeah, may be solution of https://bugs.launchpad.net/ironic/+bug/1405131 might end up fixing this problem | 06:24 |
openstack | Launchpad bug 1405131 in OpenStack Compute (nova) "Ports cannot be mapped to networks" [Low,In progress] - Assigned to Mark Goddard (mgoddard) | 06:24 |
rameshg87 | devananda: but (correct me if i am wrong) any rate are we are not supposed to enroll ports that don't have physical connectivity, right ? | 06:25 |
*** kalpase has quit IRC | 06:27 | |
*** mrda is now known as mrda-away | 06:27 | |
devananda | well, i was just going to say this bug should be easy to reproduce | 06:28 |
devananda | just enroll one MAC that works, and a second one that doesn't exist | 06:28 |
devananda | and see how often "nova boot" succeeds | 06:28 |
rameshg87 | devananda: yeah | 06:28 |
rameshg87 | half of the time shouldn't succeed | 06:28 |
devananda | right | 06:28 |
devananda | because random() is bad | 06:28 |
rameshg87 | yeah :) | 06:29 |
rameshg87 | devananda: so what best can be done for now for ilo inspection ? | 06:29 |
devananda | rameshg87: but think of this -- I'm the remote hands in the data center, I need to replace a switch. right now only one cable is plugged in (that's bad, but hey, i'm cheap) | 06:30 |
devananda | rameshg87: so I cable the second port, then unplug the first one | 06:30 |
devananda | boom. nothing can be deployed | 06:30 |
rameshg87 | devananda: yes :( | 06:30 |
devananda | that's never what I want | 06:30 |
*** kalpase has joined #openstack-ironic | 06:30 | |
rameshg87 | devananda: discoverd will be able to correct this - operator just inspects the node again after replugging the cable | 06:31 |
rameshg87 | devananda: discoverd enrolls the right ports | 06:31 |
devananda | nova & neutron should be setting up dhcp options for ALL ports, not first(port) or random(port) | 06:31 |
rameshg87 | devananda: but then dhcp options for all ports in provisioning network ? | 06:32 |
devananda | *all ports that are configured to DHCP | 06:32 |
rameshg87 | devananda: i mean create a neutron port in the provisioning network for all ironic ports | 06:32 |
devananda | which in some cases may in fact be all of tem | 06:32 |
devananda | them | 06:32 |
rameshg87 | devananda: but right know we don't isolate deployment network from tenant network | 06:33 |
rameshg87 | devananda: so that would end up connectivity to only deployment network during provisioning | 06:34 |
rameshg87 | devananda: and connectivity to tenant networks after instance is active | 06:34 |
rameshg87 | devananda: right ? | 06:34 |
devananda | rameshg87: some users dont need to separate the provisioning & tenant nets. some do. | 06:36 |
*** spandhe has quit IRC | 06:36 | |
rameshg87 | devananda: sorry, need to goto a meeting now in office. i will catch up with you tonight on this if you get away before i come back. | 06:37 |
devananda | rameshg87: k k | 06:37 |
*** dlpartain has joined #openstack-ironic | 06:37 | |
*** rameshg87 is now known as rameshg87-away | 06:37 | |
*** dlpartain has left #openstack-ironic | 06:38 | |
*** pas-ha has joined #openstack-ironic | 06:41 | |
*** takadayuiko has joined #openstack-ironic | 06:42 | |
*** saripurigopi has quit IRC | 06:46 | |
*** rameshg87-away is now known as rameshg87 | 06:46 | |
rameshg87 | devananda: i am back and ready to continue discussion if you are available | 06:47 |
devananda | that was a quick meeting. howeve,r i need to get up and get coffee and head into the office soon | 06:48 |
*** jamielennox is now known as jamielennox|away | 06:49 | |
rameshg87 | devananda: :). okay, may be i will catch up with you when are you free. | 06:49 |
rameshg87 | some time later today | 06:49 |
*** ifarkas has joined #openstack-ironic | 06:56 | |
devananda | rameshg87: https://bugs.launchpad.net/ironic/+bug/1438546 | 07:00 |
openstack | Launchpad bug 1438546 in Ironic "ilo driver inspection requires extra parameters" [Undecided,New] | 07:00 |
* devananda goes afk to get coffee, etc | 07:00 | |
*** Nisha_away has quit IRC | 07:01 | |
*** achanda has quit IRC | 07:05 | |
rameshg87 | devananda: okay, so we would rather take off the inspect_ports thing, right ? | 07:05 |
*** saripurigopi has joined #openstack-ironic | 07:17 | |
*** zhenguo has joined #openstack-ironic | 07:43 | |
*** jistr has joined #openstack-ironic | 07:43 | |
*** _lintan has joined #openstack-ironic | 07:48 | |
_lintan | hi rameshg87 | 07:49 |
rameshg87 | hey lintan_ | 07:51 |
rameshg87 | lintan_: i am just leaving my desk, will be back in a short while | 07:51 |
rameshg87 | can we talk after that ? | 07:52 |
_lintan | rameshg87, for sure :) | 07:53 |
*** mgoddard has joined #openstack-ironic | 07:58 | |
*** viktors|afk is now known as viktors | 08:01 | |
*** romcheg has joined #openstack-ironic | 08:01 | |
rameshg87 | lintan_: hi | 08:12 |
rameshg87 | lintan_: i am back :) | 08:12 |
viktors | hi folks! Is there any instruction, how to run ironic-api with apache2? | 08:13 |
*** yuanying has quit IRC | 08:13 | |
_lintan | rameshg87, I have a question about how to use IPA-Ramdisk | 08:14 |
rameshg87 | lintan_: sure .. | 08:14 |
rameshg87 | go ahead | 08:15 |
*** ndipanov has joined #openstack-ironic | 08:15 | |
rameshg87 | viktors: what do you meanby running "ironic-api with apache2" ? | 08:15 |
_lintan | rameshg87, I download these two images from https://github.com/openstack/ironic-python-agent/tree/master/imagebuild/coreos, Can I use them as replace with dib-ramdisk? | 08:15 |
_lintan | rameshg87, or I have to build IPA driver with devstack? | 08:16 |
rameshg87 | lintan_: yes, you can build a new ipa ramdisk by running make on imagebuild/coreos | 08:17 |
rameshg87 | lintan_: alternately you can use prebuilt ramdisk available here: http://tarballs.openstack.org/ironic-python-agent/coreos/files/ | 08:17 |
viktors | rameshg87: i'm looking for instruction similar to http://docs.openstack.org/developer/keystone/apache-httpd.html | 08:18 |
rameshg87 | viktors: are you looking for this - https://github.com/openstack/ironic/blob/master/etc/apache2/ironic ? | 08:20 |
*** lucasagomes has joined #openstack-ironic | 08:20 | |
ifarkas | devananda, hi, could you please take a look at https://review.openstack.org/#/c/165860/ I refactored the patch to avoid introducing new config options and the bug is targeted for rc1 | 08:20 |
rameshg87 | i haven't done this myself, but it might be what you are looking for | 08:20 |
_lintan | rameshg87, Yes, this is what I mean, I download the prebuild ramdisks. My question is can I directly replace dib-ramdisk with IPA-Ramdisk in my previous pxe-* environment. It looks like agent-* driver need swift support | 08:21 |
viktors | rameshg87: yes thats correct config, but I'm not familiar with apache, so I looked for some instruction, if any :) | 08:21 |
rameshg87 | lintan_: yes, you can do that | 08:21 |
rameshg87 | :) | 08:21 |
_lintan | Cool, thanks. | 08:21 |
rameshg87 | lintan_: no need of swift or any extra stuffs | 08:21 |
rameshg87 | just replace dib ramdisk with ipa ramdisk | 08:21 |
rameshg87 | that's all | 08:21 |
rameshg87 | credits to lucasagomes :) | 08:21 |
rameshg87 | morning lucasagomes :) | 08:22 |
* lucasagomes is lost heh | 08:22 | |
lucasagomes | rameshg87, good morning! | 08:22 |
lucasagomes | morning all | 08:22 |
_lintan | Morning lucasagomes | 08:22 |
openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Add config drive documentation https://review.openstack.org/166201 | 08:23 |
rameshg87 | lucasagomes: lintan_ was asking if it will work for pxe_* drivers if we replace dib ramdisk with ipa ramdisk | 08:23 |
lucasagomes | yeah :) cool | 08:23 |
lucasagomes | ipa ftw | 08:23 |
_lintan | Really awesome work, lucasagomes | 08:24 |
lucasagomes | _lintan, thanks, it's everyone's work :) | 08:25 |
rameshg87 | viktors: i am not sure if there's something more from ironic documentation | 08:25 |
rameshg87 | viktors: i haven't done it myself too | 08:26 |
* rameshg87 never worked on anything other than devstack configuration :( | 08:26 | |
rameshg87 | viktors: if you had some other specific questions on configuring apache, may be you can check with other folks ehre | 08:26 |
rameshg87 | lucasagomes: any updates on that grub2 stuff ? | 08:28 |
rameshg87 | lucasagomes: i am still not able to pack both bios and efi grub binaries on same image | 08:28 |
viktors | rameshg87: ok, thanks! I know, that devananda added apache config, maybe he knows something... | 08:28 |
lucasagomes | rameshg87, not really I tried some things but had to change the focus yesterday on something else | 08:28 |
lucasagomes | installing like: DEBIAN_FRONTEND=noninteractive apt-get install -y -o Dpkg::Options::="--force-confdef" -o Dpkg::Options::="--force-confold" grub2 | 08:28 |
lucasagomes | is non-interactive, which is good | 08:29 |
rameshg87 | lucasagomes: oh okay | 08:29 |
lucasagomes | but DIB complain :/ | 08:29 |
rameshg87 | :( | 08:29 |
*** tiagogomes_ has joined #openstack-ironic | 08:29 | |
lucasagomes | the grub-probe command says /dev is not mounted | 08:29 |
lucasagomes | so I have to debug it | 08:29 |
rameshg87 | okay, let me try that too | 08:29 |
*** derekh has joined #openstack-ironic | 08:29 | |
lucasagomes | ok, lemme get some coffee | 08:30 |
viktors | mrda-away: Hi! Please ping me as for bug 1438531, when you'll be here | 08:33 |
openstack | bug 1438531 in Ironic "Ironic database migration tests not running" [Undecided,New] https://launchpad.net/bugs/1438531 - Assigned to Michael Davies (mrda) | 08:33 |
rameshg87 | lucasagomes: please check the today's meeting minutes and let me know what you think | 08:35 |
rameshg87 | lucasagomes: we decided to give a try fixing the issue that dib ramdisk doesn't report back the status of bootloader installation | 08:35 |
rameshg87 | lucasagomes: and bug is on me right now | 08:36 |
rameshg87 | :) | 08:36 |
lucasagomes | rameshg87, will do | 08:39 |
lucasagomes | rameshg87, but I'm +1 on fixing that | 08:39 |
lucasagomes | ur adding a new endpoint to the pxe vnedor interface right? | 08:39 |
rameshg87 | lucasagomes: yes | 08:39 |
lucasagomes | that will be called after the bootloader is installed or something | 08:39 |
rameshg87 | lucasagomes: yes | 08:39 |
lucasagomes | make sure we don't reboot on the ramdisk as well, let's ironic do the reboot | 08:39 |
rameshg87 | lucasagomes: in all cases ? | 08:40 |
rameshg87 | lucasagomes: the second call back will be only for localboot | 08:40 |
rameshg87 | lucasagomes: and we will wait for second call back only if boot_option == "local" | 08:41 |
lucasagomes | rameshg87, oh I see, could be only for bootloader | 08:41 |
rameshg87 | lucasagomes: in that way i think we can preserve backward compatibility | 08:41 |
lucasagomes | I think it would be good to be all cases. Because for example our reboot in ironic is a power off and power on again | 08:41 |
lucasagomes | which is good | 08:41 |
lucasagomes | when you have VMs, if you only reboot from within the ramdisk some changes may not get applied, including the boot device | 08:41 |
lucasagomes | on my test env it's a pain | 08:41 |
rameshg87 | lucasagomes: but old dib ramdisks don't know it has to do second call back | 08:41 |
lucasagomes | because virsh only applies the confg changes after you power off that vm >.< | 08:42 |
lucasagomes | rameshg87, oh I see | 08:42 |
rameshg87 | oh | 08:42 |
lucasagomes | rameshg87, well that's good anyway, if it's only for localboot that would solve the problem already | 08:42 |
rameshg87 | lucasagomes: can we preserve backward compatibility if we do that for all cases ? | 08:42 |
lucasagomes | rameshg87, hmm I don't think so | 08:42 |
lucasagomes | because before it just called pass_deploy_info and then rebooted | 08:43 |
rameshg87 | lucasagomes: yeah | 08:43 |
lucasagomes | so maybe yeah making it only for the bootloader seems fair | 08:43 |
rameshg87 | lucasagomes: yeah, will do that | 08:43 |
rameshg87 | lucasagomes: make second call back from ramdisk if boot_option == "local" | 08:43 |
lucasagomes | rameshg87, good stuff, thanks! | 08:43 |
lucasagomes | rameshg87, yeah sounds good | 08:43 |
rameshg87 | okay | 08:43 |
rameshg87 | thanks :) | 08:43 |
lucasagomes | thank u for fixing it :D | 08:44 |
lucasagomes | gate seems broken | 08:44 |
*** dtantsur|afk is now known as dtantsur | 08:44 | |
dtantsur | morning Ironic | 08:45 |
lucasagomes | it looks likes it's devstack | 08:45 |
*** pelix has joined #openstack-ironic | 08:46 | |
dtantsur | lucasagomes, what's with gate? | 08:46 |
lucasagomes | dtantsur, not sure, just saw it now | 08:46 |
lucasagomes | it seems something in devstack has broken all jobs | 08:46 |
dtantsur | oh | 08:47 |
yuriyz | morning Ironic | 08:47 |
dtantsur | yuriyz, o/ | 08:47 |
*** _lintan has quit IRC | 08:49 | |
*** yog__ has joined #openstack-ironic | 08:53 | |
-openstackstatus- NOTICE: CI Check/Gate pipelines currently stuck due to a bad dependency creeping in the system. No need to recheck your patches at the moment. | 08:53 | |
*** ChanServ changes topic to "CI Check/Gate pipelines currently stuck due to a bad dependency creeping in the system. No need to recheck your patches at the moment." | 08:53 | |
rameshg87 | there you go ^^^^ :) | 08:54 |
*** athomas has joined #openstack-ironic | 08:57 | |
openstackgerrit | Yuiko Takada proposed stackforge/ironic-discoverd: Add unit tests for firewall.py https://review.openstack.org/168304 | 08:57 |
rameshg87 | dtantsur: hi | 08:58 |
rameshg87 | dtantsur: wanted to ask something regarding discoverd | 08:58 |
rameshg87 | dtantsur: are we creating ironic ports for all the nics in the system ? | 08:59 |
rameshg87 | (may be i asked it before) | 08:59 |
rameshg87 | there was a discussion surrounding this in the meeting (not w.r.t discoverd) | 08:59 |
*** jiangfei|2 has joined #openstack-ironic | 09:01 | |
dtantsur | rameshg87, it's configurable for discoverd | 09:02 |
* dtantsur hasn't looked at meeting minutes yet | 09:02 | |
dtantsur | rameshg87, default being: only one PXE booted port | 09:02 |
rameshg87 | dtantsur: okay, so they have option to configure so that all ports are enrolled as well, right ? | 09:03 |
dtantsur | rameshg87, exactly | 09:03 |
rameshg87 | dtantsur: there was this bug https://bugs.launchpad.net/ironic/+bug/1405131 | 09:04 |
openstack | Launchpad bug 1405131 in OpenStack Compute (nova) "Ports cannot be mapped to networks" [Low,In progress] - Assigned to Mark Goddard (mgoddard) | 09:04 |
rameshg87 | dtantsur: which was being discussed | 09:04 |
dtantsur | I see, yeah | 09:04 |
*** jiangfei has quit IRC | 09:04 | |
rameshg87 | dtantsur: not to be fixed on kilo, but a solution to this might bring in a solution which works when all ports are enrolled | 09:04 |
*** rameshg87 is now known as rameshg87-brb | 09:05 | |
dtantsur | rameshg87-brb, then we'll change the default in discoverd and adopt this solution :) | 09:05 |
*** jiangfei|3 has joined #openstack-ironic | 09:06 | |
*** jiangfei|2 has quit IRC | 09:06 | |
lucasagomes | ok the bug seems to affect all projects | 09:10 |
*** rameshg87-brb is now known as rameshg87 | 09:11 | |
rameshg87 | lucasagomes: dtantsur: one question regarding proliantutils | 09:11 |
rameshg87 | we are planning to release proliantutils==3.0.0 with ironic kilo | 09:11 |
rameshg87 | we are planning to delay that till april 9 so that we can fix any bugs that we may find in it | 09:12 |
rameshg87 | is it okay to update ironic now only saying we need proliantutils>=3.0.0 without it being released officially :D | 09:12 |
rameshg87 | i mean update ironic 3-4 days before releasing 3.0.0 | 09:12 |
lucasagomes | rameshg87, that's a documentation? | 09:13 |
rameshg87 | lucasagomes: yes | 09:13 |
rameshg87 | lucasagomes: we want to delay it till april 9, but can't wait till last moment for ironic doc change | 09:13 |
lucasagomes | it's odd, but since we have a time constraint in Ironic I think it's ok | 09:14 |
rameshg87 | lucasagomes: only constraint is driver-requirements for whoever packages it | 09:14 |
lucasagomes | seems fine | 09:14 |
rameshg87 | lucasagomes: they might depend on the kilo-released-version of ironic | 09:14 |
