Monday, 2015-05-11

openstackgerritNaohiro Tamura proposed openstack/ironic: Add whole disk image support for iscsi_irmc driver  https://review.openstack.org/18177100:02
*** lsmola_ has joined #openstack-ironic01:01
*** dtantsur|pto has quit IRC01:03
*** dtantsur has joined #openstack-ironic01:03
*** lsmola has quit IRC01:04
mrdaThis is a small Nova review affecting the Ironic Driver.  https://review.openstack.org/#/c/179960  You're welcome to review this (trivial) change.01:06
*** yuanying has joined #openstack-ironic01:10
*** dtantsur has quit IRC01:10
*** dtantsur has joined #openstack-ironic01:11
*** lsmola_ has quit IRC01:11
*** lsmola_ has joined #openstack-ironic01:25
*** chenglch has joined #openstack-ironic01:28
*** achanda has joined #openstack-ironic01:55
*** Shrews has joined #openstack-ironic02:00
*** kan_ has joined #openstack-ironic02:24
*** achanda has quit IRC03:01
*** achanda_ has joined #openstack-ironic03:04
*** achanda_ has quit IRC03:14
*** alex_xu has quit IRC03:20
*** alex_xu has joined #openstack-ironic03:21
*** subscope has joined #openstack-ironic03:38
*** Nisha has joined #openstack-ironic03:54
*** rameshg87 has joined #openstack-ironic04:14
rameshg87good morning ironic04:14
*** ramineni has joined #openstack-ironic04:17
pshigerameshg87: morning :)04:19
*** coolsvap|afk is now known as coolsvap04:23
rameshg87morning pshige04:33
openstackgerritMichael Davies proposed openstack/python-ironicclient: Cache negotiated api microversion for this server  https://review.openstack.org/17367404:37
mrdaMorning pshige and rameshg8704:45
rameshg87morning mrda04:46
mrdao/04:46
openstackgerritMerged stackforge/proliantutils: Python 3 compatibility  https://review.openstack.org/18107204:51
*** saripurigopi has joined #openstack-ironic04:58
*** kalpase has joined #openstack-ironic05:02
*** chenglch has quit IRC05:03
*** chlong has quit IRC05:09
openstackgerritNisha Agarwal proposed stackforge/proliantutils: test cases for get_nic_capacity and ipmi.py  https://review.openstack.org/17971005:10
openstackgerritNisha Agarwal proposed stackforge/proliantutils: parsing storage data fails in some scenarios  https://review.openstack.org/18004305:13
*** achanda has joined #openstack-ironic05:37
*** saripurigopi has quit IRC05:37
pshigemrda: morining :)05:37
mrdao/05:38
*** saripurigopi has joined #openstack-ironic05:57
*** ukalifon1 has joined #openstack-ironic06:23
*** athomas has joined #openstack-ironic06:24
*** jcoufal has joined #openstack-ironic06:31
*** yog__ has joined #openstack-ironic06:35
openstackgerritNisha Agarwal proposed stackforge/proliantutils: test cases for get_nic_capacity and ipmi.py  https://review.openstack.org/17971006:37
openstackgerritMerged openstack/ironic: Fix spelling error in docstring  https://review.openstack.org/18148106:48
*** Marga_ has joined #openstack-ironic07:01
*** dlpartain has joined #openstack-ironic07:03
*** Marga_ has quit IRC07:04
*** subscope has quit IRC07:05
openstackgerritNaohiro Tamura proposed openstack/ironic: Add IPA support for iscsi_irmc driver  https://review.openstack.org/18180707:06
*** dlpartain has left #openstack-ironic07:08
openstackgerritKan proposed openstack/ironic: Enable 'text' format sort for enterprise databases  https://review.openstack.org/18180907:12
openstackgerritKan proposed openstack/ironic: Enable 'text' format sort for enterprise databases  https://review.openstack.org/18180907:15
*** achanda has quit IRC07:15
*** takadayuiko has joined #openstack-ironic07:19
*** Marga_ has joined #openstack-ironic07:19
*** Marga_ has quit IRC07:19
openstackgerritSHIGEMATSU Mitsuhiro proposed openstack/ironic: Fix typos in Ironic docs  https://review.openstack.org/18181307:42
*** Nisha_away has joined #openstack-ironic07:44
*** Nisha has quit IRC07:44
*** achanda has joined #openstack-ironic07:45
*** ifarkas has joined #openstack-ironic07:49
*** uggla has joined #openstack-ironic07:52
ugglaHello07:54
*** rsalevsky has joined #openstack-ironic07:56
*** jistr has joined #openstack-ironic07:58
*** achanda has quit IRC08:01
*** lucasagomes has joined #openstack-ironic08:04
*** derekh has joined #openstack-ironic08:22
*** yog__ has quit IRC08:28
takadayuikodtantsur: is there?08:28
*** mgoddard has joined #openstack-ironic08:29
kan_ lucasagomes: hi, lucas :)08:35
lucasagomeskan_, hi there08:37
lucasagomestakadayuiko, dtantsur is on PTO still I believe08:37
kan_lucasagomes: I have a patch about the sort keys   https://review.openstack.org//#/c/177578/08:37
kan_lucasagomes: Clis gave an advice about not to 'make resource_fields DRYer'08:37
lucasagomeskan_, right, will take a look soonish08:38
kan_lucasagomes: We want the suggestions from cores, can you take a little time to have a look and give some comments?08:38
takadayuikolucasagomes: Hi :) Thanks for reply. PTO??08:38
lucasagomeskan_, sure no problem!08:38
kan_lucasagomes: thx08:38
kan_:)08:38
lucasagomestakadayuiko, oh it means holidays :-) PTO = paid time off08:38
lucasagomeskan_, thank you for the patch :D08:39
kan_:D08:39
*** Nisha has joined #openstack-ironic08:39
takadayuikolucasagomes: Ah, OK, thank you. And are you also on holiday?08:39
*** Nisha_away has quit IRC08:40
lucasagomestakadayuiko, I wish! But unfortunately not :-(08:40
lucasagomesafter canada I will take some for sure08:40
takadayuikolucasagomes: Ahaha :) Not national holyday, but "indivisual" holiday, I understand.08:41