rameshg87 | okay | 09:14 |
*** pradipta has quit IRC | 09:20 | |
*** pradipta has joined #openstack-ironic | 09:24 | |
dtantsur | rameshg87, I would say: release on Apr 8, make people land version bump | 09:24 |
dtantsur | depending on non-existing version might be confusing | 09:24 |
dtantsur | though I don't have a string opinion on this | 09:25 |
rameshg87 | dtantsur: okay | 09:26 |
rameshg87 | dtantsur: may be apr 7 or apr 8, i can keep bugging people on irc for this :) | 09:26 |
dtantsur | yep | 09:26 |
dtantsur | I'll be releasing discoverd 1.1 on Apr 9, but it's not a hard dep for Kilo Ironic | 09:26 |
devananda | ifarkas: done. thanks for the ping | 09:36 |
*** mgoddard has quit IRC | 09:36 | |
ifarkas | devananda, thank for the update! | 09:36 |
devananda | lucasagomes, dtantsur: related to discovery -- https://bugs.launchpad.net/ironic/+bug/1438546 | 09:37 |
openstack | Launchpad bug 1438546 in Ironic "ilo driver inspection requires extra parameters" [Undecided,New] | 09:37 |
lucasagomes | devananda, morning | 09:37 |
* lucasagomes looks | 09:37 | |
devananda | i filed after discussion with rameshg87 this morning | 09:37 |
dtantsur | devananda, hmm yeah. discoverd works around the situation somewhat differently | 09:38 |
dtantsur | devananda, I agree that's bad situation, but I've no idea how to fix until until https://bugs.launchpad.net/ironic/+bug/1405131 is fixed | 09:38 |
openstack | Launchpad bug 1405131 in OpenStack Compute (nova) "Ports cannot be mapped to networks" [Low,In progress] - Assigned to Mark Goddard (mgoddard) | 09:38 |
devananda | right | 09:39 |
rameshg87 | dtantsur: same with ilo :( | 09:39 |
rameshg87 | devananda: dtantsur: so may be we should get it fixed soon in liberty | 09:39 |
dtantsur | rameshg87, yeah, it's about ilo. discoverd, being in-band, can at least detect which port we can PXE boot from | 09:39 |
devananda | so the approach you've both taken is, effectively, dont register ports that are offline // only register one port | 09:39 |
dtantsur | rameshg87, we MUST :) | 09:39 |
dtantsur | devananda, right | 09:40 |
rameshg87 | devananda: yeah exactly :) | 09:40 |
dtantsur | devananda, (for discoverd is configurable) | 09:40 |
devananda | but the way this is done is different in each driver, and this difference is exposed in the API <<< that's what bothers me | 09:40 |
devananda | for a standard interface, every driver MUST present the same API | 09:40 |
*** jamielennox|away is now known as jamielennox | 09:41 | |
*** romcheg has quit IRC | 09:41 | |
dtantsur | devananda, right... I wonder how to fix it though | 09:42 |
dtantsur | I could potentially make discoverd accept port list too. but I've no idea what these "port numbers" mean | 09:42 |
devananda | dtantsur: you and most operators | 09:43 |
dtantsur | :) | 09:43 |
devananda | IMO, we should really not have landed that "port number" thing | 09:43 |
devananda | I understand the bug and the need for a way to work around it for now, but we ought to have caught that in the review | 09:44 |
devananda | even on HP hardware, I'm not sure what it means | 09:44 |
dtantsur | devananda, yeah :( then we can't land ilo (any OOB?) inspection though | 09:44 |
devananda | dtantsur: sure we can. it discovers all ports | 09:44 |
* lucasagomes remembers seem that port numbers list, I thought it was some iLO thing | 09:44 | |
*** romcheg has joined #openstack-ironic | 09:45 | |
dtantsur | devananda, and then we ask users to delete "the wrong ones"? | 09:45 |
devananda | if the operator knows that some ports are not connected, ... yup | 09:45 |
rameshg87 | yeah that's the only other way | 09:45 |
lucasagomes | rameshg87, what informations you get from the ports in iLO? | 09:46 |
devananda | dtantsur: does discoverd support any option to say "dont enroll ports" ? | 09:46 |
lucasagomes | maybe we should add the ability of passing a regex to Ironic | 09:46 |
lucasagomes | like if the port is connect to the network X | 09:47 |
lucasagomes | register it in Ironic | 09:47 |
rameshg87 | lucasagomes: the port number is actually marked where the lan cable is plugged | 09:47 |
*** jamielennox is now known as jamielennox|away | 09:47 | |
lucasagomes | ignore all the rest | 09:47 |
devananda | rameshg87: what about in a blade system? | 09:47 |
dtantsur | devananda, it does not, probably it should. what's the use case? | 09:47 |
lucasagomes | rameshg87, right | 09:47 |
rameshg87 | devananda: :D | 09:47 |
dtantsur | lucasagomes, "the port is connect to the network X" any ideas how to implement OOB? | 09:47 |
devananda | dtantsur: I actually dont have a use case. I think Ironic should always, when inspecting, update the ports. | 09:47 |
rameshg87 | devananda: when enclosure comes it's a different story :( | 09:48 |
lucasagomes | dtantsur, so that's why I asked which information they get frm the ports OOB | 09:48 |
devananda | dtantsur: but the ilo driver has three options here: update all ports, do not update any ports, only update ports [...] | 09:48 |
lucasagomes | if the network it's connected is one, fine otherwise :/ | 09:48 |
devananda | rameshg87: we already have folkstrying to use this with blades. "when" isn't the issue... | 09:48 |
dtantsur | devananda, oh. and again it's different behavior... | 09:48 |
devananda | rameshg87: the API for any solution we implement today needs to account for what other situations we will meet tomorrow | 09:49 |
dtantsur | that's probably my fault, I should have reviewed the whole inspection thingy better... | 09:49 |
devananda | the hard thing about making APIs is that you cant change them without breaking someone | 09:49 |
rameshg87 | yeah i understand now | 09:49 |
rameshg87 | we SHOULD make it consistent | 09:50 |
rameshg87 | devananda: but then ilo inspection will have to remain with the same behaviour - either UPDATE all ports OR UPDATE none | 09:50 |
devananda | dtantsur: discoverd always updates the ports (the ones it finds) right? | 09:51 |
devananda | as far as I can tell, that's what the spec says all drivers should do | 09:51 |
devananda | thre's nothing here http://specs.openstack.org/openstack/ironic-specs/specs/kilo/ironic-node-properties-discovery.html#rest-api-impact | 09:52 |
dtantsur | devananda, "update" = "create"? yes. it can optionally delete missing ports (this "optionally" is one more behavior difference) | 09:52 |
devananda | that says "pass this extra option to skip discovering ports" | 09:52 |
devananda | dtantsur: :( | 09:52 |
devananda | nothing here either about optionally doing something else http://specs.openstack.org/openstack/ironic-specs/specs/kilo/ironic-node-properties-discovery.html#driver-api-impact | 09:52 |
dtantsur | devananda, it's not late to make it default to "delete missing" for 1.1. If you feel like filing a but against https://bugs.launchpad.net/ironic-discoverd I'll change the default to strictly match Ironic behavior | 09:53 |
dtantsur | still, I can't make "add all ports" the default, because it's the broken default :( | 09:54 |
devananda | nothing here either - http://specs.openstack.org/openstack/ironic-specs/specs/kilo/inband-properties-discovery.html | 09:54 |
devananda | dtantsur: from an operator standpoint, if I ask Ironic to "inspect" the hardware, and it comes back saying there is only one NIC in the machine, when I know there are two ... | 09:55 |
devananda | dtantsur: I'm going to think that the second NIC is bad. or that Ironic is defective. | 09:55 |
dtantsur | devananda, and if we add both ports and then deployment fails for vague reason, what is operator supposed to think? | 09:56 |
dtantsur | I mean, Ironic (and Nova) IS broken: it's affected by bug https://bugs.launchpad.net/ironic/+bug/1405131 | 09:57 |
openstack | Launchpad bug 1405131 in OpenStack Compute (nova) "Ports cannot be mapped to networks" [Low,In progress] - Assigned to Mark Goddard (mgoddard) | 09:57 |
devananda | oh, totally | 09:57 |
devananda | that's really a much bigger bug than it looked like when mgoddard first mentioned it to me | 09:57 |
devananda | I have mistakenly thought that nova/neutron was setting the dhcp boot options on all known ports, not random(1) | 09:58 |
*** naohirot has quit IRC | 09:58 | |
devananda | but both of the proposed fixes here create additional problems | 09:59 |
devananda | with discoverd, i could get different results if I run inspect with different cables plugged in | 09:59 |
devananda | and no changes to the hardware being inspected | 09:59 |
devananda | and with ilo, inspect will flat-out error if I dont pass in this additional parameter | 10:00 |
devananda | if I'm writing some automation to inspect my hardware, I have to special case what driver I'm using | 10:01 |
rameshg87 | :( | 10:01 |
dtantsur | devananda, is it correct that we treat Ironic ports as equivalent to NIC inventory? | 10:02 |
dtantsur | I mean, maybe " i could get different results if I run inspect with different cables plugged in" is the correct thing to do? | 10:02 |
devananda | dtantsur: i believe so, yes | 10:02 |
dtantsur | especially while the bug in question is still present | 10:02 |
devananda | dtantsur: ironic should represent the hardware effectively | 10:02 |
devananda | dtantsur: and it's quite reasonable for an operator to temporarily disconnect some network cables, pull a switch out, etc | 10:03 |
devananda | as part of routine maintenance | 10:03 |
openstackgerrit | Merged stackforge/ironic-discoverd: Add unit tests for firewall.py https://review.openstack.org/168304 | 10:03 |
dtantsur | aren't we getting to CMDB border here? if we need one port, why add all present? | 10:03 |
openstackgerrit | Dmitry Tantsur proposed stackforge/ironic-discoverd: First take on integration test for using on devstack https://review.openstack.org/169241 | 10:03 |
devananda | dtantsur: CMDB typically holds a lot more information than just "what is my hardware" | 10:03 |
*** jiangfei|3 has quit IRC | 10:04 | |
*** jiangfei|3 has joined #openstack-ironic | 10:04 | |
dtantsur | devananda, last summer we agreed that trying to represent hardware information that we're not using means crossing CMDB borders | 10:04 |
devananda | dtantsur: why NOT add all ports? we should use them all ... | 10:04 |
dtantsur | devananda, what for? especially if only one can PXE boot? | 10:05 |
devananda | if all the ports are cabled, and on the same network, then ironic can use all of them | 10:05 |
dtantsur | that's a strong assumptions for bare metal | 10:05 |
devananda | certainly the user who is deploying an instance to a node with 4 ports may want to, you know, bond them for HA and extra throughput | 10:05 |
*** achanda has joined #openstack-ironic | 10:06 | |
dtantsur | are we planning to support this case in the near future? | 10:06 |
devananda | and as the operator of that cloud, I kinda want to know what all the MAC addresses are so that neutron puts them all on the right VLAN | 10:06 |
devananda | uh... i thought we already did | 10:06 |
*** chenglch has quit IRC | 10:06 | |
dtantsur | I'm not sure, but it doesn't sound easy | 10:06 |
devananda | dtantsur: you're saying ironic doesn't support more than one NIC per node?? | 10:06 |
devananda | I'm quite sure it does | 10:06 |
dtantsur | I'm not sure what's required for bonding... | 10:07 |
dtantsur | Ironic support more than one NIC only if they are 1. on one network; 2. all can PXE boot | 10:07 |
dtantsur | referring to the same bug as above | 10:07 |
devananda | does not support != affected by bug | 10:07 |
devananda | we need to support servers | 10:08 |
devananda | every server I've used in the last 15 years has had >1 NIC | 10:08 |
dtantsur | I agree. But we have problems right now, and they are pretty hard to solve in a proper way | 10:09 |
*** pradipta has quit IRC | 10:09 | |
devananda | some times, only one NIC can PXE boot. some times, more than one can. that needs to be (and isn't right now) accounted for. but Ironic needs to model the hardware that's there, be able to inform neutron (or other network providers) about what MACs belong to which machine, etc | 10:09 |
*** jcoufal has joined #openstack-ironic | 10:09 | |
devananda | I agree. but hard or easy doesn't mean we should not try | 10:09 |
dtantsur | we should. but let's get back to Kilo situation | 10:10 |
devananda | right | 10:10 |
*** achanda has quit IRC | 10:10 | |
dtantsur | we're not fixing it in Kilo, and we have some workaround on the table | 10:10 |
devananda | if discoverd only finds MACs that are cabled and have LINK UP right now, I think that's a bug, however, I recognize that it's a workaround, and it doesn't affect the API | 10:11 |
devananda | the way ilo is wokring around this is, IMO, incompatible, because it's changed the API | 10:11 |
rameshg87 | devananda: yeah, now it requires driver_info['inspect_ports'] to be filled in before triggering inspect | 10:12 |
devananda | if both drivers enrolled all NICs, and we document and widely advertise "hey, discovery will find all the NICs, but if they're on separate networks, E_BUG. you should delete some manually" | 10:13 |
devananda | then we have a consistent API, which will still work after we fix the bug, and clear messaging to users as to why there is a work around | 10:13 |
devananda | none of these options are perfect, I know | 10:13 |
rameshg87 | devananda: one sec, let me confirm | 10:13 |
rameshg87 | devananda: all nics = nics which have link up + nics which have link down | 10:14 |
rameshg87 | right ? | 10:14 |
devananda | rameshg87: yes | 10:14 |
rameshg87 | okay | 10:14 |
dtantsur | devananda, then on my discoverd side I will update inspection docs saying one should put "add_ports = all" to the discoverd.conf to match Ironic expectations | 10:15 |
devananda | dtantsur: actually I think that parameter should be removed from ilo driver | 10:15 |
dtantsur | devananda, I meant discoverd configuration. +1 for removing from ilo. | 10:16 |
devananda | that parameter was never discussed in the spec, and AFAIK it only exists so the ilo driver can work around this byg | 10:16 |
rameshg87 | devananda: if we choose to enroll all ports, it's not required | 10:16 |
devananda | dtantsur: oh. gotcha. you'll make it optional in discoverd | 10:16 |
dtantsur | yep | 10:17 |
rameshg87 | devananda: so i will go ahead and remove all references of inspect_ports from ilo inspection | 10:18 |
devananda | sometimes the right long-term solution isn't the best short-term solution, but the short-term solution makes it much harder in the long term :( | 10:18 |
rameshg87 | devananda: may be worth discussing in the summit ? | 10:20 |
*** ramineni has quit IRC | 10:22 | |
*** ramineni has joined #openstack-ironic | 10:23 | |
openstackgerrit | Dmitry Tantsur proposed openstack/ironic: Document ports creating configuration for in-band inspection https://review.openstack.org/169253 | 10:23 |
dtantsur | devananda, kind of ^^^ wdyt? | 10:23 |
dtantsur | ++ for summit (provided I'll be granted a visa :D ) | 10:24 |
openstackgerrit | Dmitry Tantsur proposed openstack/ironic: Document ports creating configuration for in-band inspection https://review.openstack.org/169253 | 10:27 |
devananda | definitely needs discussion at summit | 10:29 |
devananda | would be good for some folks to test it ahead of time | 10:29 |
devananda | afaict, the current nova/network code handles multiple NICs, as long as they're all in the same subnet, just fine | 10:30 |
dtantsur | as to testing, our test labs have separate networks | 10:30 |
takadayuiko | dtantsur, is there? | 10:31 |
dtantsur | takadayuiko, o/ | 10:31 |
takadayuiko | dtantsur, hi :D | 10:31 |
devananda | OH! no, it doesn't | 10:31 |
dtantsur | :( | 10:31 |
dtantsur | takadayuiko, I'm a bit surprised you got DHCP request from some 192.168.100 network. did you change it? | 10:32 |
dtantsur | the default is different | 10:32 |
devananda | https://github.com/openstack/nova/blob/master/nova/network/neutronv2/api.py#L299 | 10:32 |
dtantsur | takadayuiko, anyway, could you paste both logs somewhere? | 10:33 |
devananda | cmpute/manager is geting a list of macs from ironic, passing that list to _create_port() there | 10:33 |
takadayuiko | dtantsur, No, I didn't do any change about neutron. Should I? | 10:33 |
devananda | but crete_port only sets up one | 10:33 |
takadayuiko | dtantsur, OK. screen-ironic-discoverd-dhcp.log: | 10:34 |
-takadayuiko- sudo dnsmasq --conf-file=/etc/ironic-discoverd/dnsmasq.conf | 10:34 | |
-takadayuiko- dnsmasq: started, version 2.68 DNS disabled | 10:34 | |
-takadayuiko- dnsmasq: compile time options: IPv6 GNU-getopt DBus i18n IDN DHCP DHCPv6 no-Lua TFTP conntrack ipset auth | 10:34 | |
-takadayuiko- dnsmasq-dhcp: DHCP, IP range 192.168.100.13 -- 192.168.100.20, lease time 1h | 10:34 | |
-takadayuiko- dnsmasq-dhcp: DHCP, sockets bound exclusively to interface br-discoverd | 10:34 | |
-takadayuiko- dnsmasq-dhcp: DHCPDISCOVER(br-discoverd) 52:54:00:58:b7:a5 | 10:34 | |
-takadayuiko- dnsmasq-dhcp: DHCPOFFER(br-discoverd) 192.168.100.18 52:54:00:58:b7:a5 | 10:34 | |
-takadayuiko- dnsmasq-dhcp: DHCPDISCOVER(br-discoverd) 52:54:00:58:b7:a5 | 10:34 | |
-takadayuiko- dnsmasq-dhcp: DHCPOFFER(br-discoverd) 192.168.100.18 52:54:00:58:b7:a5 | 10:34 | |
-takadayuiko- dnsmasq-dhcp: DHCPREQUEST(br-discoverd) 192.168.100.18 52:54:00:58:b7:a5 | 10:34 | |
-takadayuiko- dnsmasq-dhcp: DHCPACK(br-discoverd) 192.168.100.18 52:54:00:58:b7:a5 | 10:34 | |
dtantsur | takadayuiko, you'd better use http://paste.openstack.org/ for pasting :) | 10:34 |
takadayuiko | dtantsur, but my local.conf is a little bit different from your. | 10:34 |
dtantsur | takadayuiko, please provide it as well (also on http://paste.openstack.org/) | 10:35 |
takadayuiko | dtantsur, OK, sorry | 10:35 |
takadayuiko | dtantsur, http://paste.openstack.org/show/197674/ Can you see this? it's first time for me to use this | 10:36 |
dtantsur | takadayuiko, yeah, it's fine | 10:36 |
takadayuiko | dtantsur, and my local.conf: http://paste.openstack.org/show/197675/ | 10:37 |
dtantsur | takadayuiko, oh, IRONIC_DISCOVERD_DHCP_RANGE and IRONIC_DISCOVERD_INTERNAL_IP should be from some random subnet, not your real one. better leave them as default values (i.e. drop from local.conf) | 10:38 |
* devananda proposes https://review.openstack.org/169258 to Nova | 10:40 | |
takadayuiko | dtantsur, Oh! OK! I'll fix to be same as etherpad's. | 10:40 |
dtantsur | yes please :) | 10:40 |
* dtantsur hunts for food, brb | 10:41 | |
tiagogomes_ | Why is Ironic creating absolute symbolic links in the tftp folder? That makes impossible to use --secure option of tftp-hpa server no? | 10:46 |
openstackgerrit | Nisha Agarwal proposed stackforge/proliantutils: Adding functionality required by ilo-inspection https://review.openstack.org/164142 | 10:54 |
*** teju has joined #openstack-ironic | 10:55 | |
*** ramineni1 has joined #openstack-ironic | 11:02 | |
*** ramineni has quit IRC | 11:02 | |
*** ramineni1 has quit IRC | 11:03 | |
openstackgerrit | Vladyslav Drok proposed openstack/ironic: Removed hardcoded IDs from "chassis" test resources https://review.openstack.org/133205 | 11:05 |
*** jcoufal has quit IRC | 11:08 | |
*** jcoufal has joined #openstack-ironic | 11:08 | |
*** dmellado has quit IRC | 11:08 | |
*** jcoufal has quit IRC | 11:10 | |
takadayuiko | Hmm, it doesn't work well even though changing IRONIC_DISCOVERD_DHCP_RANGE and IRONIC_DISCOVERD_INTERNAL_IP... | 11:12 |
openstackgerrit | Yuriy Zveryanskyy proposed openstack/ironic: Do not save auth token on TFTP server in PXE driver https://review.openstack.org/159819 | 11:14 |
dtantsur | takadayuiko, please paste logs and local.conf again | 11:15 |
takadayuiko | dtantsur, thanks. Are you OK now? lunch time? local.conf is as this; http://paste.openstack.org/show/197684/ | 11:19 |
*** vdrok_afk is now known as vdrok | 11:19 | |
dtantsur | takadayuiko, already had lunch, thanks | 11:20 |
dtantsur | takadayuiko, local.conf looks fine, now logs please | 11:20 |
*** romcheg has quit IRC | 11:23 | |
takadayuiko | in this time, node doesn't boot maybe because of blacklisted. screen-ironic-discoverd.log: http://paste.openstack.org/show/197685/ | 11:23 |
takadayuiko | I deleted messages a little because it was too long | 11:23 |
*** romcheg has joined #openstack-ironic | 11:23 | |
tiagogomes_ | Can some one explain to me the purpose of the remapping rules on http://docs.openstack.org/developer/ironic/deploy/install-guide.html#configure-neutron-to-communicate-with-the-bare-metal-server | 11:29 |
*** jistr is now known as jistr|english | 11:32 | |
*** jistr|english is now known as jistr|class | 11:33 | |
dtantsur | takadayuiko, hmm, how did you start the introspection? | 11:33 |
takadayuiko | dtantsur, just executed API, like curl -g -i -X POST http://192.168.100.10:5050/v1/introspection/50d273af-1197-4336-95b5-666b7e3f8694 -H "X-Auth-Token: 79133057ea8c4a20b795cd6418e24309" | 11:33 |
dtantsur | takadayuiko, the thing is, I don't see this call in logs. did it succeed? | 11:34 |
dtantsur | takadayuiko, also updated https://etherpad.openstack.org/p/DiscoverdDevStack with instructions on how to avoid CURL :) | 11:36 |
takadayuiko | dtantsur, The reason is I removed. Well, API-executed-log was written in the log; INFO:werkzeug:192.168.100.10 - - [31/Mar/2015 12:30:59] "POST /v1/introspection/50d273af-1197-4336-95b5-666b7e3f8694 HTTP/1.1" 202 - | 11:36 |
takadayuiko | dtantsur, Wao! thanks! | 11:37 |
dtantsur | takadayuiko, could post the full log please? (also note that debug mode is not very helpful, it's safe to disabled it) | 11:38 |
takadayuiko | dtantsur, sure. I'll stack.sh again, so that it will take a little. | 11:38 |
dtantsur | ifarkas, hey, mind reviewing https://review.openstack.org/#/c/168354/ before you go on PTO? :) | 11:46 |
ifarkas | dtantsur, sure thing ;-) | 11:46 |
*** dmellado has joined #openstack-ironic | 11:47 | |
*** saripurigopi has quit IRC | 11:48 | |
*** ChanServ changes topic to "Bare Metal Provisioning | Status: http://bit.ly/ironic-whiteboard | Docs: http://docs.openstack.org/developer/ironic/ | Bugs: https://bugs.launchpad.net/ironic" | 11:49 | |
-openstackstatus- NOTICE: Check/Gate unstuck, feel free to recheck your abusively-failed changes. | 11:49 | |
*** Haomeng|2 has joined #openstack-ironic | 11:50 | |
*** Haomeng has quit IRC | 11:51 | |
*** romcheg1 has joined #openstack-ironic | 11:51 | |
*** romcheg has quit IRC | 11:52 | |
dtantsur | devananda, are we granting string freeze exception to https://review.openstack.org/#/c/165788/ ? | 11:54 |
dtantsur | ... and to https://review.openstack.org/#/c/168030/ | 11:54 |
*** romcheg has joined #openstack-ironic | 11:56 | |
takadayuiko | dtantsur, I've tried, and it fails with "502 Server Error: Bad Gateway" at "openstack baremetal introspection start <uuid>". | 11:57 |
openstackgerrit | jiangfei proposed openstack/python-ironicclient: Log the SHA1 hash of X-Auth-Token value https://review.openstack.org/162517 | 11:58 |
takadayuiko | it's same as directly executing "openstack baremetal introspection star" command. | 11:58 |
*** romcheg1 has quit IRC | 11:58 | |
dtantsur | O_o | 11:58 |
takadayuiko | it's weired. BTW, it's PM9::00 here. I have to go home and cook for my hasband :( | 11:59 |
dtantsur | takadayuiko, oh sorry for keeping you here :( | 11:59 |
dtantsur | I think I'll approve the devstack thing and then let's fix your case... | 12:00 |
dtantsur | I think you're not the last one for whom this plugin won't work :D | 12:00 |
dtantsur | g'night | 12:00 |
takadayuiko | dtantsur, no problem at all, It's sorry taking your time and thank you very much. | 12:00 |
dtantsur | :) | 12:01 |
takadayuiko | I'll investigate tomorrow ;) | 12:01 |
takadayuiko | Bye. | 12:01 |
*** takadayuiko has quit IRC | 12:01 | |
*** ndipanov has quit IRC | 12:03 | |
*** teju has left #openstack-ironic | 12:04 | |
*** lucasagomes is now known as lucas-hungry | 12:04 | |
*** mgoddard has joined #openstack-ironic | 12:06 | |
*** zhenguo has quit IRC | 12:07 | |
*** dprince has joined #openstack-ironic | 12:09 | |
*** pas-ha has quit IRC | 12:18 | |
openstackgerrit | Merged stackforge/ironic-discoverd: Proper errors handling for client https://review.openstack.org/168354 | 12:19 |
openstackgerrit | Merged stackforge/ironic-discoverd: Add DevStack plugin for ironic-discoverd https://review.openstack.org/164782 | 12:19 |
*** mgoddard has quit IRC | 12:23 | |
* rameshg87 goes home | 12:23 | |
*** rameshg87 has quit IRC | 12:23 | |
devananda | dtantsur: 169253 seems fine to me | 12:26 |
dtantsur | good | 12:27 |
openstackgerrit | jiangfei proposed openstack/python-ironicclient: Log the SHA1 hash of X-Auth-Token value https://review.openstack.org/162517 | 12:31 |
*** pas-ha has joined #openstack-ironic | 12:32 | |
*** trown|outttypeww is now known as trown | 12:42 | |
*** mgoddard has joined #openstack-ironic | 12:51 | |
*** lucas-hungry is now known as lucasagomes | 12:53 | |
*** mgoddard has quit IRC | 12:56 | |
*** jistr|class is now known as jistr | 12:58 | |
*** rameshg87 has joined #openstack-ironic | 13:01 | |
openstackgerrit | Merged openstack/ironic: DRAC: add retry capability to wsman client operations https://review.openstack.org/165860 | 13:04 |
*** Marga_ has joined #openstack-ironic | 13:11 | |
*** Marga_ has quit IRC | 13:11 | |
tiagogomes_ | Hi, the image deployed by Ironic is hanging at "Trying to unpack rootfs image as initramfs". Any ideas? I used disk-image-builder to create the images | 13:11 |
*** Marga_ has joined #openstack-ironic | 13:12 | |
*** logan2 has quit IRC | 13:16 | |
*** rloo has joined #openstack-ironic | 13:19 | |
*** logan2 has joined #openstack-ironic | 13:19 | |
*** kkoski has joined #openstack-ironic | 13:21 | |
*** kkoski has quit IRC | 13:22 | |
*** kkoski has joined #openstack-ironic | 13:22 | |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Remove inspect_ports from ilo inspection https://review.openstack.org/169303 | 13:29 |
rameshg87 | tiagogomes_: how much memory does your system have ? | 13:30 |
rameshg87 | tiagogomes_: and how bug is the ramdisk ? | 13:30 |
rameshg87 | tiagogomes_: is it a virtual machine ? | 13:30 |
tiagogomes_ | rameshg87 yes, it is a VM for the time being | 13:30 |
tiagogomes_ | ramdisk is 83M | 13:31 |
tiagogomes_ | the VM has allocated 256M of RAM | 13:31 |
Shrews | tiagogomes_: i think DiB created images require at least 3GB of RAM on the VM. I'm not quite sure why, but it stinks, I know :( | 13:34 |
rloo | morning ironic'ers | 13:35 |
Shrews | rloo: morning | 13:35 |
rloo | Does anyone understand the string freeze? I thought we weren't supposed to approve any new strings: https://review.openstack.org/#/c/165860/ | 13:36 |
Shrews | rloo: hrm, i thought the freezes were very short lived... like, during the week of the release | 13:36 |
rloo | Shrews: I thought it was til end of kilo | 13:37 |
tiagogomes_ | Shrews ta, I'll try that | 13:37 |
TheJulia | tiagogomes_: If your using IPA, I think you need at least 500MB-1GB to be able to expand the agent image | 13:37 |
Shrews | "The string freeze is maintained for every module affected by the freeze until the release branch is cut. When that happens, the string freeze is lifted from master." | 13:37 |
rloo | Shrews: yeah, so until kilo release | 13:38 |
Shrews | *sigh* | 13:38 |
rloo | Shrews: +2 | 13:38 |
TheJulia | tiagogomes_: from there, the file your writing out is downloaded into RAM, and the file is written out to disk. | 13:38 |
rloo | Shrews: so we're not supposed to approve new strings, right? | 13:38 |
devananda | dtantsur: i've updated mgoddard's nova fix - https://review.openstack.org/#/c/153230/ | 13:39 |
*** rameshg871 has joined #openstack-ironic | 13:39 | |
rloo | Shrews: unless we accidentally do it. Hmm... | 13:39 |
devananda | dtantsur: if you have hardware with >1 NIC that you have been testing with, I'd like to know what you find | 13:39 |
devananda | string freeze is "dont approve unnecessary or very large changes to the translated strings" | 13:40 |
devananda | it's not "dont approve any at all" | 13:40 |
*** ukalifon1 has quit IRC | 13:40 | |
rloo | devananda: so we can use our judgement? | 13:40 |
devananda | and yes, it is from k3 tag until rc1 | 13:40 |
dtantsur | interesting, will test. lucasagomes, also have a look at https://review.openstack.org/#/c/153230/ | 13:40 |
devananda | so that translation teams have time to, you know, translate it | 13:40 |
rloo | devananda: or do/should we ask you? | 13:40 |
tiagogomes_ | TheJulia, I am using the pxe_ipmi driver, does that uses IPA? | 13:41 |
lucasagomes | will do | 13:41 |
devananda | rloo: preferrably, ask me, but there is a part of my mind right now that's wondering whether there is any benefit to Ironic in following the string freeze this cycle | 13:41 |
devananda | https://www.transifex.com/projects/p/ironic/ | 13:41 |
rloo | devananda: that was in my mind in Juno too. I think not. | 13:41 |
*** romcheg1 has joined #openstack-ironic | 13:41 | |
devananda | are all quite low still | 13:41 |
*** romcheg has quit IRC | 13:41 | |
devananda | unless they hit the threshold (80% IIRC) the translations won't be included at all | 13:42 |
*** rameshg87 has quit IRC | 13:42 | |
Shrews | pfft | 13:42 |
rloo | devananda: right. And I'm doubtful that they will do much this time. And what's the odds that they will have translated something that we might change in the near future | 13:42 |
devananda | ah, 75% | 13:43 |
devananda | https://wiki.openstack.org/wiki/Translations/Infrastructure#Workflow | 13:43 |
devananda | " The script connects to Transifex and asks to download for each resource all new "sufficiently translated" files. We define "sufficiently" as at least 75 per cent of strings are translated." | 13:43 |
TheJulia | tiagogomes_: if memory serves, iscsi, so I think your max required is something like 1GB | 13:44 |
rloo | devananda: maybe we can translate the english (UK) and english (australia) ones. I mean, how hard could those be :-) | 13:45 |
devananda | rloo: ;) | 13:45 |
TheJulia | heh | 13:45 |
lucasagomes | dtantsur, O.O I thought we only needed to filter the macs returned on the macs_for_instance() method? | 13:46 |
TheJulia | Somebody should just write a script that translates the strings to latin | 13:46 |
rloo | devananda: so I think I spend more time wondering about whether we can approve a patch cuz of strings, than is warranted given our current xlated string situation. So I'm going to ahh, use my own judgement to decide ;) | 13:46 |
dtantsur | lucasagomes, not sure I understand the question... | 13:46 |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Remove inspect_ports from ilo inspection https://review.openstack.org/169303 | 13:46 |
rameshg871 | dtantsur: ^^ tested :) | 13:46 |
devananda | rloo: great :) | 13:46 |
dtantsur | rameshg871, cool! | 13:46 |
lucasagomes | dtantsur, the patch you told me to take a look | 13:46 |
*** rameshg871 is now known as rameshg87 | 13:47 | |
dtantsur | lucasagomes, I've no idea about it's contents :) devananda has updated it and I though you might be interested | 13:47 |
jroll | morning all :) | 13:47 |
dtantsur | lucasagomes, we can apply it to our nova and try | 13:47 |
lucasagomes | I thought that the nova part, would be just to filter out the ports in Ironic which are bootable and return their mac addresses to nova | 13:47 |
rameshg87 | morning jroll | 13:47 |
lucasagomes | but that patch is doing way more than it | 13:47 |
dtantsur | jroll, TheJulia, Shrews, morning | 13:48 |
lucasagomes | dtantsur, I see | 13:48 |
lucasagomes | yeah wroth trying | 13:48 |
rameshg87 | morning everyone | 13:48 |
lucasagomes | jroll, TheJulia rloo morning | 13:48 |
jroll | devananda: so um. do we want to commit to that node.extra['physical_network'] thing? that's a bit of a data model change that we might want to think on | 13:48 |
lucasagomes | devananda, https://review.openstack.org/#/c/153230/ perhaps I need to read more about the problem this patch is trying to fix | 13:48 |
lucasagomes | but would be enough to just filter out the macs that can be PXE booted in the macs_for_instance() method? | 13:49 |
rloo | oh yeah, morning devananda (wherever you are), jroll, lucasagomes, dtantsur, TheJulia, rameshg87 | 13:49 |
dtantsur | rloo, morning! | 13:49 |