lucasagomestakadayuiko, oh yes, he took some personal holidays :-)08:42
lucasagomesbut we also live in diff countries, dtantsur is based in czech republic08:42
takadayuikolucasagomes: And you're living in Ireland, right?08:43
lucasagomesso national holidays doesn't for both usually08:43
lucasagomestakadayuiko, yup08:43
lucasagomesdoesn't apply*08:43
takadayuikoDo you know when he will back? in this week?08:44
takadayuikoI'd like to discuss about discoverd in Summit... of course, with you :)08:44
lucasagomestakadayuiko, :D hmm lemme try to find out when he will be back. I think probably at the end of this week08:45
*** achanda has joined #openstack-ironic08:45
*** yog__ has joined #openstack-ironic08:45
openstackgerritNisha Agarwal proposed stackforge/proliantutils: parsing storage data fails in some scenarios  https://review.openstack.org/18004309:01
*** tiagogomes has joined #openstack-ironic09:05
*** edcragg has joined #openstack-ironic09:08
*** yog__ has quit IRC09:12
kan_lucasagomes: Hi, lucas. Thanks for your comment in patch 181809. Actually it is the patch 177578 need your comments. could you take a loot at this?09:17
lucasagomeskan_, yup09:17
kan_lucasagomes: https://review.openstack.org//#/c/177578/09:17
kan_lucasagomes: thanks :D09:17
*** kalpase has quit IRC09:23
*** yog__ has joined #openstack-ironic09:23
openstackgerritNisha Agarwal proposed stackforge/proliantutils: parsing storage data fails in some scenarios  https://review.openstack.org/18004309:24
*** derekh has quit IRC09:27
*** derekh has joined #openstack-ironic09:28
*** kalpase has joined #openstack-ironic09:38
*** achanda has quit IRC09:40
*** subscope has joined #openstack-ironic09:43
*** naohirot has quit IRC09:50
openstackgerritNisha Agarwal proposed stackforge/proliantutils: parsing storage data fails in some scenarios  https://review.openstack.org/18004309:51
*** athomas has quit IRC10:00
*** Nisha has quit IRC10:01
*** athomas has joined #openstack-ironic10:08
*** mrmoje has quit IRC10:14
*** mrmoje has joined #openstack-ironic10:16
*** kan_ is now known as kan_afk10:17
*** kalpase has quit IRC10:24
*** kalpase has joined #openstack-ironic10:25
*** pelix has joined #openstack-ironic10:30
*** kalpase has quit IRC10:33
openstackgerritMerged openstack/ironic-specs: Fix accumulated nits in iRMC power, mgmt, and deploy specs  https://review.openstack.org/17706510:34
*** yuanying has quit IRC10:34
*** kalpase has joined #openstack-ironic10:34
*** coolsvap is now known as coolsvap|afk10:48
* rameshg87 goes home10:55
*** rameshg87 has quit IRC10:55
*** ramineni1 has joined #openstack-ironic11:00
*** ramineni has quit IRC11:01
openstackgerritShivanand Tendulker proposed openstack/ironic: Update iLO documentation for UEFI secure boot  https://review.openstack.org/17858211:02
*** kan_afk has quit IRC11:06
*** zhenguo has joined #openstack-ironic11:10
*** EmilienM|afk is now known as EmilienM11:28
*** lucasagomes is now known as lucas-hungry11:41
openstackgerritchenglch proposed openstack/ironic-specs: Add console log support  https://review.openstack.org/16458611:43
*** subscope has quit IRC11:44
*** subscope has joined #openstack-ironic11:47
*** trown|outttypeww is now known as trown11:51
*** kbs has quit IRC11:55
*** dprince has joined #openstack-ironic12:02
*** kalpase has left #openstack-ironic12:02
*** ramineni1 has quit IRC12:04
*** Marga_ has joined #openstack-ironic12:12
*** _dguerri is now known as dguerri12:13
*** Marga_ has quit IRC12:21
*** Marga_ has joined #openstack-ironic12:22
*** TheJulia has joined #openstack-ironic12:24
*** nagyz has joined #openstack-ironic12:25
nagyzhi12:25
nagyzI've started to play with ironic (kilo) and I can get as far as booting a node. however, after successful kernel + ramdisk load I get an initramfs shell. Isn't it supposed to boot from a ramdisk and dd from iSCSI to the local disk?12:25
TheJulianagyz: that would largely depend on the settings and the driver.  It sounds like there may be something is mixed up with the settings.12:27
nagyzideally I wouldn't like to have local disks at all just iSCSI, but sounds like I must have local disks for ironic to work with?12:27
nagyzthe drivers are IPMI + PXE12:27
TheJulianagyz: That is how ironic is designed, to burn an image to the local storage to provision the node into whatever role it is required12:29
TheJuliaThere is talk of supporting block storage, but there are many questions of how that would be orchustrated appropriately12:29
nagyzfor now I'd be happy to just get local storage working :-)12:30
TheJuliaI've not done much with the pxe_ipmitool driver, largely I've played with agent_ipmitool12:32
TheJuliathe dropping to a shell makes me wonder if the deployment image is correct, additionally I can't help but wonder if command line being passed for boot is correct12:32
nagyzso what's the difference between the agent and the normal one? (I'm happy to RTFM but the docs I'm reading isn't that great...)12:33
TheJuliabottom line, the pxe_ipmitool driver bottlenecks on the ironic conductor, and thus iscsi gets used to write the image out to the disk.  The agent essentially calls back to the ironic api and says "I'm here, what are my orders" and it goes and does the required steps locally12:34
TheJuliahopefully that 10,000ft difference makes sense12:34
TheJulia:)12:34
nagyzsure12:34