rameshg87 | rloo: morning | 13:49 |
jroll | lucasagomes: see the bug there, tl;dr nova has one network and picks a NIC at random to attach it to... your idea might be ok for now but will break later when we try to support multiple networks | 13:49 |
devananda | re the multiple NIC issue,https://review.openstack.org/#/c/153230/ looks like it should address the issue where different NICs are on different networks | 13:49 |
devananda | however there are two issues still not directly solved by it | 13:50 |
devananda | 1) which NIC(s) are bootable via PXE isn't explicit, but by creating different networks and associating the port to the right one, you can model it | 13:50 |
jroll | devananda: side note, according to docstrings it looks like you're still changing the API there (but I don't think you actually are) | 13:51 |
devananda | it appears that you need to request multiple networks to get multiple ports via Nova | 13:51 |
jroll | yes. and there's a false assumption that number of networks == number of ports | 13:52 |
devananda | which makes sense in VM-land but no sense in physcal-server-land. however, I'm not sure how much we would need to change in Nova to fix this | 13:52 |
devananda | jroll: exactly | 13:52 |
devananda | jroll: so even if we expose >1 port here, if they're all on one network, nova will only process one port | 13:53 |
* rameshg87 wonders if nova will have problems with breaking macs_for_instance for out-of-tree drivers | 13:53 | |
jroll | a long-standing comment in a downstream hack: "# TODO(russell_h): this is based on the broken assumption that the number of Neutron ports will match the number of physical ports." | 13:53 |
devananda | jroll: fwiw, this is still requiring a change to ironic. one must pass some extra info on the port object so nova knows what network to put it on | 13:53 |
devananda | jroll: aaahahaha | 13:53 |
devananda | jroll: so what's the code that follows it? :) | 13:53 |
*** logan2 has quit IRC | 13:53 | |
jroll | devananda: right, which is why I think we need to punt on it and do it right? | 13:54 |
jroll | s/?// | 13:54 |
devananda | jroll: yes | 13:54 |
jroll | devananda: for port in ports: attach_tenant_net(port) | 13:54 |
jroll | by the grace of $deity, we happen to have two NICs on our servers and two networks that we attach | 13:55 |
devananda | jroll: and no bonding | 13:56 |
jroll | yes bonding | 13:56 |
devananda | oh | 13:56 |
firl | devananda: I went through and put my flat network on the same management network so that the TFTP clients would be able to reach the IP. Now the qdhcp namespace responds to the DHCP requests, however it tries to pull the boot files ( TFTP ) off from the qdhcp ip instead of the IP on the network. Any ideas? | 13:56 |
*** logan2 has joined #openstack-ironic | 13:56 | |
devananda | firl: your dhcp config is pointing to the wrong ip? | 13:57 |
firl | devananda: the dhcp is auto configured based upon the subnet create | 13:58 |
devananda | firl: using neutron? | 13:58 |
firl | am I missing a config option in ironic / neutron? | 13:58 |
firl | ya | 13:58 |
firl | neutron subnet-create dot8 10.0.8.0/24 --name dot8 --ip-version=4 --gateway=10.0.8.1 --allocation-pool start=10.0.8.100,end=10.0.8.200 --enable-dhcp | 13:59 |
devananda | firl: using which driver in ironic? | 14:00 |
firl | ipmi_pxe | 14:01 |
firl | should the qdhcp namespace be able to respond on the TFTP? | 14:01 |
devananda | firl: /etc/ironic/ironic.conf should have a [pxe] section with tftp_server=x.x.x.x | 14:01 |
firl | got it | 14:02 |
firl | tftp_server=$my_ip | 14:02 |
devananda | firl: that may or may not be related to your issue. pxe should try to infer that | 14:02 |
devananda | yea | 14:02 |
firl | cool I will try that | 14:02 |
devananda | that IP will get passed to Neutron for dhcp next-server response | 14:03 |
devananda | sounds like it isn't, so that's where i'd look | 14:03 |
*** absubram has joined #openstack-ironic | 14:04 | |
*** yog__ has quit IRC | 14:05 | |
*** ChuckC has quit IRC | 14:09 | |
*** Marga_ has quit IRC | 14:20 | |
*** enikanorov has quit IRC | 14:23 | |
firl | TFTP.....P:110.0.8.14 MASK: 255.255.255.0 DHCP IP: 10.0.8.101 TFTP.32: TFTP open timeout | 14:24 |
firl | looks like no dice still | 14:24 |
*** enikanorov has joined #openstack-ironic | 14:27 | |
*** anderbubble has joined #openstack-ironic | 14:28 | |
*** viktors has quit IRC | 14:30 | |
openstackgerrit | Merged openstack/ironic: Use node UUID in logs instead of node ID https://review.openstack.org/167623 | 14:30 |
*** absubram has quit IRC | 14:31 | |
*** zz_jgrimm is now known as jgrimm | 14:31 | |
*** BadCub_Hiding is now known as BadCub | 14:35 | |
BadCub | morning folks | 14:35 |
lucasagomes | devananda, jroll so lemme see if I understood that nova patch right. With that nova will create a network for each physical_network that is passed with the mac addresses? | 14:36 |
lucasagomes | I'm wondering in the case we have 2 ports, and they have the same physical_network | 14:36 |
devananda | lucasagomes: right. it does not handle that case | 14:37 |
lucasagomes | will nova create the 2 network for both ports? | 14:37 |
lucasagomes | BadCub, morning | 14:37 |
lucasagomes | devananda, right | 14:37 |
devananda | lucasagomes: it only addresses the case where i have >1 network, and one port on each network | 14:37 |
devananda | or the case where I have 1 network, and only one port on tht network, and iwant to ignore all the other ports | 14:37 |
TheJulia | BadCub: good morning | 14:37 |
devananda | g'rmonring BadCub | 14:38 |
* devananda tyep gdoo | 14:38 | |
lucasagomes | devananda, I see, in that case wouldn't another metadata be useful? | 14:38 |
devananda | gah | 14:38 |
lucasagomes | like primary=True | 14:38 |
lucasagomes | meaning that, a port which has that is the one pick | 14:38 |
devananda | lucasagomes: well, maybe. but there are a lot of other cases to consider too. and whether or not nova will allow us to model the case where >1 port is on the same phys network AND we want them all to work | 14:39 |
*** ChuckC has joined #openstack-ironic | 14:39 | |
jroll | if they don't allow that, I'll be really sad | 14:41 |
lucasagomes | devananda, right yeah fair enough | 14:41 |
lucasagomes | jroll, yeah I hope they do, I'd love to be able to specify precisely which port I want to use | 14:42 |
lucasagomes | among many ports | 14:42 |
jroll | lucasagomes: I mean like, take my ball and go home kind of sad | 14:42 |
lucasagomes | but that said, it also makes us to look into ironic itself. why we register many ports if not of all of them are going to be used | 14:43 |
lucasagomes | should we change something in our part too? | 14:43 |
* lucasagomes brainstorms | 14:43 | |
jroll | I mean, multiple networks is pretty dang valid for bare metal | 14:44 |
jroll | and you'll likely only be booting once | 14:44 |
lucasagomes | sure | 14:44 |
jroll | e.g. at rackspace our hypervisors have at least two different networks | 14:45 |
jroll | pxe booting one* | 14:45 |
dtantsur | in our case it's also different networks | 14:45 |
*** anderbubble has left #openstack-ironic | 14:45 | |
NobodyCam | good morning Ironic | 14:46 |
*** edcragg has joined #openstack-ironic | 14:46 | |
BadCub | ugh. morning TheJulia , devananda and lucasagomes | 14:46 |
devananda | morning | 14:46 |
lucasagomes | NobodyCam, hey ya morning | 14:46 |
NobodyCam | morning lucasagomes, dtantsur jroll and devananda | 14:47 |
jroll | morning NobodyCam :) | 14:47 |
dtantsur | NobodyCam, morning | 14:47 |
NobodyCam | :) | 14:48 |
devananda | lucasagomes: given our earlier conversation re: string freeze ... 167122 ? | 14:48 |
devananda | lucasagomes: just posing the question. i dont feel strongly either way on that one | 14:48 |
*** romcheg has joined #openstack-ironic | 14:48 | |
lucasagomes | devananda, right, since it wasn't fixing any bugs just enhancing the message I thought it felt into the string freeze | 14:49 |
lucasagomes | I don't have a strong feeling on that too | 14:49 |
lucasagomes | plus, knowing that we are way low % on the translations | 14:49 |
lucasagomes | and it won't even be included in the release | 14:49 |
lucasagomes | we could give it a go | 14:50 |
devananda | lucasagomes: meh. you're right. not actually fixing anything. | 14:50 |
devananda | *any particular bug | 14:50 |
lucasagomes | yeah | 14:50 |
lucasagomes | that's why I blocked it | 14:50 |
*** romcheg1 has quit IRC | 14:50 | |
lucasagomes | I mean it's a nice improvement on the log message to know the number of attempts | 14:51 |
lucasagomes | but not critical or anything | 14:51 |
*** hj-hp has joined #openstack-ironic | 14:51 | |
*** ndipanov has joined #openstack-ironic | 14:51 | |
devananda | yea, well, after using this a bit more i kinda think that we should move all of those sort of things to DEBUG level | 14:51 |
devananda | because hardware *is* flaky. ironic's job (one of htem) is to handle that | 14:52 |
*** mgoddard has joined #openstack-ironic | 14:52 | |
devananda | so logging a WARNING every time a bmc is a litte slow --- this just adds noise for the operators who are looking for service problems in the logs | 14:52 |
devananda | jroll: JayF: thoughts ^ ? | 14:52 |
jroll | devananda: do we expect most people to run without DEBUG level? | 14:53 |
devananda | or perhaps s/DEBUG/NOTICE/ | 14:53 |
jroll | devananda: in our experience, slow BMCs can *really* mess up an environment | 14:53 |
jlvillal | Good morning Ironic. | 14:54 |
NobodyCam | good morning jlvillal | 14:54 |
devananda | jroll: beyond the node on which that bmc exists? | 14:54 |
BadCub | mornin jlvillal | 14:54 |
devananda | jroll: ie, because it causes timeouts in the periodic tasks? | 14:54 |
lucasagomes | jlvillal, morning | 14:54 |
jroll | devananda: yeah, if the power loop lags things go badly, worker threads get tied up, long locks etc | 14:54 |
devananda | gotcha | 14:54 |
devananda | we should fix that | 14:54 |
devananda | power loop shouldn't (b)lock | 14:55 |
jroll | devananda: when we started standing up prod, it was unusable until we disabled a few slow nodes | 14:55 |
jroll | yeah | 14:55 |
* devananda adds to priorities | 14:55 | |
* jlvillal thinks he will working on setting up ZoZ (Znc on Znc) today | 14:56 | |
jroll | O.o | 14:58 |
lucasagomes | there's a way to see how many times a job in gate failed or passed? | 14:58 |
lucasagomes | jenkins just shows that? /me looks | 14:58 |
devananda | lucasagomes: the job, not the change? | 15:01 |
lucasagomes | yeah the job | 15:01 |
lucasagomes | I want to see if the if the pxe_ipa is stable | 15:01 |
lucasagomes | it seems it is, but I wanted to know % | 15:02 |
devananda | logstash, or look at adam's stats page | 15:02 |
lucasagomes | ah cool, will take a look! devananda thanks | 15:02 |
devananda | lucasagomes: http://no-carrier.net/~adam/openstack/ironic_gate_status.html | 15:02 |
lucasagomes | nice, the job doesn't seem to be there we should include it :) | 15:03 |
lucasagomes | the page is cool anyway | 15:03 |
devananda | http://no-carrier.net/~adam/openstack/ironic_gate.html | 15:03 |
devananda | oh - pxe_ipa | 15:04 |
devananda | right, yea, not there | 15:04 |
devananda | lucasagomes: have a look at graphite.openstack.org -- you can make what ever graph you want | 15:04 |
lucasagomes | oh will take a look | 15:06 |
lucasagomes | thanks | 15:06 |
devananda | lucasagomes: http://graphite.openstack.org/render/?width=1086&height=384&_salt=1399585848.568&target=sumSeries%28stats_counts.zuul.pipeline.check.job.check-tempest-dsvm-ironic-pxe_ipa-nv.FAILURE%29%29&target=sumSeries%28stats_counts.zuul.pipeline.check.job.check-tempest-dsvm-ironic-pxe_ipa-nv.SUCCESS%29&lineMode=connected&connectedLimit=&areaMode=all&colorList=red%2Cgreen | 15:07 |
devananda | that's a start -- it doesn't show a comparison for where that job failed but others passed | 15:07 |
devananda | which is what you really want to see | 15:08 |
lucasagomes | devananda, heh i was just trying to get used to that interface, now I can see the folders on the left side with the data you can grab | 15:08 |
lucasagomes | devananda, thanks for that :D | 15:08 |
devananda | yvm! | 15:08 |
devananda | yvw, even | 15:08 |
devananda | also, i just edited the URL to change the test name ;) | 15:09 |
devananda | going to drop off for dinner in about 20 minutes | 15:09 |
devananda | taking another qiuck look at our RC1 bug list ... it's long ... perhaps I've targeted too much to it? | 15:10 |
lucasagomes | devananda, enjoy dinner | 15:12 |
lucasagomes | http://graphite.openstack.org/render/?width=1411&height=794&_salt=1427814736.728&target=stats_counts.zuul.pipeline.check.job.check-tempest-dsvm-ironic-pxe_ipa-nv.FAILURE&target=stats_counts.zuul.pipeline.check.job.check-tempest-dsvm-ironic-pxe_ipa-nv.SUCCESS&from=00%3A00_20150310&until=23%3A59_20150331 | 15:12 |
lucasagomes | looks pretty stable | 15:12 |
dtantsur | devananda, we can start dropping things from the list later :) | 15:12 |
dtantsur | also enjoy your meals | 15:13 |
* lucasagomes liks graphite | 15:13 | |
* dtantsur drops off for food shopping for some time | 15:13 | |
devananda | huh. I thought this was already addressed? https://bugs.launchpad.net/ironic/+bug/1412561 | 15:16 |
openstack | Launchpad bug 1412561 in Ironic "IPA/DIB is not compatible to create the uefi capable images" [Medium,Confirmed] | 15:16 |
devananda | or should that be targeted as a bug to DIB, not Ironic? | 15:16 |
*** a1exhughe5 has joined #openstack-ironic | 15:18 | |
*** Marga_ has joined #openstack-ironic | 15:19 | |
mjturek1 | hey all, I'm trying to use the python apis. Can anyone point me to the methods in the python APIs to get the extra, driver_info, and properties of an ironic node? | 15:19 |
jroll | JoshNang: left a comment here https://review.openstack.org/#/c/168197/ | 15:21 |
lucasagomes | devananda, yeah indeed it looks like it's related to the image building process | 15:22 |
firl | devananda: http://pastebin.com/z4ercBK4 I seem to be stuck. Does this seem to be an issue with ovs / neutron? | 15:22 |
jroll | mjturek1: it should be in the node show response /me looks | 15:22 |
lucasagomes | so ironic + dib seems correct | 15:22 |
lucasagomes | ironic because ipa uses the ironic bug tracker | 15:22 |
lucasagomes | devananda, https://bugs.launchpad.net/nova/+bug/1177655 I think this, would need to be L | 15:22 |
openstack | Launchpad bug 1177655 in Ironic "kernel boot command line for baremetal assumes block device UUID is correct root" [Medium,Triaged] | 15:22 |
jroll | mjturek1: sorry, node.get | 15:22 |
lucasagomes | because that also needs to change in the way the images are builded, so that we can build the root partition with a specific label | 15:23 |
*** zigo__ is now known as zigo | 15:23 | |
mjturek1 | jroll oh sweet, perfect! thanks :) | 15:24 |
jroll | np :) | 15:24 |
jroll | JoshNang: also curious why we need to bump this so high (this may have been asked before) https://review.openstack.org/#/c/168175/4/lib/tempest | 15:25 |
JoshNang | jroll: your comment is correct, i meant to remove that code and add it as a separate patch with some extra handling. | 15:27 |
jroll | oh! | 15:27 |
jroll | yay, I found a thing | 15:27 |
JoshNang | i think i left a comment on that review...i put it to 10 mins because it was timing out before the agent even got booted | 15:27 |
JoshNang | i'll check how long it actually took and bump it down | 15:27 |
jroll | ah ok | 15:28 |
jroll | right, we have to boot the agent again | 15:28 |
jroll | no worries then :) | 15:28 |
lucasagomes | someone understand this bug here? https://bugs.launchpad.net/ironic/+bug/1327260 | 15:28 |
openstack | Launchpad bug 1327260 in Ironic "Disk layout is not being verified on re-provisioning" [Medium,Triaged] | 15:28 |
lucasagomes | I mean, we destroy the disk layour when provisioning the node, so... | 15:28 |
jroll | JoshNang: "I bumped the timeout way up just to test. 10 minutes is definitely unreasonable." they probably won't merge with that comment there :P | 15:28 |
devananda | lucasagomes: "rebuild" ? | 15:28 |
jroll | lucasagomes: sounds like rebuild | 15:28 |
jroll | that's an interesting bug | 15:29 |
JoshNang | jroll: heh and it's not my patch, so i couldn't workflow :P | 15:29 |
devananda | yea, if someone changes the data in ironic then starts a rebuild, it'll break | 15:29 |
lucasagomes | oh | 15:29 |
lucasagomes | gotcha | 15:29 |
jroll | JoshNang: oh, oops. tests should catch it if it's too short right? | 15:29 |
*** spandhe has joined #openstack-ironic | 15:30 | |
lucasagomes | like root_gb ephemeral_gb and so on | 15:30 |
lucasagomes | yeah | 15:30 |
lucasagomes | makes sense | 15:30 |
JoshNang | jroll: for sure | 15:30 |
devananda | like if I change root_gb and swap_mb | 15:30 |
jroll | devananda: I think it's the reverse here, someone changes the disk layout on the disk? | 15:30 |
devananda | jroll: oh. that too | 15:30 |
jroll | maybe not | 15:30 |
*** maurosr has quit IRC | 15:30 | |
jroll | at any rate | 15:30 |
lucasagomes | that can get really funny... heh | 15:30 |
devananda | jroll: if I parted and shrink root, for example, then try to rebuild, yea, I think that would break | 15:30 |
jroll | you can't preserve ephemeral in that case | 15:30 |
*** ramineni has joined #openstack-ironic | 15:30 | |
devananda | well, right :) | 15:30 |
lucasagomes | not sure whether we should allow changing such things when the machine is deployed | 15:31 |
jroll | which is what the bug is about | 15:31 |
devananda | the whole preserve ephemeral is somewhat .... awkward | 15:31 |
jroll | and an oxymoron | 15:31 |
jroll | I hate it | 15:31 |
devananda | indeed | 15:31 |
lucasagomes | lol | 15:31 |
lucasagomes | yeah | 15:31 |
devananda | i think we all do | 15:31 |
devananda | *have some issues with it | 15:31 |
lucasagomes | is tripleo still using it? | 15:31 |
jroll | I'm sure they are, but I really have no idea | 15:31 |
jroll | looks like fuel is using it based on this bug ;D | 15:32 |
devananda | in any case, I need to run (car's waiting). I targeted some of these bugs so we'd look at them as they're old and sound bad | 15:33 |
lucasagomes | yeah, imo the only way to safely fix that bug is to prevent those sizes from being updated if the node is already deployed | 15:33 |
lucasagomes | it's not only recreating the partition layout if, say the first partition is now bigger | 15:33 |
*** spandhe_ has joined #openstack-ironic | 15:33 | |
lucasagomes | we would have to move data around | 15:33 |
lucasagomes | which I'm pretty sure we don't want to do | 15:33 |
jroll | +1 | 15:33 |
devananda | maybe they're not valid any more, or maybe the right thing to do is just more error checking :) | 15:34 |
*** spandhe has quit IRC | 15:34 | |
*** spandhe_ is now known as spandhe | 15:34 | |
devananda | lucasagomes: blocking changes to node.properties while it is state=active makes some sense. i think we tried that a couple cycles ago though and had a problem? | 15:34 |
* devananda really goes afk now | 15:34 | |
*** maurosr has joined #openstack-ironic | 15:35 | |
lucasagomes | devananda, hmm can't recall. But yeah I think we should prevent only a subset of it | 15:36 |
*** mgoddard has quit IRC | 15:36 | |
lucasagomes | static things like disk layout, things like memory_gb still can be updated | 15:36 |
* lucasagomes have to think more about it | 15:37 | |
firl | devananda: the TFTP server isn’t listening on the qdhcp ip namespace. Should it be? right now my PXE client gets an IP from the qdhcp namespace but can’t connect to the tftp server because it is passing the same IP. and the TFTP IP config change didn’t seem to matter | 15:44 |
*** ifarkas has quit IRC | 15:44 | |
*** ramineni has quit IRC | 15:45 | |
*** dprince has quit IRC | 15:46 | |
*** saripurigopi has joined #openstack-ironic | 15:47 | |
* BadCub has to reboot for lovely updates. BRB | 15:50 | |
*** a1exhughe5 has quit IRC | 16:00 | |
*** rloo has quit IRC | 16:02 | |
openstackgerrit | Josh Gachnang proposed openstack/ironic: Fix cleaning nits https://review.openstack.org/168197 | 16:05 |
*** jistr has quit IRC | 16:09 | |
jroll | JoshNang: you missed the other nits there :P | 16:10 |
JoshNang | jroll: oh gah, got distracted by the weirdness. | 16:10 |
*** Sukhdev has joined #openstack-ironic | 16:11 | |
jroll | also uh | 16:11 |
jroll | so do we not handle reboots for in-band things at all right now? | 16:11 |
JoshNang | yeahhh | 16:11 |
*** saripurigopi_ has joined #openstack-ironic | 16:12 | |
JoshNang | i was working on a patch for it yesterday but got distracted by the decom->cleaning thing | 16:12 |
jroll | ok | 16:13 |
*** saripurigopi has quit IRC | 16:14 | |
jroll | thanks, +2 | 16:14 |
*** saripurigopi_ is now known as saripurigopi | 16:14 | |
*** dprince has joined #openstack-ironic | 16:14 | |
openstackgerrit | Josh Gachnang proposed openstack/ironic: Fix cleaning nits https://review.openstack.org/168197 | 16:15 |
JoshNang | jroll: oh blah, i was just waiting on tests after fixing the nits from NobodyCam | 16:16 |
jroll | oh, no worries :) | 16:16 |
JayF | Meeting today in 40m, correct? | 16:19 |
jroll | no? | 16:19 |
jroll | it was last night | 16:19 |
jroll | it's tuesday :) | 16:19 |
*** derekh has quit IRC | 16:21 | |
JayF | HAH | 16:21 |
JayF | It's monday in my head, doesn't that count for anything | 16:22 |
jroll | no, because this week was the late meeting ;D | 16:22 |
*** mgoddard has joined #openstack-ironic | 16:26 | |
*** mtanino has joined #openstack-ironic | 16:27 | |
NobodyCam | so many calls this morning :-p | 16:27 |
* BadCub no like late meeting *grumbles at coffee cup* | 16:30 | |
*** spandhe has quit IRC | 16:35 | |
*** harlowja_away is now known as harlowja_ | 16:45 | |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Check status of bootloader installation for DIB ramdisk https://review.openstack.org/169419 | 16:46 |
*** absubram has joined #openstack-ironic | 16:47 | |
rameshg87 | folks ^^^^^ | 16:47 |
rameshg87 | above is a proposed fix for our last critical bug for rc-1 | 16:47 |
rameshg87 | https://bugs.launchpad.net/ironic/+bug/1422723 | 16:47 |
openstack | Launchpad bug 1422723 in Ironic "Ironic sets instance as active even if bootloader installation fails when boot_option is 'local'" [Critical,In progress] - Assigned to Ramakrishnan G (rameshg87) | 16:47 |
rameshg87 | (last for now) | 16:48 |
rameshg87 | haven't fixed the tests/flake8 | 16:48 |
rameshg87 | but wanted to know if jenkins goes through and would love to get some quick feedback from someone | 16:48 |
rameshg87 | i will fix the tests and rest of the things tomorrow | 16:48 |
rameshg87 | thanks | 16:48 |
rameshg87 | and good night all | 16:49 |
rameshg87 | see you tomorrow | 16:49 |
rameshg87 | bye | 16:49 |
*** rameshg87 has quit IRC | 16:49 | |
*** Marga_ has quit IRC | 16:49 | |
jroll | devananda: hahahaha, I found my thing with the conductors. clock skew. | 16:57 |
*** rloo has joined #openstack-ironic | 16:57 | |
*** saripurigopi has quit IRC | 16:57 | |
BadCub | hey folks... Going through our old bugs... Is anyone still seeing this issue: https://bugs.launchpad.net/ironic/+bug/1319858 | 16:59 |
openstack | Launchpad bug 1319858 in Ironic "default api loop count / intervals can't cope with 40 machine clusters" [High,Confirmed] | 16:59 |
jroll | BadCub: the default is better, but doing callbacks could make it even better https://github.com/openstack/nova/blob/master/nova/virt/ironic/driver.py#L84 | 17:00 |
jroll | which is why we left that open | 17:01 |
openstackgerrit | Jay Faulkner proposed openstack/ironic: Cleaning docs followup: Add that it's new in Kilo https://review.openstack.org/169423 | 17:02 |
*** saripurigopi has joined #openstack-ironic | 17:03 | |
JayF | https://review.openstack.org/#/c/168189/ is ready to land, and ^ 169423 is the followup for one of the nits | 17:03 |
JayF | the first patch only needs one more +2A, lots of +1s | 17:03 |
NobodyCam | jroll: think that old bug still need to be left at High? | 17:05 |
*** andreykurilin_ has joined #openstack-ironic | 17:05 | |
JayF | jroll: ^ you could land 168189 fwiw :D | 17:05 |
* BadCub adds followup cleaning doc to pad | 17:05 | |
jroll | NobodyCam: dunno, it's going to end up as a spec | 17:06 |
NobodyCam | ya | 17:07 |
BadCub | jroll: NobodyCam I changed it to Wishlist and noted that it will end up requiring a spec | 17:08 |
jroll | cool, thanks | 17:09 |
BadCub | yw :) | 17:09 |
*** kalpase has left #openstack-ironic | 17:11 | |
NobodyCam | JayF: look at the new doc (which is awesome TY) | 17:11 |
NobodyCam | paragraph 5 under Troubleshooting | 17:12 |
NobodyCam | teh last sentence : It will restart cleaning from the first step | 17:12 |
NobodyCam | oh NM nodes got thru cleading moving from MANAGEABLE to AVAILABLE | 17:13 |
openstackgerrit | Julia Kreger proposed openstack/ironic: Add retry logic to _exec_ipmitool https://review.openstack.org/168120 | 17:13 |
NobodyCam | w00 h00 TheJulia :) | 17:14 |
*** dtantsur is now known as dtantsur|afk | 17:14 | |
TheJulia | That... was not fun :) | 17:14 |
JayF | NobodyCam: ah, I'll add that fix to my followup | 17:14 |
openstackgerrit | Shivanand Tendulker proposed openstack/ironic: boot_mode is overwritten in node properties https://review.openstack.org/167952 | 17:14 |
NobodyCam | JayF: ack I'll +a then :) | 17:14 |
openstackgerrit | Jay Faulkner proposed openstack/ironic: Cleaning docs followup: Add that it's new in Kilo https://review.openstack.org/169423 | 17:15 |
openstackgerrit | Shivanand Tendulker proposed openstack/ironic: pxe deploy fails for whole disk images in UEFI https://review.openstack.org/167882 | 17:15 |
openstackgerrit | Jay Faulkner proposed openstack/ironic: Followup to comments on Cleaning Docs https://review.openstack.org/169423 | 17:15 |
openstackgerrit | Shivanand Tendulker proposed openstack/ironic: grub2 bootloader support for uefi boot mode https://review.openstack.org/166192 | 17:16 |
openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for pxe_ilo driver https://review.openstack.org/154808 | 17:16 |
NobodyCam | bam | 17:17 |
lucasagomes | hey folks I will call it a day | 17:19 |
lucasagomes | have a good night everyone | 17:19 |
*** lucasagomes is now known as lucas-dinner | 17:19 | |
NobodyCam | have a great night lucas-dinner | 17:19 |
BadCub | g'night lucas-dinner | 17:20 |
* NobodyCam gets more offee | 17:20 | |
JayF | night lucas | 17:20 |
*** achanda has joined #openstack-ironic | 17:23 | |
*** Marga_ has joined #openstack-ironic | 17:24 | |
*** pas-ha has quit IRC | 17:27 | |
*** Sukhdev has quit IRC | 17:35 | |
NobodyCam | didn't we fix this already: https://bugs.launchpad.net/ironic/+bug/1403887 ? | 17:38 |
openstack | Launchpad bug 1403887 in Ironic "No such file or directory error during deployment" [High,Triaged] | 17:38 |
*** jerryz_ has quit IRC | 17:39 | |
JayF | NobodyCam: that was fixed in /ipa/ | 17:39 |
JayF | NobodyCam: was it fixed in Ironic-proper? | 17:39 |
JayF | Thinking fix for this is `udevadm settle`; right? | 17:39 |
NobodyCam | with the udevadm settle? | 17:39 |
JayF | the udevadm settle 100% landed in IPA | 17:39 |
TheJulia | NobodyCam: could you be thinking of a similar issue downstream ? | 17:39 |
JayF | idk about Ironic-pxe | 17:39 |
NobodyCam | ya.. | 17:40 |
NobodyCam | TheJulia: I could be | 17:40 |
TheJulia | NobodyCam: I remember talking about it, I'm thinking it was something like the iscsi connection was blipping in the rack due to the switchports, I remember we requested information w/r/t the switch configuration and I think thats where it was left off. | 17:45 |
NobodyCam | ya I think I'm thinking of https://bugs.launchpad.net/ironic/+bug/1415117 | 17:46 |
openstack | Launchpad bug 1415117 in Ironic "iSCSI login depends on fix length sleep to wait for connections to become active" [Medium,In progress] - Assigned to Chris Krelle (nobodycam) | 17:46 |
*** pelix has quit IRC | 17:50 | |
NobodyCam | I'll ask lucas-dinner tomorrow if this is still a issue | 17:51 |
* NobodyCam heads out for quick smoke | 17:53 | |
*** penick has joined #openstack-ironic | 17:54 | |
*** mtanino has quit IRC | 17:56 | |
*** spandhe has joined #openstack-ironic | 17:59 | |
*** saripurigopi has quit IRC | 18:01 | |
*** jamielennox|away is now known as jamielennox | 18:04 | |
*** edcragg has quit IRC | 18:09 | |
JayF | Does anyone know what the Friday of the summit is supposed to be? Nothing at all on the sched for it -- basically wondering if I should be good to fly out Friday afternoon rather than staying overnight to get all day fri in | 18:09 |
JayF | In case you haven't seen it; summit schedule: https://openstacksummitmay2015vancouver.sched.org/ | 18:09 |
jroll | JayF: design sessions? | 18:10 |
jroll | I don't see design sessions on the schedule | 18:10 |
JayF | hmm | 18:11 |
*** Haomeng has joined #openstack-ironic | 18:11 | |
JayF | Design summit is listed as Tues-Fri | 18:11 |
JayF | https://www.openstack.org/summit/vancouver-2015/details-and-updates/ | 18:11 |
rloo | I thought Fri was going to be sort of like a mini mid-cycle thingy. small pods/groups. up to us/ironic to decide what to do then i think. | 18:12 |
JayF | I guess I'd need to konw if Ironic sessions were on Fri to make a decision? | 18:12 |
JayF | Aha. Okay, like it was in Paris, then? | 18:12 |
*** Haomeng|2 has quit IRC | 18:12 | |
jroll | hrm, this looks interesting: https://openstacksummitmay2015vancouver.sched.org/event/7461140343cb709e3113d353c984842e | 18:12 |
JayF | I'm just trying to avoid flying on Sun-Sat | 18:12 |
rloo | yeah, although 'more so'. i think. fewer large design sessions, more smaller things. | 18:12 |
jroll | JayF: I think it's safe to say friday will be valuable | 18:12 |
jroll | I'm probably flying out friday evening | 18:12 |
rloo | by fri aft, i suspect people will be tired/ready to leave. | 18:12 |
JayF | jroll: there aren't really flights | 18:12 |
JayF | jroll: which is why I was looking at leaving on the 5-6am flight Sat if I stayed on Fri | 18:13 |
jroll | JayF: hrm, I'm going to pdx, hopefully there is one | 18:13 |
JayF | Maybe so, that'll be shorter for sure | 18:13 |
rloo | where is pdx? | 18:13 |
jroll | portland | 18:13 |
rloo | jroll: oh. thx. | 18:14 |
jroll | going to visit friends for the long weekend | 18:14 |
jroll | :) | 18:14 |
JayF | Nothing YVR->SFO in our corporate travel booking thing that leaves after 3pm on Friday | 18:14 |
rloo | oh, is that a long weekend in the US? | 18:14 |
JayF | so I'll just book Sun-Sat | 18:14 |
JayF | oh, is it? I don't think it is... | 18:14 |
BadCub | as of yesterday the 4Seasons still had some rooms at the conference rate for Vancouver, but you have to call them direct | 18:14 |
JayF | First Monday in May is memorial day | 18:14 |
jroll | oh, right. cwt. | 18:14 |
* jroll cries | 18:14 | |
JayF | yeeep | 18:15 |
jroll | rloo: yeah, memorial day | 18:15 |
jroll | last monday in may | 18:15 |
JayF | wait, so we have that Monday off? Seriously, that's memorial day? | 18:15 |
JayF | hrm. | 18:15 |
jroll | ya | 18:15 |
rloo | jroll: my long weekend ends the day the summit starts :-( | 18:15 |
* JayF wonders if $wife would want to join him for a pleasant weekend in vancouver | 18:15 | |
*** dencaval has joined #openstack-ironic | 18:15 | |
jroll | JayF: doeet, vancouver is great from what I hear | 18:15 |
jroll | rloo: :( | 18:15 |
BadCub | Isn't conference the 3rd week of May? | 18:16 |
JayF | no | 18:16 |
jroll | https://www.google.com/search?q=memorial+day+2015&ie=utf-8&oe=utf-8 | 18:16 |
JayF | well, yes | 18:16 |
jroll | it is | 18:16 |
JayF | Mon May 18- Fri MAy-22 | 18:16 |
jroll | memorial day is the monday after | 18:16 |
JayF | and then Mon May 24 is US holiday | 18:17 |
jroll | 25* | 18:17 |
jroll | :P | 18:17 |
BadCub | oh hell. That really messes things up | 18:17 |
JayF | I math real good | 18:17 |
jroll | JayF: I just assume you still think today is yesterday | 18:17 |
JayF | Heh | 18:18 |
* BadCub has to rethink packing up the RV to head to Washington for summer because of conference *frowns* | 18:19 | |
JayF | Why not take the RV? | 18:20 |
JayF | It's in Vancouver | 18:20 |
JayF | if you're already in Washington... right there | 18:20 |
BadCub | That may end up being the plan | 18:20 |