nagyzbut in order to run the agent it has to run in-memory before any local disk - so does it get loaded as part of the ramdisk?12:35
nagyzI can change the driver and not touch anything else so we can try with that - does that make sense to do?12:35
TheJuliaIt is a special image that is loaded into RAM and executed, like the ramdisk, but different12:35
TheJuliaI honestly would try running with the agent, there is information on troubleshooting, and it is the preferred path moving forward12:36
TheJuliaYou'll need to have the IPA agent kernel/ramdisk image on hand for it to work though12:36
nagyzlet me look into the docs on that12:37
nagyzhttp://docs.openstack.org/developer/ironic/deploy/install-guide.html12:37
*** lucas-hungry is now known as lucasagomes12:37
nagyzthis is the one I'm using12:37
nagyzthere is 0 match for "IPA" :)12:37
TheJulia IPA is shorthand for ironic python agent12:38
*** saripurigopi has quit IRC12:38
TheJuliahttp://tarballs.openstack.org/ironic-python-agent/coreos/files/  Pre-built kernel and ramdisk :)12:39
*** kbs has joined #openstack-ironic12:40
nagyzgreat, let me look into changing the config and using this image12:41
nagyzis coreos used just to provision an image to the local disk or is this going to actually deploy coreos?12:41
nagyzI saw there is a separate "deploy image" and "image"12:41
TheJuliait is used to deploy the image12:42
TheJuliathe Ironic Python Agent is loaded into the image as a container12:42
TheJuliaso coreos fires up, loads the container, the agent calls home, and magic begins happening12:42
nagyzsounds good to me. I just need to look into reconfiguring the node's agent I guess and loading the image to glance12:45
TheJuliaIPA can load images from HTTP12:46
TheJuliaor HTTPS12:46
nagyzbut the IPA image needs to be added to glance, I guess?12:46
nagyzI see it's in cpio which is not any of glance's formats, right...?12:46
TheJulianope, as long as it is available and referenced approprately for the pxe/ipxe boot sequence12:47
nagyzso is iPXE preferred instead of the NIC's built-in PXE?12:47
*** rloo has joined #openstack-ironic12:47
TheJulianagyz: the preferred method is PXE to bootstrap ipxe, to retrieve the image from wherever it is located12:48
nagyzoh I see the coreos files are referenced in the iPXE config12:48
*** early has quit IRC12:56
*** early has joined #openstack-ironic12:57
nagyzso I've added a new node with the agent and when I run validate, it tells me this12:57
nagyz| deploy     | False  | Node cd1091f5-9129-46d2-988f-2e7959cf01b7 failed to validate deploy image info. Some parameters were missing. Missing are: ['driver_info.deploy_kernel', 'driver_info.deploy_ramdisk', 'instance_info.image_source'] |12:57
nagyzbut since iPXE is using loading the drivers, what should I specify here?12:57
nagyzas they don't need to be in glance12:57
*** kkoski has joined #openstack-ironic12:58
TheJulianagyz: so you will need to define them.  The image source is the full disk image that you want to have written out to the server. deploy_kernel and deploy_ramdisk should point to where the IPA kernel and ramdisk can be found12:59
nagyzright but if I upload the images to glance how will iPXE reference them?13:01
nagyzalso, what format is a cpio from glance's perspective?13:01
TheJuliayou can't put the coreos image image in glance13:01
TheJuliayou'll need to feed it out via tftp or http most likely13:01
TheJuliathe image_source can be the reference to the image in glance if memory serves13:01
nagyzso I can just set the deploy_kernel and deploy_ramdisk to a http url, you mean?13:02
nagyzsounds wicked :)13:02
TheJuliayup13:02
*** ukalifon1 has quit IRC13:02
*** ukalifon has joined #openstack-ironic13:02
jrollTheJulia: nagyz: the ramdisk image can certainly be in glance :)13:03
jrollalso, g'morning13:03
TheJuliaWell, yeah, it can13:05
TheJuliabecause the conductor will retreieve it13:05
lucasagomesmorning jrist TheJulia nagyz13:06
jrolldon't forget about me lucasagomes :P morning13:06
TheJuliagood morning13:06
nagyzgood morning13:07
nagyzso I've added http links to both the ramdisk and the kerner image, and this is what I get for validation: Node cd1091f5-9129-46d2-988f-2e7959cf01b7 is configured to use the agent_ipmitool driver which currently does not support deploying partition images.13:07
lucasagomesjroll, ohhh damn... always hah13:07
lucasagomesjrist, sorry again hah13:08
nagyz(I've linked the coreos cpio as the kernel... which feels weird at 203MB.)13:08
trownlucasagomes: I was thinking a lot about distributed introspection this weekend, I think it would be much simpler if we were using the agent instead of a discovery ramdisk as we get the "check-in" for free13:08
trownhave you thought about the distributed case much?13:09
jrollnagyz: right, so the agent doesn't support partition images yet, you'd need to use whole disk images13:09
lucasagomestrown, the DAHC?13:09
trownalso, good morning jroll TheJulia nagyz lucasagomes13:09
TheJulianagyz: can you share a link to a paste of your ironic node-show output?13:09
trownlucasagomes: ya13:09
TheJulianagyz: naturally, with anything sensitive redacted13:09
lucasagomestrown, I haven't thought much about it since the last time we talked. But yeah I think that IPA would facilitate much of the stuff we currently do with the discovery/dib ramdisks13:09