BadCub | we are booked in Sequim, which is 20 minutes south of the ferry to Vancouver | 18:20 |
BadCub | but that means the kids will be with us, so I won't be able to go to conference :( | 18:22 |
openstackgerrit | Merged openstack/ironic: Ensure configdrive isn't mounted for ilo drivers https://review.openstack.org/167700 | 18:26 |
*** mgoddard has quit IRC | 18:27 | |
openstackgerrit | Merged openstack/ironic: Add IPA to enable drivers doc's page https://review.openstack.org/168095 | 18:27 |
* NobodyCam needs to run to the cornor store .. brb | 18:31 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/ironic-lib: Updated from global requirements https://review.openstack.org/169453 | 18:32 |
BadCub | hey adam_g are our current tempest tests addressing this bug: https://bugs.launchpad.net/ironic/+bug/1252824 if so, I will close it out | 18:34 |
openstack | Launchpad bug 1252824 in Ironic "RPC exception deserialization is not tested (need integration tests)" [Low,Triaged] | 18:34 |
adam_g | BadCub, hmm no. im not sure thats something we can test, at least according to the test example described there | 18:36 |
BadCub | hmm probably why it has been hanging around for so long :-p | 18:39 |
* NobodyCam is back-ish | 18:43 | |
openstackgerrit | Nisha Agarwal proposed stackforge/proliantutils: Adding functionality required by ilo-inspection https://review.openstack.org/164142 | 18:46 |
*** achanda has quit IRC | 18:49 | |
*** Marga_ has quit IRC | 18:51 | |
NobodyCam | rloo: happen to be around? | 18:53 |
rloo | NobodyCam: yeah, in a meeting. what's up? | 18:54 |
NobodyCam | oh just had a question on your comment on 164369 | 18:54 |
NobodyCam | it can wait until you done, just wanted to know if the review you pointed out just needed to be added to the comments on that patch or if more code was needed to guard them | 18:55 |
* BadCub is going blind sifting through year-old bugs | 18:56 | |
rloo | NobodyCam: there are those two endpoints that will be available in v1.0 | 18:56 |
NobodyCam | BadCub: :( | 18:56 |
NobodyCam | rloo: ahh ok I'll add them :) | 18:56 |
NobodyCam | thank you :) | 18:56 |
rloo | NobodyCam: the GET .../vendor_passthru/methods etc | 18:56 |
NobodyCam | and good morning rloo :) | 18:56 |
rloo | NobodyCam: good morning! | 18:57 |
NobodyCam | ya, I'll add them | 18:57 |
NobodyCam | :) | 18:57 |
rloo | you need to add code somewhere to check the version. | 18:57 |
rloo | NobodyCam: in v1/node.py and v1/driver.py I think. | 18:59 |
*** Sukhdev has joined #openstack-ironic | 19:01 | |
*** Marga_ has joined #openstack-ironic | 19:02 | |
NobodyCam | :) was just looking there | 19:04 |
*** Sukhdev has quit IRC | 19:05 | |
firl | anyone know if I should be able to see my SoL console inside the openstack webview of the console / log? | 19:11 |
*** spandhe has quit IRC | 19:13 | |
NobodyCam | jlvillal: have you looked at the faliures on 169109? not sure they are real but the ibm 3rd party tet have been mostly stable? | 19:16 |
NobodyCam | firl: I have not tested that | 19:17 |
firl | do you know what the reasons we have the IPMI console parameters then? | 19:17 |
firl | just future dev? or is there another way to access it | 19:17 |
jlvillal | NobodyCam: I have glanced at them. I am going to try a recheck. | 19:18 |
jroll | NobodyCam: I've never noticed the xcat stuff being stable | 19:20 |
*** athomas has quit IRC | 19:21 | |
NobodyCam | jroll: humm | 19:22 |
NobodyCam | jlvillal: Thank you | 19:22 |
NobodyCam | firl: we do have console support | 19:22 |
jlvillal | NobodyCam: Thanks for pointing it out :) | 19:22 |
NobodyCam | I was just pointing out that I have not tested it in horzion | 19:22 |
*** hj-hp has quit IRC | 19:24 | |
firl | ahh gotcha | 19:24 |
*** hj-hp has joined #openstack-ironic | 19:24 | |
firl | Is there a doc / guide to working with the console instead of me just SoL from command line? the ironic-conductor keeps stealing the session | 19:25 |
firl | just don’t know where to view the conductors session | 19:25 |
jroll | disable console for the node and the conductor shouldn't grab a session from you? | 19:26 |
*** achanda has joined #openstack-ironic | 19:27 | |
*** Marga_ has quit IRC | 19:27 | |
firl | jroll: yeah, I can but if there is an “ironic” way to use it | 19:28 |
firl | I was trying to leverage that | 19:28 |
NobodyCam | firl: that is a area our doc's could use improvment | 19:28 |
*** spandhe has joined #openstack-ironic | 19:28 | |
jroll | firl: I see, I've never used ironic's console thing but I thought it gives you some sort of url for the console? | 19:29 |
firl | jroll: NobodyCam said it hasn’t been tested on horizon. Guess It just might not be visible yet? I don’t mind looking into it, I just don’t know where to start hah | 19:29 |
*** Sukhdev has joined #openstack-ironic | 19:29 | |
jroll | firl: I have no idea about horizon, sorry :) | 19:30 |
firl | haha | 19:30 |
firl | if there is a non horizon url that’d be awesome | 19:30 |
jroll | see if ironic node-show $node has some sort of url, maybe in driver_info | 19:30 |
firl | it’s deploying now, will see if it shows up | 19:32 |
firl | no url’s, just instance_info | 19:32 |
jroll | hrm, yeah, dunno. I always just use ipmitool when I need a console :) | 19:33 |
* NobodyCam steps away to find his mind and maybe some food stuffs too | 19:35 | |
*** spandhe has quit IRC | 19:35 | |
*** Sukhdev has quit IRC | 19:36 | |
*** Sukhdev has joined #openstack-ironic | 19:37 | |
firl | yeah, I am getting a post PXE boot kernel load error. I figure now I just need to make sure I provisioned the images properly | 19:41 |
*** alexpilotti has joined #openstack-ironic | 19:41 | |
*** hj-hp has quit IRC | 19:50 | |
*** wanyen has joined #openstack-ironic | 19:52 | |
openstackgerrit | Merged openstack/ironic: Add documentation for Cleaning https://review.openstack.org/168189 | 19:53 |
*** Sukhdev has quit IRC | 19:54 | |
wanyen | Hi, JoshNang, I have a question regarding node zapping and cleaning | 19:55 |
*** Marga_ has joined #openstack-ironic | 19:58 | |
BadCub | heya jroll is https://bugs.launchpad.net/ironic/+bug/1300939 still an issue we are looking at? | 19:59 |
openstack | Launchpad bug 1300939 in Ironic "API does not translate errors based on Accept-Language header" [Low,Confirmed] - Assigned to Jim Rollenhagen (jim-rollenhagen) | 19:59 |
*** hj-hp has joined #openstack-ironic | 20:01 | |
JoshNang | wanyen: hi! ok, i hopefully have an answer :) | 20:02 |
*** Marga_ has quit IRC | 20:03 | |
*** andreykurilin_ has quit IRC | 20:06 | |
jroll | BadCub: lol, I picked that up at the end of kilo and never finished it, I know Haomeng was working on the same thing but maybe fined a different bug? | 20:07 |
BadCub | hmmm | 20:08 |
jroll | s/fined/filed | 20:08 |
*** dencaval has quit IRC | 20:09 | |
BadCub | Haomeng: are you working on something similar to/or dup of ^^^ | 20:09 |
BadCub | thanks jroll :-) I will wait to hear from Haomeng on that one | 20:10 |
jroll | BadCub: https://review.openstack.org/#/c/167507/ | 20:12 |
jroll | it was that one | 20:12 |
wanyen | JoshNang, node zapping design spec says "The API will be changed to prevent changing power state or provision state while the node is in a ZAPPING state". Can a driver power on bare-metal node while performing execute_cleaning_steps()? | 20:12 |
BadCub | Coolness | 20:12 |
BadCub | I will mark the original as a dup then | 20:13 |
JoshNang | wanyen: heh we were just talking about this yesterday. i'm going to put up a patch to prevent it. | 20:15 |
jroll | BadCub: usually the latter is the dupe | 20:15 |
jroll | BadCub: also that change was abandoned | 20:15 |
BadCub | I see that | 20:15 |
BadCub | leaving it as-is for now | 20:15 |
jroll | k | 20:15 |
wanyen | JoshNang, OOB node cleaning may need to power on the bare-metal node during execute_cleaning-steps so I am actually hoping that we won't restrict driver to power on a bm node. | 20:17 |
JoshNang | wanyen: it should only be restricted at the api level | 20:18 |
wanyen | JoshNang, so the driver can power on/off the node during execute_cleaning-steps as needed. right? | 20:19 |
*** achanda has quit IRC | 20:20 | |
JoshNang | wanyen: i think that's reasonable. the agent needs the same thing | 20:20 |
wanyen | JoshNang, +1 | 20:20 |
jroll | wanyen: yes, the goal is to stop a user from turning off the node while cleaning is happening. for example, when firmware updates are running | 20:21 |
*** alexpilotti has quit IRC | 20:21 | |
wanyen | jroll, do we allow a user to cancel node cleaning? | 20:21 |
JoshNang | wanyen: hmm, i don't think so, but let me check | 20:22 |
JoshNang | wanyen: no we don't | 20:23 |
wanyen | JoshNang, it's probaly ok for kilo but we might want to consider that for Liberty. It will be complicated as some of the steps may not be canacellable after certain point. | 20:24 |
jroll | why would we need to cancel cleaning? | 20:25 |
JoshNang | wanyen: right, that's why i hadn't included it. you could potentially brick a node if you cancel and it isn't handled properly | 20:25 |
wanyen | jroll, sometine cleaning takes too long and user could change their mind. | 20:25 |
TheJulia | I think cancel is a bad description, abort maybe? | 20:26 |
jroll | wanyen: change their mind about what? what state would the node go to if it was cancelled? | 20:26 |
wanyen | TheJulia, abort indeed. | 20:26 |
wanyen | jroll, for instance, they might decide not to do firmware update. | 20:27 |
TheJulia | cancel has this implication something may be recoverable, abort should kind of just stop things | 20:27 |
*** ijw has quit IRC | 20:27 | |
jroll | wanyen: ok, so cleaning is something automatic, where I do not believe the user should have a choice. zapping is something triggered by a user, and I'm ok with being able to abort a zap process. | 20:28 |
jlvillal | Does anyone know if the IBM xCAT test honors the 'recheck" comment? | 20:29 |
wanyen | jroll, sounds reasonable for now. I need to think more about it. | 20:29 |
jlvillal | I'm curious since IBM xCAT posted a comment that it was starting the test and ending the test when I uploaded the patch. | 20:29 |
jlvillal | But I didn't see the comment on starting after I did 'recheck" | 20:29 |
wanyen | JoshNang and Jroll, ty for answering my questions. | 20:30 |
JayF | jlvillal: most 3rd part CIs are supposed to have their own keyword | 20:30 |
jroll | wanyen: no problem, I hope I was helpful | 20:31 |
jlvillal | JayF: Oh :( How does someone figure out that keyword? | 20:31 |
NobodyCam | off the wall question if you were going to check for a service, such as tftp, would netstat be the best option? | 20:31 |
JayF | https://wiki.openstack.org/wiki/ThirdPartySystems/IBM_xCAT_CI how to recheck should maybe go here | 20:31 |
jroll | jlvillal: I ignore the xcat CI so I have no idea | 20:31 |
jroll | NobodyCam: that or ps | 20:31 |
openstackgerrit | Merged openstack/ironic: Removed hardcoded IDs from "chassis" test resources https://review.openstack.org/133205 | 20:31 |
NobodyCam | ya ps has to many options | 20:31 |
jlvillal | JayF: Thanks. I had read that page earlier and didn't see any info. | 20:31 |
jroll | NobodyCam: what do you actually want to know? :P | 20:32 |
jlvillal | jroll: Okay. I won't worry about it too much :) | 20:32 |
NobodyCam | xinit.d tftpd bob's trival file transfer | 20:32 |
jroll | jlvillal: I'm also kind of a bad person | 20:32 |
jlvillal | :D | 20:32 |
jroll | take advice at your own risk :) | 20:32 |
NobodyCam | jroll: I was looking at https://bugs.launchpad.net/ironic/+bug/1282836 | 20:32 |
openstack | Launchpad bug 1282836 in Ironic "PXE driver does not manage tftpd" [Low,Confirmed] | 20:32 |
jroll | NobodyCam: oh. fun. | 20:33 |
NobodyCam | and figured I could just add a cehck to my https://review.openstack.org/#/c/167451 patch | 20:33 |
jroll | NobodyCam: so we'll refuse to start if tftp isn't running? | 20:34 |
jroll | seems heavy-handed, idk | 20:34 |
NobodyCam | jroll: nope fir that I was just going to log "tftp appears to not be running.. please start it" type message | 20:35 |
jroll | tftp can also be managed outside of ironic as well, TheJulia does this (with ironic managing /tftpboot), I do it without ironic managing /tftpboot | 20:35 |
wanyen | jroll and JoshNang, yes you are very helpful. :) | 20:35 |
JoshNang | :) | 20:36 |
jroll | NobodyCam: in my case, tftp runs on a different host than the conductor | 20:36 |
NobodyCam | ack, so maybe just the directory check is good enough | 20:36 |
jroll | maybe | 20:37 |
TheJulia | NobodyCam: locally or via tftp? | 20:37 |
jroll | just throwing out some cases to think about | 20:37 |
NobodyCam | TheJulia: I was trying to think of a pre-flight style check for running tftp service | 20:38 |
NobodyCam | but as jroll pointed out that might not be so easy | 20:38 |
firl | anyone able to help me diagnose why the node isn’t booting? http://bit.ly/1NB5eYR http://picpaste.com/Screen_Shot_2015-03-31_at_3.34.16_PM-YenPc3qN.png | 20:38 |
firl | it’s able to get to the point where it pulls a deploy image from ironic | 20:38 |
TheJulia | NobodyCam: what about a pre-flight check before powering on the node to double check that something is present via tftp? | 20:39 |
NobodyCam | TheJulia: I got kick back checking before each action | 20:40 |
NobodyCam | so I was thinking about a startup check | 20:40 |
NobodyCam | but .. | 20:40 |
*** david-lyle has quit IRC | 20:40 | |
jroll | firl: that looks pretty normal to me? | 20:41 |
TheJulia | firl: Looks normal to me, did you look in the conductor log? | 20:41 |
jroll | firl: have any conductor logs? | 20:41 |
firl | jroll: it just turns off after that | 20:41 |
firl | let me clear my log space and re run it | 20:41 |
jroll | firl: right, at that point the image writing is done and it should try to boot the instance | 20:41 |
firl | ( it turns off right after that ) | 20:41 |
TheJulia | firl: what does ironic node-show indicate? | 20:42 |
jroll | firl: yeah, would love to see conductor logs and know what driver you're running | 20:42 |
firl | http://pastebin.com/iHTCdAe0 | 20:43 |
firl | ipmi_pxe | 20:43 |
firl | regenerating conductor logs from start to finish now | 20:43 |
TheJulia | your instance_info is empty | 20:43 |
TheJulia | firl: your missing the information required in instance info for ironic to know what to put on the host, if that makes sense | 20:44 |
firl | gotcha | 20:44 |
jroll | I would think it wouldn't boot without that | 20:44 |
TheJulia | btw, you should now change your ipmi password | 20:44 |
jroll | are you using nova? | 20:44 |
firl | lol yes thank you | 20:44 |
firl | I forgot that it put it in clear text | 20:44 |
firl | yes nova | 20:44 |
jroll | yeah, the deploy should break way earlier with nova involved | 20:45 |
jroll | I'm guessing the deploy failed, nova rescheduled (which tears down the first node) | 20:45 |
firl | http://pastebin.com/JXQx4b6X | 20:45 |
firl | that is the instance info during boot | 20:45 |
*** mrda-away is now known as mrda | 20:46 | |
firl | maybe the swap being 0 is an issue? | 20:46 |
mrda | Morning Ironic | 20:46 |
jroll | firl: yeah, that lgtm, swap=0 should be fine | 20:46 |
firl | jroll kk | 20:46 |
firl | now it is deleting | 20:47 |
jroll | conductor logs should have all the answers | 20:47 |
firl | http://pastebin.com/mG4x8phN | 20:47 |
jroll | and maybe nova-compute logs, but mostly conductor | 20:47 |
firl | jroll: just finished the conductor log | 20:47 |
jroll | Mar 31 15:46:39 granger ironic-conductor: 2015-03-31 15:46:39.100 11885 ERROR ironic.drivers.modules.deploy_utils [-] StdErr: 'Error: The location 74753 is outside of the device /dev/sdb.\n' | 20:47 |
jroll | looks like maybe the disk is smaller than the flavor? | 20:48 |
firl | jroll: compute logs http://pastebin.com/z9vJtA7g | 20:49 |
firl | ok I will try to resize it down | 20:49 |
firl | giving it another go | 20:51 |
*** arif-ali has joined #openstack-ironic | 20:57 | |
* BadCub is done with bugs for today. Review Day pad has been updated with some new goodies | 20:59 | |