nagyzTheJulia, I thought I should just use the coreos images as is? but sure, give me a moment to do a redacted pastebin13:10
trownlucasagomes: the problem that seems a little bit tricky with the discovery ramdisk is coordination, but the agent is already coordinated13:10
jrollnagyz: your deploy image is fine, it's the image_source that needs to be a full disk image13:11
lucasagomestrown, yeah, the only thing is that, for DAHC it requires a server to orchestrate the machines booting right?13:11
lucasagomesI don't think this server would be Ironic tho13:11
nagyzjroll, right I should read more carefully.13:11
nagyzdoes it actually checks the partitionedness of the image?13:11
*** kan_ has joined #openstack-ironic13:11
jrollnagyz: indeed13:11
trownlucasagomes: I am not to sure about DAHC proper, I am more thinking about how to provide the same information in a more generic way13:12
lucasagomestrown, IIRC and if we are mimic'ing what eDeploy does for DAHC. The server get's a file that describe the tests as inut13:12
lucasagomesinput*13:12
lucasagomesand it contains things like what tests it should run and the number of machines it will use13:12
trowni.e. as a cloud installer I want to make sure my network topology has connectivity13:12
lucasagomesso it goes and power on those machines and orchestrate them13:12
trownlucasagomes: ya, that is how edeploy does it13:13
nagyzTheJulia, here's the pastebin: http://pastebin.com/K6GZB8DN13:13
lucasagomestrown, but yeah, that said, when reporting back to the nodes it would be good to have a lookup13:13
nagyzTheJulia, I guess I need to look into getting an unpartitioned source image...13:13
lucasagomesand findout which node is it13:13
nagyzwould the stock ubuntu qcow2 image be such an image, actually?13:13
lucasagomestrown, perhaps we should brainstorm how this is going to work for the distributed case13:13
lucasagomestrown, but yeah +1 for ipa :-)13:14
trownlucasagomes: ya, I have some ideas...really need dtantsur though13:14
lucasagomestrown, he will be back on thursday I think13:14
lucasagomestrown, maybe we can set up a call or something to discuss it13:14
TheJulianagyz: you will actually want an image that has a partition table burned into it13:14
TheJulianagyz: and a boot loader13:14
nagyzTheJulia, I guess I should build such an image with disk-image-builder?13:15
trownlucasagomes: my idea is to have an endpoint on discoverd that takes a list of checks we want to perform, and then that sends the checks to the agents13:15
trownwhere checks could be implemented by many different drivers13:15
TheJulianagyz: good idea :)13:15
trownlucasagomes: +1 to a call to discuss with dtantsur13:16
lucasagomestrown, oh, right. Yeah sounds good. I like the idea of having a dummy ramdisks that waits the server inputs to tell it what to do13:16
lucasagomestrown, sounds more sane13:16
trownya13:16
lucasagomestrown, mind sending an invite ? On friday maybe?13:16
trownlucasagomes: sure thing13:16
nagyzTheJulia, so I see disk-image-create builds a qcow2 image plus a ramdisk plus the kernel separately. should looking at it's help (which I just did) tell me exactly how to get it to build the "right" image? I don't see any mention of "unpartitioned" or "built-in kernel"13:19
*** ramineni has joined #openstack-ironic13:20
TheJulianagyz: pass the vm element into the list which will load the bootloader.  If memory serves dib still writes out the ramdisk/kernel separately, but the resulting qcow2 is bootable13:21
*** chlong has joined #openstack-ironic13:22
*** cinerama has quit IRC13:22
*** cinerama has joined #openstack-ironic13:26
nagyzTheJulia, I'm not sure I follow.13:26
nagyzwhat do you mean by the vm element? :)13:27
TheJulianagyz: When you use diskimage-builder, on the command line, you pass a list of elements to be used in the construction of the image.  One of the elements is named vm.  You can view it https://github.com/openstack/diskimage-builder/tree/master/elements/vm  An example of where you can see this being done is located at https://github.com/juliakreger/bifrost/blob/master/playbooks/roles/ironic-instal13:28
TheJulial/tasks/create_dib_image.yml13:28
*** jgrimm is now known as zz_jgrimm13:32
nagyzTheJulia, thanks let me see13:37
nagyzTheJulia, so we're telling dib to actually create a vm image but we're using it for bare-metal..13:41
TheJuliakind of.. yeah :(13:41
TheJuliaIt is a badly named element13:41
TheJulianagyz: the important part is that it puts a bootloader at the beginning of the image13:42
nagyzso I see in the playbook you're also passing in the http url for the deploy image?13:43
BadCubmorning folks13:44
TheJuliaYes13:44
TheJuliagood morning BadCub13:44
BadCubmorning TheJulia13:45
NobodyCamgood morning Ironicers13:45
TheJuliagood morning NobodyCam13:46
NobodyCammorning TheJulia :)13:46
BadCubmorning NobodyCam13:47
*** yog__ has quit IRC13:47
NobodyCammorning BadCub13:47
NobodyCam:)13:47
nagyzTheJulia, ok, built one, validate runs. let's see if it boots :-)13:48
nagyzis that normal that nova sees each ironic host as a hypervisor?13:50
nagyzwell I have only one but that is shown with it's ID as a hypervisor13:50
jrollyes13:50
TheJuliaYes13:50
NobodyCammorning jroll :)13:50
jrollheya NobodyCam :)13:51
NobodyCam:)13:53
nagyzvalidate should check if CPU and memory is specified... as I don't have it in my node config right now and nova scheduler throws an error (can't find available nodes)13:54
nagyzor at least I think that's what triggers in13:54
*** saripurigopi has joined #openstack-ironic13:55
nagyz*it13:55
nagyzalso not very clear if the cpus property is cores or sockets13:56
nagyzeven after adding these properties it can't find any host13:57
openstackgerritOpenStack Proposal Bot proposed openstack/ironic: Updated from global requirements  https://review.openstack.org/18120413:58
openstackgerritLucas Alvares Gomes proposed openstack/ironic: Fix and enhance "Exercising the Services Locally" docs  https://review.openstack.org/18016314:00
nagyzok found 2 issues14:02
nagyz1, after adding the memory/cpu/disk I needed to restart nova-compute to pick it up in the scheduler - is that normal?14:02
nagyz2, I forgot to add the ports in ironic so it went into an error state with a nice stacktrace due to it getting 404 from neutron - this should be pointed out at validation, no?14:02
jroll1) it should pick it up "eventually", when the next resource tracker loop runs14:03
*** mgoddard has quit IRC14:03
jroll2) since ironic can be used without nova, we may not want to validate that, I'm not sure14:03
nagyzbut even without nova wouldn't people want to make sure the mac addresses are in ironic?14:04
nagyzlet me see if adding it helps14:04
*** mgoddard has joined #openstack-ironic14:04
jrolltrue14:04
jrollfeel free to file a bug that we can bike-shed about :)14:04
nagyz:D14:04
*** saripurigopi has quit IRC14:04
nagyzsure14:04
*** openstackgerrit has quit IRC14:06
*** openstackgerrit has joined #openstack-ironic14:07
*** saripurigopi has joined #openstack-ironic14:09
openstackgerritYuiko Takada proposed stackforge/ironic-discoverd: Use keystonemiddleware auth credentials  https://review.openstack.org/17078214:12
openstackgerritYuiko Takada proposed stackforge/ironic-discoverd: Use keystonemiddleware auth credentials  https://review.openstack.org/17078214:18
openstackgerritYuiko Takada proposed stackforge/ironic-discoverd: Load authenticate token from HTTP header  https://review.openstack.org/17423114:18
*** openstack has joined #openstack-ironic14:20
BadCubmornin rloo :)14:20
jrollgood morning rloo :)14:20
NobodyCammornng rloo :) yes it is... :)14:20
* jroll has a two-day week this week \o/14:20
* NobodyCam throws things at jroll 14:20
NobodyCamlol14:20
jrollI'll be hanging out with my parents those days, don't get too jealous14:21
rloolucky jroll. which two days? will make sure we ask you lots of questions :-)14:21
jrollrloo: today and tomorrow :P14:21
NobodyCamhehehe... I may take friday off to pack14:21
jrolloh, we have a mtg today eh14:21
* BadCub plans to take Thursday and Friday off this week14:22
lucasagomesrloo, hi there!14:23
NobodyCammorning lucasagomes :)14:23
lucasagomesNobodyCam, BadCub morning14:23
rlooI think we need to help finalize the ironic sessions. or give feedback before it is finalized ;)14:23
BadCubmornin lucasagomes :)14:23
openstackgerritOpenStack Proposal Bot proposed openstack/ironic: Updated from global requirements  https://review.openstack.org/18120414:24
BadCubWe also need to finalize the proposed Ironic Dinner :)14:25
*** bnemec has joined #openstack-ironic14:25
ShrewsToday is "Throw Things at jroll Day"?  Finally!14:26
BadCubLOL14:26
NobodyCamlol morning Shrews14:26
BadCubmornin Shrews :)14:26
NobodyCam:)14:26
Shrewsmorning all14:27
* BadCub grabs a supply of soggy lamas to throw at jroll lol14:27
NobodyCamlol14:27
rloomorning Shrews. (I'm more tempted to throw things at Shrews...)14:28
* BadCub passes some soggy lamas to rloo to throw at Shrews14:28
Shrews:-D14:28
lucasagomesBadCub, oh when will be the dinner?14:29
NobodyCamare they Dalai Lama's lol14:29
BadCublucasagomes: I put a section up on the Planning pad. Trying to get input from everyone so I can make reservations for us14:29
jrollShrews: hahaha14:30
lucasagomesgotcha will take a look thanks!14:30
BadCubcoolness :)14:30
* NobodyCam gits mor kofi 14:30
devanandag'morning, all14:30
jrollBadCub: which pad, not seeing it14:30
rloojroll: https://etherpad.openstack.org/p/liberty-ironic-design-summit-ideas14:31
jrollhm, that's where I am14:31
* jroll needs ethergrep14:31
jrollmorning devananda \o14:31
BadCubjroll: starts at line 3514:31
rloomorning devananda14:31
BadCubmornin devananda :)14:32
* devananda makes the coffee happen14:32
rlooBadCub: are you looking for comments about date/time, and/or do you want name/count?14:33
lucasagomesdevananda, morning14:33
BadCubrloo: yes on all of those points14:33
NobodyCammorning devananda14:33
*** ukalifon has quit IRC14:33
*** mtanino_away has quit IRC14:35
*** mtanino has joined #openstack-ironic14:35
rlooBadCub: looks like 6:30pm might be better :-)14:36
BadCubrloo: Yeah. I think 630 would be better14:36
*** zz_jgrimm is now known as jgrimm14:37
lucasagomes+1 to 6:3014:37
jrollor even 7, might take a while to round up 20 people or whatever14:37
BadCubIndeed. I changed it to 630 on the pad14:38
*** ramineni has quit IRC14:38
devanandatuesday seems like it'd be rushed - folks will be missing the "supernatural evening" event14:39
BadCubdevananda: Wednesday is the Core party too.14:40
devanandaindeed14:40
jrollif I miss pop rocks ice cream, I will be upset.14:40
jrollI know mordred would agree.14:40
jroll:P14:40
devanandawe can't have that :)14:40
*** ukalifon1 has joined #openstack-ironic14:41
devanandathursday doesn't have any large parties announced yet, afaik14:41