*** Marga_ has joined #openstack-ironic | 20:59 | |
openstackgerrit | Josh Gachnang proposed openstack/ironic: Allow IPA to request reboot after clean steps https://review.openstack.org/169504 | 20:59 |
*** trown is now known as trown|outttypeww | 21:01 | |
*** kkoski has quit IRC | 21:03 | |
*** Marga_ has quit IRC | 21:04 | |
firl | How do I have the ironic node show up as having a console? | 21:04 |
NobodyCam | firl: ironic node-set-console-mode | 21:06 |
firl | can’t set it while it’s locked by the host hah | 21:07 |
NobodyCam | ironic node-validate will check that console can be enabled | 21:07 |
NobodyCam | :-p | 21:07 |
firl | it has True on it | 21:08 |
firl | so I just assumed that the node would automatically have it | 21:09 |
NobodyCam | ahh :) needs to be enabled :) | 21:09 |
firl | yeah, I should probably put a wiki together on what all I had to do to get this all working lol | 21:10 |
NobodyCam | firl: my standard /shamelessPlug is we love to review doc patches :) | 21:10 |
firl | yeah, I am doing a PoC for my company for automated testing of build OS images on baremetal | 21:11 |
firl | so if I get it working end to end I will do that, then you can pick through and tell me where I didn’t do it proper haha | 21:11 |
NobodyCam | firl: if you see any area where our docs can be improved ... hint ... wink.. lol | 21:12 |
NobodyCam | :) | 21:12 |
* BadCub snickers at NobodyCam's shamless plugs and hints LOL | 21:12 | |
firl | I can put my thoughts in yeah, however things like having to specify dhcp options inside a port, or diagnosing why the dhcp client is passing a certain parameter when using the dnsmasq driver | 21:12 |
*** hj-hp has quit IRC | 21:13 | |
firl | my next task, once this is done, is to go through and figure out how to create the deploy images for our custom centos OS. In which we have a custom anaconda installer UI, which didn’t quite work with Oz | 21:13 |
*** achanda has joined #openstack-ironic | 21:14 | |
openstackgerrit | Josh Gachnang proposed openstack/ironic: Prevent power actions on node in cleaning https://review.openstack.org/169510 | 21:15 |
*** Marga_ has joined #openstack-ironic | 21:16 | |
openstackgerrit | Josh Gachnang proposed openstack/ironic: Prevent power actions on node in cleaning https://review.openstack.org/169510 | 21:17 |
NobodyCam | oh fooy :( I believe linc would prevent us from using :https://github.com/msoulier/tftpy | 21:17 |
JoshNang | JayF: ^ rebased | 21:17 |
BadCub | JoshNang: I added ^^ to the pad | 21:17 |
JoshNang | BadCub: thanks! | 21:18 |
BadCub | yw! | 21:18 |
*** ndipanov has quit IRC | 21:18 | |
*** mgoddard has joined #openstack-ironic | 21:19 | |
*** mgoddard has quit IRC | 21:20 | |
openstackgerrit | Jay Faulkner proposed openstack/ironic: Update docs: No power actions during cleaning https://review.openstack.org/169514 | 21:20 |
*** Sukhdev has joined #openstack-ironic | 21:22 | |
JoshNang | JayF: nice, thanks! +1 | 21:23 |
JayF | yeah lets try to, for the stuff that we took the time to document, to update in or as a followup patch | 21:23 |
JayF | that was easy to do now, pita to do it in Liberty when $who_knows_what has changed | 21:24 |
*** jgrimm is now known as zz_jgrimm | 21:26 | |
jroll | wow nice | 21:27 |
*** mjturek1 has quit IRC | 21:28 | |
JoshNang | JayF: ++ | 21:28 |
*** hj-hp has joined #openstack-ironic | 21:32 | |
openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent: Add sphinx build + basic documentation https://review.openstack.org/168503 | 21:33 |
JayF | All comments addressed on that ^ and I'd love to land it today if someone with core can take a look. | 21:34 |
* JayF reminds folks that all Ironic cores are IPA cores, and since I stol^W adopted the docs buidler from Ironic, it should be a good review :P | 21:34 | |
JayF | NobodyCam: ^ | 21:34 |
firl | A node that is stuck in deploying, is the only way to remove it via sql still? I found some older bugs that reference it | 21:34 |
JayF | firl: yes, and yes that does make all of us sad | 21:35 |
*** dprince has quit IRC | 21:35 | |
firl | womp womp | 21:35 |
firl | If I wanted to write a python method to force a delete ( and this would be valued ) where would I start | 21:35 |
JayF | I'd look more at something like the nova reset-state CLI command | 21:36 |
JayF | which doesn't force a delete, but puts a node in a sane state so things (like delete) can work | 21:36 |
firl | gotcha, nova also has a force command | 21:36 |
jroll | it needs to be in the API | 21:36 |
JayF | so if I had an instance in Nova stuck spawning; nova reset-state UUID (as an admin) would put the node in task_state:none/prov_state:error | 21:36 |
jroll | you can't directly update, say, target_provision_state | 21:36 |
jroll | (through the API) | 21:37 |
jroll | so it would have to be a new api endpoint | 21:37 |
openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent: Add sphinx build + basic documentation https://review.openstack.org/168503 | 21:38 |
openstackgerrit | Merged openstack/ironic: Prevent updating the node's driver if console is enabled https://review.openstack.org/168030 | 21:42 |
mrda | devananda: Not sure if you've seen my reply to your comments on 165666 but I think that's fine to proceed with now. | 21:44 |
openstackgerrit | Josh Gachnang proposed openstack/ironic: Fix cleaning nits https://review.openstack.org/168197 | 21:44 |
mrda | rloo: Review 169101 fixing up some backwards compatibility is now also ready for your re-review :) | 21:46 |
rloo | mrda: in a meeting, will try to look | 21:47 |
devananda | JayF: yay docs!!! | 21:48 |
mrda | np rloo | 21:48 |
TheJulia | NobodyCam: fwiw, that bug I was bashing my head against for a few days, I just removed the WIP, Happilly worked across 35 modes. | 21:49 |
NobodyCam | TheJulia: awesome | 21:50 |
NobodyCam | !!!! | 21:50 |
openstack | NobodyCam: Error: "!!!" is not a valid command. | 21:50 |
NobodyCam | lol | 21:50 |
mrda | ha! | 21:51 |
TheJulia | lol | 21:51 |
JayF | devananda: anyone you can suggest as a resource for docs builds failing with giant horrible tracebacks followed by "TypeError" | 21:56 |
JayF | heh | 21:56 |
* JayF suspects something is None somewhere, but no idea what | 21:56 | |
BadCub | TheJulia: I changed your patch on the pad from (WIP) to needs review | 21:58 |
firl | NobodyCam: how do i edit the docs for adding content for review? | 21:59 |
devananda | JayF: run locally? | 21:59 |
JayF | this is local | 21:59 |
JayF | Very intermittant failuires | 21:59 |
NobodyCam | firl: do you have a launchpad account? | 21:59 |
devananda | JayF: o.0 | 21:59 |
JayF | https://bugs.launchpad.net/ironic/+bug/1361295 | 21:59 |
openstack | Launchpad bug 1361295 in Ironic "gate-ironic-docs failed due to TypeError from sphinx" [High,Fix released] | 21:59 |
JayF | appears to be it | 21:59 |
JayF | but the fix implemented there no longer exists in ironic afaict | 22:00 |
firl | NobodyCam: no, signing up now | 22:00 |
openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent: Add sphinx build + basic documentation https://review.openstack.org/168503 | 22:01 |
NobodyCam | firl: :) also take a look at http://docs.openstack.org/developer/ironic/dev/contributing.html | 22:01 |
firl | ty | 22:01 |
NobodyCam | :) | 22:01 |
NobodyCam | the files are in the dos/source directory | 22:01 |
JayF | devananda: in the realm of docs, you should look at the cleaning docs that just landed | 22:01 |
NobodyCam | s/dos/docs/ | 22:01 |
BadCub | heya gang.. We gots a growing list of goodies that need reviews on the pad :-) | 22:02 |
devananda | JayF: indeed. i saw (but didn't do a full review) | 22:02 |
*** igordcard_ has joined #openstack-ironic | 22:03 | |
BadCub | jroll: NobodyCam & anyone else interested. Is there a good day this week to maybe pull together a bug squash? | 22:04 |
devananda | bugsquash ++ | 22:04 |
*** harlowja_ is now known as harlowja_away | 22:04 | |
NobodyCam | BadCub: tomorrow morning is good.. I have eye doctor in the afternoon so I expect I wont be able to see after that. or thurs, Fri is also good for me | 22:05 |
devananda | this would be a fairly low hanging task if there are newer contributors that want to help | 22:05 |
devananda | https://bugs.launchpad.net/ironic/+bug/1271317 | 22:05 |
openstack | Launchpad bug 1271317 in Ironic "XML support not required for ironic" [Low,Triaged] - Assigned to Michael Davies (mrda) | 22:05 |
BadCub | NobodyCam: I imagine you'll need a ride from Eye Dr tomorrow LOL | 22:05 |
*** absubram has quit IRC | 22:05 | |
NobodyCam | si | 22:06 |
firl | NobodyCam: sounds good, I will go through and see what I might be able to add | 22:06 |
NobodyCam | firl: awesome! thank you :) | 22:06 |
firl | I didn’t realize that I needed shellinabox to have the console enabled, was going to add that for example | 22:06 |
NobodyCam | :) ahh | 22:06 |
BadCub | How does Friday sound for bug squash then? | 22:07 |
*** harlowja_away is now known as harlowja_ | 22:07 | |
mrda | re: 1271317 will require touching every wsexpose method, and will need a bit of testing effort too. But certainly doable. | 22:07 |
NobodyCam | BadCub: works for me | 22:08 |
NobodyCam | morning mrda | 22:08 |
devananda | mrda: what if we monkeypatch wsexpose | 22:08 |
devananda | mrda: in the api base class | 22:08 |
devananda | or something | 22:08 |
BadCub | I will put a note on the pad for Friday :-) | 22:08 |
mrda | ooh, devananda, that's an idea | 22:08 |
devananda | should be like 10 lines | 22:08 |
mrda | if no-one else jumps on that, I'm happy to play on that today (not sure how it ended up on my plate again, but hey, happy to pick it up again) | 22:09 |
mrda | once I re-roll 163730 again | 22:09 |
devananda | mrda: I think the bug just didn't get untargeted from you? | 22:10 |
mrda | someone in channel told me they were working on it, but I forgot to follow thru and assign them the bug | 22:10 |
devananda | anyway, it ought to be fairly easy now, and it'd be nice if we officially drop xml rather than having it around another cycle | 22:10 |
mrda | that's all ancient histroy now though | 22:10 |
devananda | heh | 22:10 |
mrda | mea culpa | 22:10 |
openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent: Add sphinx build + basic documentation https://review.openstack.org/168503 | 22:11 |
devananda | on https://bugs.launchpad.net/bugs/1326269 i have found out that certain companies who shall remain nameless are using the testing drivers essentially in production | 22:11 |
openstack | Launchpad bug 1326269 in Ironic "Developer docs should include information about fake drivers" [Low,Triaged] - Assigned to Shraddha Pandhe (shraddha-pandhe) | 22:11 |
devananda | for reasons that i can't fathom right now | 22:12 |
devananda | i dont htink now's the right time to push https://review.openstack.org/152056 through -- but want to bring this to folks attention (again) | 22:13 |
BadCub | devananda: do you want me to add ^ to L pad? | 22:14 |
mrda | BadCub: can you post the link for that | 22:14 |
mrda | ? | 22:14 |
devananda | BadCub: definitely for discussion post RC. like i said, i dont think RC is the right time to push that | 22:14 |
BadCub | #link https://etherpad.openstack.org/p/ironic-liberty-priorities | 22:14 |
devananda | i think it's there already | 22:14 |
mrda | ta | 22:14 |
devananda | oh. it's not | 22:15 |
*** mtanino has joined #openstack-ironic | 22:15 | |
BadCub | devananda: kk. Will double check it for sanity sake :-) | 22:15 |
BadCub | added | 22:16 |
*** hj-hp has quit IRC | 22:18 | |
*** harlowja_ has quit IRC | 22:21 | |
NobodyCam | lol : https://tools.ietf.org/html/rfc7168 | 22:21 |
NobodyCam | last years april 1st rfc | 22:22 |
*** penick has quit IRC | 22:23 | |
BadCub | LOL | 22:23 |
mrda | oh rats, I forgot to get my patch ready - to change all spellings in Ironic to the English, rather than US dictionary. i.e. add back in all the 'u's and s/z/s/ conditionally :) | 22:24 |
NobodyCam | mrda: :-p shhh *dont tell BadCub* | 22:24 |
mrda | lol | 22:25 |
NobodyCam | hehehe | 22:25 |
mrda | the i18n team won't be happy with me... | 22:25 |
* BadCub looks up.. Say what? | 22:25 | |
BadCub | lol | 22:25 |
NobodyCam | hehehe | 22:25 |
NobodyCam | why is there no native python support for tftp file retrival | 22:26 |
mrda | NobodyCam: sounds like an opportunity for upstream contribution :-P | 22:26 |
*** hj-hp has joined #openstack-ironic | 22:26 | |
*** harlowja has joined #openstack-ironic | 22:26 | |
NobodyCam | there are several packages that do it. just didn't want to add anymore deps | 22:27 |
TheJulia | NobodyCam: umm.... it is a relatively simple protocol | 22:27 |
devananda | NobodyCam: https://tools.ietf.org/html/rfc7168#section-2.3.3 418 IM A TEAPOT | 22:27 |
NobodyCam | devananda: ya... | 22:27 |
NobodyCam | lol | 22:27 |
NobodyCam | i love that error | 22:28 |
firl | devananda, NobodyCam, jroll: Thanks again for all your help | 22:29 |
NobodyCam | :) | 22:29 |
firl | Just succesfully deployed the first bare metal ubuntu os | 22:29 |
NobodyCam | firl: w00t | 22:29 |
BadCub | firl: congrats :-) | 22:29 |
mrda | Oh, so we don't want 406s anymore, but we want 418s instead? I can fix thaty... | 22:30 |
NobodyCam | mrda: ++++ | 22:30 |
firl | thanks! :) | 22:30 |
NobodyCam | lol | 22:30 |
devananda | mrda: the error message body should be short | 22:30 |
devananda | and stout | 22:30 |
NobodyCam | lol | 22:30 |
BadCub | LOL | 22:30 |
mrda | good job firl! | 22:30 |
mrda | lol | 22:30 |
devananda | firl: woot! | 22:30 |
mrda | stout as in beer, or in stature, or size? | 22:31 |
BadCub | or all of the ^^ | 22:31 |
*** penick has joined #openstack-ironic | 22:31 | |
TheJulia | NobodyCam: of course I show the RFC to my girlfriend and she immediately complains that there must be an addendum for hot chocolate. | 22:32 |
mrda | or as in 'standard out'? | 22:32 |
mrda | TheJulia: next year is only 365 days away | 22:32 |
BadCub | LOL TheJulia | 22:32 |
NobodyCam | :) | 22:32 |
BadCub | If we are making addendum for hot chocolate, we should include hot buttered rum as well | 22:32 |
devananda | TheJulia: naturally, hot chocolate is prepared on a stove, not brewed in a teapot | 22:33 |
BadCub | devananda: well... technically... one could boil water/milk in a tea pot and pour into coco-mix | 22:33 |
TheJulia | mrda: I think the requirement is to have it filed by mid-march, but I don't remember the rules anymore. | 22:33 |
devananda | the original HTCPCP - https://tools.ietf.org/html/rfc2324 | 22:34 |
* BadCub ducks | 22:34 | |
TheJulia | BadCub: I'll put you two in touch and you can write the addendum. | 22:34 |
TheJulia | oh, wait a second, that might not be a good idea | 22:34 |
TheJulia | nvmd! | 22:34 |
devananda | Coffee is brewed using coffee pots. Networked coffee pots require a control protocol if they are to be controlled. | 22:34 |
BadCub | TheJulia: yeah, that could be dangerous! | 22:34 |
mrda | TheJulia: Just plan to have it up for review by summit, and then the reviews will be done in L, ready for next year with a cycle to spare :) | 22:35 |
* mrda thinks that everyone should run to the OpenStack cadence | 22:36 | |
TheJulia | *facepalm* | 22:36 |
mrda | That would make my life easier :) | 22:36 |
NobodyCam | brb | 22:39 |
firl | anyone have a good resource on how to make all the needed ironic image files from a iso boot source? | 22:40 |
openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent: Fix Sphinx Autodoc WARNING/ERROR in docs build https://review.openstack.org/169545 | 22:40 |
*** romcheg has quit IRC | 22:42 | |
devananda | firl: you said ISO | 22:44 |
*** andreykurilin_ has joined #openstack-ironic | 22:45 | |
firl | Yes, sorry, centos 6 image source specifically ( we generate a usb boot image, and ISO image ) | 22:45 |
devananda | firl: you realize ironic doesn't use scripted installation media? | 22:47 |
firl | I thought that I could script the install in qemu and create the image to test it on a physical machine | 22:50 |
*** devlaps has joined #openstack-ironic | 22:50 | |
NobodyCam | firl: have you ever looked at Disk-Iage-builder? | 22:51 |
NobodyCam | Disk-Image-bulider even? | 22:51 |
firl | lol | 22:51 |
NobodyCam | :-p | 22:51 |