jrollthursday wfm provided we party afterward14:41
jroll:D14:41
devanandahowever I'll be busy (TC dinner)14:41
BadCubIf we are gonna make a dinner work, we are probably gonna have to pick what party to be late to14:41
devanandaBadCub: indeed14:42
jrollseems like devananda should be there14:42
lucasagomesyeah14:42
BadCubbrb14:42
NobodyCambrb wouse workers here14:42
lucasagomeswhat if we do the dinner party on sunday?14:42
NobodyCamhouse even14:42
*** zz_natorious has joined #openstack-ironic14:42
jrolllucasagomes: you win the game14:42
lucasagomesI think everybody will be around sunday, no?14:42
*** zz_natorious is now known as natorious14:42
jrollI will14:42
BadCubnot sure everyone arrives Sunday14:42
lucasagomesright14:42
lucasagomesyeah... maybe we should ask if people will be around. I'll be there14:43
jrolllucasagomes: if we don't do it sunday, you and I are going to the pub14:43
devanandahm. I can bail on my sunday plans14:43
jrollI'll be settled in by noon14:43
lucasagomesjroll, +1 works for me14:43
devanandaI'll be there because board meeting all day sunday14:43
BadCubWe arrive Sunday early day14:43
BadCubTheJulia: What day do you and cinerama arrive?14:43
BadCubbrb14:44
mordredjroll: I didn't do it14:44
devanandawho CANT make a sunday dinner?14:44
jrollmordred: are you saying you would be ok with missing pop rocks ice cream balls?14:44
mordreddevananda: you know there's a board/tc dinner too, yeah?14:44
mordreddevananda: oh - wait - nevermind. *facepalm*14:44
mordreddevananda made the right choice ...14:45
*** anteaya has joined #openstack-ironic14:45
devanandamordred: yep. I know.14:45
openstackgerritLucas Alvares Gomes proposed openstack/ironic-specs: iPXE to use Swift Temporary URLs  https://review.openstack.org/18192014:46
jrolllucasagomes: ooooooooooo14:46
* rloo wonders what's wrong with the board/tc dinner but won't ask14:46
jrollJayF: ^^^ dat patch14:46
trownlucasagomes: shiny14:46
devanandarloo: nothing.14:46
rloodevananda: :D14:46
lucasagomesjroll, :-) take a look at the dependent one too when u have some time14:46
lucasagomeswants to try to have something working for L-114:46
jrolllucasagomes: yep, will do14:47
trownI was looking at swift temp urls for AHC as well14:47
* jroll would like to run his env this way14:47
lucasagomestrown, cool, yeah. We already have all the code needed to generate swift temp urls in the Ironic tree and all14:47
trownoh, nice14:47
lucasagomestrown, take a look at the swift.py module :-)14:47
lucasagomestrown, https://github.com/openstack/ironic/blob/0d9be37daebdb994e0d907952b58d0bc406f28f9/ironic/common/swift.py14:48
trownlucasagomes: awesome, thanks!14:48
NobodyCammorning trown and mordred14:50
trownmorning NobodyCam14:50
*** hemnafk is now known as hemna14:51
TheJuliaBadCub: I arrive sunday14:53
TheJuliaBadCub: cinerama arrives monday afaik14:53
BadCubTheJulia: okay. I think I remember cinerama telling me that too14:53
*** rameshg87 has joined #openstack-ironic14:53
BadCubwhat is going on Monday night? Is that an option?14:54
rameshg87lucasagomes: hi14:54
NobodyCammorning rameshg8714:54
rameshg87morning NobodyCam14:54
NobodyCam:)14:54
rameshg87lucasagomes: did you have a chance to test https://review.openstack.org//#/c/167929/ ?14:54
lucasagomesrameshg87, hi there, yes! I tested with fedora seems good14:56
rameshg87oh okay..thanks :)14:57
rameshg87i will ping tripleo folks for reviews14:57
lucasagomesack14:58
* lucasagomes jumps in a call14:58
devanandaBadCub: monday has the "booth crawl" until 7:30, but I haven't seen any event invites for after that yet15:02
BadCubdevananda: hmmm... Then maybe a later dinner on Monday would be the ticket15:03
devanandaBadCub: corporate parties notwithstanding15:03
* BadCub grumbles15:04
jrolldamn corporations15:04
* TheJulia wonders if we need a meter that translates badcub's grumles... something along the lines of a geiger counter15:05
rlooexcept HP? where's the RH party? :)15:05
rlooTheJulia: I thought it was part of the job description for BadCub15:05
BadCubTheJulia: LOL15:05
BadCubrloo: why yes, it is! LOL15:06
TheJuliarloo: grumbles, yes... but I think we need some sort of grumbles per-hour measureing system15:06
* BadCub grumbles about metering systems15:08
BadCubI don;t grumble that often lol15:08
*** saripurigopi has quit IRC15:08
devanandaas far as thursday sessions and big ticket items we need to discuss -- I'd like to put "nova's virt driver" on the agenda, and since nova hasn't made room for it, we'll have to host it15:08
jrolldevananda: what about it?15:09
devanandajroll: less of a "i want to add this feature" and more of a "it's an important part of our project that we dont have an HA story for, and haven't been tracking bugs on"15:12
jrolldevananda: oh, that15:12
jrollyou only said "nova's virt driver"15:12
jroll:P15:12
devanandaoh, i mean nova's ironic virt driver :p15:12
devanandais there any other one?15:13
jrollstill15:13
jrollthat doesn't say anything about what you want to talk about15:13
jrollmakes sense now :P15:13
devanandasomehow, i think if we just get together with that topic, everyone will have something to say ...15:14
jrollindeed :)15:14
jrollI still want to make the compute lock manager distributed15:14
rloodevananda: that reminds me. did you see the email thread from nova about it (wanting an ironic person to liason with). i was going to put it on the agenda for the meeting.15:14