firl | yeah I have, I was looking at that earlier, I was going to use that as a starting point, just didn’t know if anyone here had a better idea | 22:52 |
devananda | firl: diskimage-builder is sort of the common base line most of us use | 22:53 |
firl | sounds good. I will go through and research / utilize it. I just wanted to have a known good ironic system before going down that route | 22:53 |
firl | hopefully it will take less than 4 days to get it up like ironic did hah | 22:53 |
NobodyCam | firl: looks like a lot of steps: https://www.ibm.com/developerworks/community/wikis/home?lang=en#!/wiki/OpenStack/page/Creating+qcow2+CentOS+Image+for+OpenStack | 22:54 |
*** devlaps has quit IRC | 22:54 | |
devananda | firl: this is probably the simplest way to get started: http://docs.openstack.org/developer/ironic/deploy/install-guide.html#image-requirements | 22:54 |
firl | nobodycam: thanks for the link, I will check it out in depth | 22:55 |
devananda | it's also what we do CI with, so there's a fairly good chance it will _just_work_ :) | 22:55 |
firl | devananda: :) yeah I used those instructions for the fedora and ubuntu images | 22:55 |
NobodyCam | :) | 22:55 |
devananda | cool | 22:56 |
firl | ubuntu is up and running right now, however my ssh key authority didn’t get copied over, but figured that was just something I might have messed up | 22:56 |
NobodyCam | TheJulia: fyi I found a example of using python to get a tftp file | 22:56 |
TheJulia | NobodyCam: that... took you a while | 22:57 |
TheJulia | NobodyCam: was it well hidden on the internet? | 22:57 |
BadCub | TheJulia: NobodyCam is old and blind.. be nice now | 22:57 |
* BadCub ducks under desk | 22:57 | |
BadCub | lol | 22:57 |
NobodyCam | I wanted I single sctipt | 22:57 |
NobodyCam | doh | 22:57 |
TheJulia | NobodyCam: ahhh! | 22:57 |
TheJulia | BadCub: troublemaker :) | 22:57 |
* BadCub is not a troublemaker... I just give it a helping hand! LOL | 22:58 | |
NobodyCam | ehh, got a speack up.. /me is also getting hard of hearing | 23:00 |
BadCub | lol | 23:00 |
* BadCub notices NobodyCam's typos are getting worse with age too lol | 23:01 | |
NobodyCam | lol | 23:01 |
NobodyCam | doh | 23:01 |
NobodyCam | fingers are the second thing to go | 23:01 |
BadCub | lol | 23:02 |
Sukhdev | NobodyCam: can I ask a QQ | 23:03 |
NobodyCam | Sukhdev: sure :) | 23:03 |
Sukhdev | I am trying to use ironicclient API | 23:04 |
Sukhdev | trying to look for some sample use examples, have not been able to find | 23:04 |
Sukhdev | can you point me to an example? if you know of one? | 23:04 |
Haomeng | BadCub: Yes, agree with you, this 1436173 is dup with existing one 1300939, we can work wigh orig one, I found the basic translation issue with UnicodeEncodeError - https://review.openstack.org/#/c/168268/, so we have to make the basic translation work[1], and then go ahead to enable lazy translation(). | 23:05 |
Haomeng | BadCub: reply in the taht bug:) | 23:05 |
NobodyCam | Sukhdev: like the ironic nova driver? | 23:05 |
BadCub | Haomeng: total awesomeness!!!! | 23:05 |
Sukhdev | NobodyCam: yes | 23:05 |
NobodyCam | https://github.com/openstack/nova/blob/master/nova/virt/ironic/driver.py | 23:06 |
Sukhdev | NobodyCam: Basically I want to do an equivalent of "ironic port-show", "ironic node-list", etc - | 23:06 |
Sukhdev | NobodyCam: but, programatically | 23:06 |
Sukhdev | NobodyCam: from ML2 driver | 23:07 |
BadCub | Haomeng: do you want ot assign 1300939 to you? I have set it to rc-1 for K. We can tag 1426173 as a dup? | 23:07 |
Sukhdev | NobodyCam: that link does not help because I can not import it | 23:07 |
openstackgerrit | Merged openstack/ironic: Prevent power actions on node in cleaning https://review.openstack.org/169510 | 23:08 |
NobodyCam | Sukhdev: are you looking to run the command line client? | 23:08 |
openstackgerrit | Merged openstack/ironic: Followup to comments on Cleaning Docs https://review.openstack.org/169423 | 23:08 |
Sukhdev | NobodyCam: so, I thought I use pyrhon-ironicclient to achieve it - | 23:08 |
openstackgerrit | Merged openstack/ironic: Update docs: No power actions during cleaning https://review.openstack.org/169514 | 23:08 |
NobodyCam | or the python lib? | 23:08 |
NobodyCam | hahahaha DOCS | 23:09 |
Haomeng | BadCub: for this one https://bugs.launchpad.net/ironic/+bug/1300939, I worked on it for long time, but not working and no soultion now , blocking at - https://bugs.launchpad.net/wsme/+bug/1256042, so I am not sure I can find new way to support lazy translation now, you can check https://review.openstack.org/#/c/84362/ for details | 23:09 |
openstack | Launchpad bug 1300939 in Ironic "API does not translate errors based on Accept-Language header" [Low,Confirmed] - Assigned to Jim Rollenhagen (jim-rollenhagen) | 23:09 |
openstack | Launchpad bug 1256042 in WSME "Pecan's on_error() hook never gets trigged when used with WSME" [Undecided,Confirmed] - Assigned to Jianing Yang (jianingy) | 23:09 |
NobodyCam | Sukhdev: https://github.com/openstack/nova/blob/master/nova/virt/ironic/driver.py#L432-L437 | 23:09 |
NobodyCam | thats a example og node list via the Lib | 23:10 |
*** ijw has joined #openstack-ironic | 23:10 | |
NobodyCam | for the CLI it would be "ironic node-list" | 23:10 |
BadCub | Haomeng: I copied ^^ to the bug | 23:10 |
*** ijw has quit IRC | 23:10 | |
NobodyCam | Sukhdev: oh associated=True is NOT needed for fill list that is a filter for nides with instance id set | 23:11 |
Sukhdev | NobodyCam; that example is perfect - however, when I try to run it, I get the authentication error - hence, I was looking for a use case where and entrypoint (ironic end point) is passed to the ironicclient | 23:11 |
Haomeng | BadCub: https://bugs.launchpad.net/ironic/+bug/1300939 current owner is jim, not sure if he have idea to fix | 23:12 |
openstack | Launchpad bug 1300939 in Ironic "API does not translate errors based on Accept-Language header" [Low,Confirmed] - Assigned to Jim Rollenhagen (jim-rollenhagen) | 23:12 |
jroll | you can unassign me from that, I haven't looked at it in 6 months | 23:12 |
Haomeng | BadCub: I think if it is not in high prority, can we target to next-release? | 23:12 |
NobodyCam | Sukhdev: oh | 23:13 |
Haomeng | jroll: yes, i think we are blocking at 1256042, so no solution so far | 23:13 |
Sukhdev | NobodyCam: for example if I import ironicclient in python and then run that command (that you pointed me to) it fails - because of authentication error | 23:13 |
BadCub | Haomeng: I didn't target it based on convo & amount of bugs we have for rc1 | 23:13 |
*** igordcard_ has quit IRC | 23:13 | |
BadCub | jroll: I unassigned it for ya | 23:13 |
NobodyCam | Sukhdev: ironic needs admin creds | 23:14 |
jroll | BadCub: thanks | 23:14 |
BadCub | yw | 23:14 |
NobodyCam | Sukhdev: is this with in an openstack frame work? | 23:14 |
Sukhdev | NobodyCam: Yes, I am aware of it, but, where and how do I pass it to the API call so that it will allow me to access the DB? That is the example I am looking for and could not find one | 23:15 |
*** hj-hp has quit IRC | 23:15 | |
*** davideagnello has quit IRC | 23:15 | |
*** pcrews_ has joined #openstack-ironic | 23:15 | |
*** pcrews has quit IRC | 23:15 | |
BadCub | Haomeng: since you are working on the secondary, would you mind to keep an eye on the original one in case a solution pops_to_mind? | 23:16 |
*** hj-hp has joined #openstack-ironic | 23:16 | |
NobodyCam | Sukhdev: i'm not understanding, a API call to access the DB? | 23:16 |
NobodyCam | we have a http API | 23:16 |
Sukhdev | NobodyCam: I want to something like "ic = ironicclient.client(ironic-url, username, password, etc)….and can not seem to find example as to what needs to be passed to it | 23:16 |
openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent: Docs on HardwareManager, building, and workflow https://review.openstack.org/168522 | 23:17 |
NobodyCam | you can get smaples of that by adding --debug to the cli commands | 23:17 |
NobodyCam | oh | 23:17 |
JayF | jroll: I have 3 patchsets in a row. I need to make a one character change to the one in the middle; what's the best way you'd suggest to do that without breaking my world | 23:17 |
NobodyCam | Sukhdev: https://github.com/openstack/nova/blob/master/nova/virt/ironic/client_wrapper.py#L59-L89 | 23:18 |
Sukhdev | NobodyCam: once I have instance of "ic" then I can do ic.node.list(), etc.. | 23:18 |
NobodyCam | that | 23:18 |
jroll | JayF: fix the middle and rebase the end | 23:18 |
Haomeng | BadCub: yes, i will try, but not sure if we still depens on wsme defect 1256042, and try to check with wsme guys, but I can not promise now:) | 23:18 |
*** pcrews__ has joined #openstack-ironic | 23:18 | |
Sukhdev | NobodyCam: perfect - thanks - that is exactly I was looking for | 23:18 |
NobodyCam | :) | 23:18 |
*** pcrews__ has quit IRC | 23:18 | |
BadCub | Haomeng: you rock! No promises expected :-) | 23:18 |
Haomeng | BadCub: https://bugs.launchpad.net/wsme/+bug/1256042 | 23:19 |
openstack | Launchpad bug 1256042 in WSME "Pecan's on_error() hook never gets trigged when used with WSME" [Undecided,Confirmed] - Assigned to Jianing Yang (jianingy) | 23:19 |
NobodyCam | sorry I was missi understanding there | 23:19 |
Sukhdev | NobodyCam: I was looking in the tests (hoping they will set up the endpoint) | 23:19 |
Haomeng | Haomeng: :) | 23:19 |
openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent: Add sphinx build + basic documentation https://review.openstack.org/168503 | 23:19 |
mrda | JayF: Download the first review in the sequence, rebase -i master, choose to edit the middle one, git commit -a --amend, continue, hope it behaves correctly :) | 23:19 |
Sukhdev | NobodyCam: Thanks a ton... | 23:19 |
NobodyCam | :) sure | 23:19 |
openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent: Fix Sphinx Autodoc WARNING/ERROR in docs build https://review.openstack.org/169545 | 23:20 |
*** hj-hp has quit IRC | 23:20 | |
openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent: Docs on HardwareManager, building, and workflow https://review.openstack.org/168522 | 23:20 |
JayF | that was super easy, wow | 23:20 |
*** pcrews_ has quit IRC | 23:20 | |
mrda | JayF: :) | 23:21 |
JayF | I didn't know rebase button worked on dependent changes tbh | 23:21 |
JayF | that makes it way easier | 23:21 |
JayF | plus my first patch failed pep8 so I had more fixes anyway :P | 23:21 |
*** Sukhdev has quit IRC | 23:22 | |
JayF | NobodyCam: JoshNang: https://review.openstack.org/#/c/168503/7 should be landable right meow | 23:22 |
JayF | https://review.openstack.org/#/c/169545/2 is a no-brainer kind of review too :) | 23:22 |
NobodyCam | JayF: just a rebase? | 23:23 |
*** davideagnello has joined #openstack-ironic | 23:24 | |
JayF | NobodyCam: you never had a + on any of these; first one had -1 from you and lucas I fixed, 2nd patch didn't exist yesterday :) | 23:24 |
NobodyCam | ya but my - was for version number :-p | 23:24 |
JayF | enough substance has changed you should have a look | 23:24 |
NobodyCam | looking | 23:25 |
NobodyCam | :-p | 23:25 |
* devananda wanders back to the hotel to find sleep | 23:26 | |
NobodyCam | have a good night devananda | 23:26 |
BadCub | g'night devananda | 23:26 |
*** yuanying has joined #openstack-ironic | 23:26 | |
NobodyCam | JayF: PYTHONHASHSEED=0 needed? | 23:28 |
JayF | NobodyCam: yes, or the builds randomly fail. see the bug I linked earlier | 23:28 |
JayF | it clearly indicates something fucky is going on but I can't figure out what, so I used that to get it to build consistently | 23:28 |
JayF | NobodyCam: https://bugs.launchpad.net/ironic/+bug/1361295 | 23:29 |
openstack | Launchpad bug 1361295 in Ironic "gate-ironic-docs failed due to TypeError from sphinx" [High,Fix released] | 23:29 |
JayF | NobodyCam: and JoshNang confirmed that without the PYTHONHASHSEED=0 he wasn't getting things to reliably run either | 23:29 |
NobodyCam | ya thats what I was recalling .. trying to think what the fix was | 23:29 |
JoshNang | yeah, i checked because...well...that seems like an unlikely fix | 23:30 |
JayF | looks like venv is still set to that | 23:30 |
JayF | in Ironic | 23:30 |
NobodyCam | ack | 23:30 |
JayF | I might, if anything, have to add this to tox.ini for IPA testenv:venv as well once I turn jenkins on for these docs | 23:30 |
JayF | but I want to land them first | 23:30 |
JayF | then make a docs job -- one that will fail in IPA if someone makes a docstring that WARN/ERR in the build | 23:30 |
JayF | since IPA is a mostly developer-consumed project seems to me it should have great dev docs, even if most of them are generated | 23:31 |
NobodyCam | JayF: :) | 23:31 |
*** andreykurilin_ has quit IRC | 23:31 | |
*** penick has quit IRC | 23:33 | |
*** wanyen has quit IRC | 23:33 | |
NobodyCam | JayF: +a'd | 23:33 |
JayF | NobodyCam: the followup is dead simple if you wanna get that too | 23:35 |
*** rfchapman has quit IRC | 23:35 | |
JayF | that last one is about 170 lines of docs though, so you might wanna save that one for just before beer:thirty | 23:35 |
JayF | lol | 23:35 |
*** achanda has quit IRC | 23:41 | |
NobodyCam | lol | 23:41 |
* BadCub is thinking it is cocktail time already | 23:41 | |
*** rfchapman has joined #openstack-ironic | 23:42 | |
rfchapman | NobodyCam: Hey hows it going? | 23:43 |
NobodyCam | hey rfchapman doing good.. how many nodes are you deploying now? | 23:44 |
NobodyCam | :-P | 23:44 |
NobodyCam | how are you doing | 23:44 |
rfchapman | currently 3.... I have deployed upto 10 at one time, but I'm still having issues with the metadata portion of cloud-init. | 23:45 |
NobodyCam | :( | 23:45 |
NobodyCam | what kind of error? | 23:45 |
rfchapman | I tried to build an image with a default root ssh key but something doesn't seem to have worked. | 23:46 |
*** pcrews has joined #openstack-ironic | 23:47 | |
NobodyCam | rfchapman: are you seeing errors or is it just not workoing? | 23:47 |
NobodyCam | working even :-p | 23:47 |
rfchapman | [ 129.180402] cloud-init[467]: 2015-03-31 18:45:52,338 - url_helper.py[WARNING]: Calling 'http://169.254.169.254/2009-04-04/meta-data/instance-id' failed [119/120s]: request error [HTTPConnectionPool(host='169.254.169.254', port=80): Max retries exceeded with url: /2009-04-04/meta-data/instance-id (Caused by <class 'socket.error'>: [Errno 101] Network is unreachable)] | 23:47 |
*** Marga_ has quit IRC | 23:47 | |
jroll | we really need some sort of doc on setting up the metadata service | 23:47 |
jroll | some people have solved this but I have no idea how | 23:47 |
jroll | "routing magic" is all I know | 23:48 |
*** Marga_ has joined #openstack-ironic | 23:48 | |
BadCub | hey jroll did you see the mention of a Friday bug squash? | 23:50 |
rfchapman | NobodyCam: It will eventually give up.. and it tries another ip address on my network, but when I manually try to connect to that address it doesn't repond. | 23:50 |
JayF | BadCub: Friday is a nogo for j* | 23:50 |
rfchapman | Jroll: that would be nice... | 23:50 |
JayF | BadCub: Friday is our annual company outing to angel island \o/ | 23:50 |
BadCub | SWEET! | 23:50 |
BadCub | How would Thursday work? | 23:50 |
jroll | thursday is fine | 23:51 |
BadCub | NobodyCam: you good with Thursday? | 23:51 |
NobodyCam | rfchapman: your using dib | 23:52 |
NobodyCam | BadCub: ya | 23:52 |
BadCub | Awesome. Ya'll rock. TY :-) | 23:52 |
rfchapman | NobodyCam: yes | 23:52 |
rfchapman | NobodyCam: It would be nice if there were a ENV Var that I could set with the password for root while testing/debugging. | 23:54 |
rfchapman | I mean a DIB_VAR | 23:55 |
NobodyCam | add the stackuser element | 23:55 |
openstackgerrit | Michael Davies proposed openstack/ironic: New field 'name' not supported in port REST API https://review.openstack.org/163730 | 23:55 |
*** achanda has joined #openstack-ironic | 23:55 | |
NobodyCam | rfchapman: maybe something like: iptables -t nat -A PREROUTING -d 169.254.169.254/32 -i <interface> -p tcp -m tcp --dport 80 -j DNAT --to-destination <ip>:8775 | 23:59 |
jroll | how do you do that withuot access to the box | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!