jrollrloo: did you see where jlvillal volunteered? :)15:14
devanandamy long term hope is that we'll be able to bypass the nova compute process entirely, and plug in ironic at the nova conductor or scheduler layers15:15
jlvillalrloo: That is true, I did volunteer.15:15
devanandanova folks seems to generally agree, but so far, have no idea when or how that'll actually happen15:15
rloojlvillal: yeah. so i thought the agenda item would be 'thank you' or to address any concerns or whatever.15:15
NobodyCambrb15:15
jrolldevananda: that seems.... wrong15:15
devanandarloo: yes. and thank you, jlvillal :)15:17
devanandalike any liaison role, it'll be an ongoing need that we have. also - we should wiki'fy the various liaison roles15:17
jlvillaldevananda: You're welcome.  I will attend the next Nova meeting.  I will also try to stay awake and attend the Ironic meeting tonight :)15:18
* jlvillal is usually trying to go to sleep at 10pm PDT....15:18
jrollironic mtg is in the AM today :)15:18
jlvillalWoo hoo! :)15:18
devanandajroll: no, it's tonight15:18
jlvillalOh noooes!!15:18
jlvillal;)15:18
* rloo wonders if it is worth evaluating the effectiveness of the Tues meetings15:19
jrolloh?15:19
jrollwtf15:19
jrolloh wow, I'm dumb15:19
jrollanother 10am mtg was on my calendar and I assumed15:19
jrollrloo: ++15:19
devanandarloo: that is something I planned to bring up at the summit15:19
jrolldevananda: list seems better for that, summit will be biased toward US people15:19
rloodevananda: thx!15:19
* devananda adds to 'pad15:20
rloooh, jroll has a good point.15:20
*** mtanino has quit IRC15:20
devanandatrue15:20
* devananda emails15:22
rameshg87jlvillal: hi15:23
jlvillalrameshg87: Hi15:23
rameshg87jlvillal: wondering if it's time to add some kind of a default arguments decorator for https://review.openstack.org/#/c/177700/315:23
jlvillalrameshg87: Working on the patch.  Thanks for the comments!15:23
rameshg87jlvillal: oh okay :)15:23
rameshg87jlvillal: wdyt ?15:23
jlvillalrameshg87: I was going to add you as a 'co-author' :)15:24
rameshg87jlvillal: something like what we did for wsexpose15:24
rameshg87jlvillal: oh but i didn't do that much ;-)15:24
jlvillalrameshg87: I respectfully disagree :P15:24
rameshg87:)15:24
rameshg87jlvillal: since we have autospec=True and spec_set=True for most of the methods15:25
jlvillalrameshg87: I'm not sure on the decorator.  Probably a topic for a meeting I would think.15:25
jlvillalSince that is a more invasive change.15:25
rameshg87yeah .. okay15:25
jlvillalrameshg87: On the patch set you mentioned.  For spec_set=True.  I was slightly confused.15:26
jlvillalrameshg87: I thought you were talking about my patch.15:26
rameshg87jlvillal: ah no15:27
rameshg87jlvillal: i was talking about naohirot's patch15:27
rameshg87jlvillal: since he is building some work on top of yours, i thought of checking with you15:27
jlvillalrameshg87: Still think good idea for meeting, but are you thinking if 'autospec=True' then also do 'spec_set=True'?15:27
* BadCub heads off to morning call15:27
rameshg87jlvillal: ah no15:27
rameshg87jlvillal: i meant a decorator in tests/utils.py or somewhere15:27
rameshg87jlvillal: which calls mock.patch.object and defaults arguments autospec=True and spec_set=True if it isn't passed - since we recommend that for all the tests15:28
nagyzah too bad I won't be at this summit just for Liberty15:28
jlvillalrameshg87: So instead of @mock.patch.object() do  @utils.patch.object()?15:28
rameshg87jlvillal: yeah something like that15:28
*** kan_ has quit IRC15:29
jlvillalrameshg87: That seems reasonable.  As it would be explicit.  I was thinking you meant like make it so that @mock.patch.object() calls would get munged to do that.  I was scared of that, as people might not realize it was happening.15:29
jlvillalrameshg87: But an explicit call would be reasonable to me.15:29
rameshg87jlvillal: yeah i meant an explicit call15:30
jlvillalrameshg87: I'm +1 for that :)15:30
rameshg87okay, just posting comment in naohirot's review15:30
jlvillalrameshg87: Thanks15:30
*** mtanino has joined #openstack-ironic15:31
nagyzso after adding a port and NOT restarting ironic or nova-compute I still see the neutron error15:34
*** saripurigopi has joined #openstack-ironic15:35
nagyzwhich persists even after restart15:35
*** tiagogomes has left #openstack-ironic15:36
openstackgerritJohn L. Villalovos proposed openstack/ironic: ironic/tests/drivers: Add autospec=True and spec_set=  https://review.openstack.org/17704115:36
jlvillalrameshg87: I implemented your suggestions ^^^^^15:36
jlvillalrameshg87: Thanks for the feedback!15:36
NobodyCammorning jlvillal :)15:36
jlvillalNobodyCam: Good morning15:36
NobodyCam:)15:36
jlvillalrameshg87: I will update the commit message to remove autospec=False line15:37
openstackgerritJohn L. Villalovos proposed openstack/ironic: ironic/tests/drivers: Add autospec=True and spec_set=  https://review.openstack.org/17704115:37
rameshg87jlvillal: oh, that's the last thing with autospec=False in it :-)15:37
jlvillalNobodyCam: And congratulations! :)  I see your patch is in the merge queue.15:37
NobodyCamoh lol ... W00t !!!!15:38
NobodyCamhehehe15:38
jlvillalrameshg87: Yep, the last one15:38
rameshg87jlvillal: but i think i disagree there - https://review.openstack.org/#/c/177041/ . my name shouldn't have been mentioned for just pointing out somethings in test case15:43
rameshg87jlvillal: there was no way for me to check if what i suggested worked, so i just took your patch, and tried those15:44
openstackgerritMerged openstack/ironic: Check temp dir is usable for ipmitool driver  https://review.openstack.org/16038315:44
rameshg87and that's why i posted the snippets15:44
jlvillalrameshg87: I won't force you to be on the patch.  But you did provide a lot of good info to fix those last few issues.  And even provided diffs15:44
jlvillalrameshg87: So I personally think you deserve credit.15:44
jlvillalrameshg87: But if you would prefer to not be on it, then I can remove your name.15:45
nagyzso I'm running into a strange situation.15:45
nagyzI've deleted a node which autodeletedit's network port (I think, at least ironic port-list doesn't show it). now when I want to provision a new instance (for which I added a new port) it wants to delete the old neutron port15:45
*** mgoddard has quit IRC15:46
nagyzwhich doesn't exists so it gets a 404 and the provisioning dies.15:46
rameshg87jlvillal: i think you deserve the credit for that .. and as per rules can't +2 a patch that my name is put as co-author (and i surely want to +2 it) ;-)15:46
jlvillalrameshg87: Ah, good point :)15:46
openstackgerritMerged openstack/ironic: Fix typos in Ironic docs  https://review.openstack.org/18181315:46
nagyzI checked the db and the port is indeed deleted from there.. yet ironic wants to ask neutron to delete it.15:46
nagyzideas?15:46
openstackgerritJohn L. Villalovos proposed openstack/ironic: ironic/tests/drivers: Add autospec=True and spec_set=  https://review.openstack.org/17704115:47
jlvillalrameshg87: ^^^15:47
*** mgoddard has joined #openstack-ironic15:47
*** rwsu has joined #openstack-ironic15:48
rameshg87nagyz: are you sure if that is the reason why provisioning fails ?15:48
*** mgoddard has quit IRC15:48
rameshg87jlvillal: ack15:48
jlvillalrameshg87: Thanks :)15:48
*** mgoddard has joined #openstack-ironic15:48
openstackgerritOpenStack Proposal Bot proposed openstack/ironic: Updated from global requirements  https://review.openstack.org/18120415:49
*** mgoddard has joined #openstack-ironic15:49
nagyzrameshg87, looking at the error message you're right, it says no valid host is found. weird... let me debug it further.15:49
rameshg87nagyz: yeah i expect some other error should be the cause (most probably an exception in nova-compute log right above it should be there)15:50
nagyzso the scheduler log says "host blahblah failed", meaning I guess nova-compute's neutron error bubbles up?15:50
nagyzlet's see15:50
nagyzah.15:50
nagyz Error: Node cd1091f5-9129-46d2-988f-2e7959cf01b7 is configured to use the agent_ipmitool driver which currently does not support deploying partition images. (HTTP 500)15:50
rameshg87nagyz: ah no. a possible thing - so it's like nova compute tried to spawn and failed15:51
rameshg87yeah15:51
lucasagomesnagyz, for partition images you can use any of the pxe_* drivers15:51
lucasagomeseven with the agent ramdisk15:51
*** mgoddard1 has joined #openstack-ironic15:51
lucasagomess/agent/IPA15:52
*** ifarkas has quit IRC15:52
openstackgerritLucas Alvares Gomes proposed openstack/ironic: Wake-On-Lan Power interface  https://review.openstack.org/17907815:52
*** mgoddard has quit IRC15:54
* lucasagomes brb15:54
*** mgoddard1 has quit IRC15:55
*** mgoddard has joined #openstack-ironic15:56
*** e0ne has joined #openstack-ironic15:57
*** e0ne is now known as e0ne_15:58
*** e0ne_ is now known as e0ne15:58
* rameshg87 leaves for the day 16:10
rameshg87good night ironic16:10
*** rameshg87 has quit IRC16:10
devanandag'night, ramesh16:10
*** subscope has quit IRC16:12
nagyzlucasagomes, well I've built a vm image using dib16:15
nagyzwhich doesn't seem to work16:15
TheJulianagyz: I thought you were going to switch the node over to agent_ipmitool16:16
TheJuliaerr16:16
nagyzTheJulia, it's already using that16:16
TheJulianvmd16:16
* TheJulia apparently can't read16:16
nagyzbut the image is still partitioned somehow16:16
nagyz:-)16:16
TheJuliawell16:16
*** trown is now known as trown|lunch16:16
nagyzI just need to find a good image, or to build a new image I guess16:17
TheJuliaIt is interpretting it as a partition image, which means no partition table, just the raw contents of a volume16:17
*** jistr has quit IRC16:17
nagyzright but shouldn't the vm image build put in the partition table?16:17
TheJuliait should, so it should be in the image that it is attempting to deploy16:17
TheJuliaIt might be worthwhile to compare what its trying to use on disk and verify that though16:18
*** jcoufal has quit IRC16:18
nagyzyeah I'll do that from home. I have to catch my bus home so I'll be back later. :)16:18
nagyzthanks very much for your help!16:18
TheJuliaok, no problem, have a safe ride16:18
*** jcoufal has joined #openstack-ironic16:18
nagyzswiss buses are very reliable. :-)16:18
*** jcoufal has quit IRC16:23
*** jcoufal has joined #openstack-ironic16:23
openstackgerritZhenguo Niu proposed openstack/ironic: When boot option is not persisted, set boot on next power on  https://review.openstack.org/17764216:31
*** Sukhdev has joined #openstack-ironic16:35
*** absubram has joined #openstack-ironic16:40
*** Marga_ has quit IRC16:42
*** derekh_ has joined #openstack-ironic16:44
*** derekh has quit IRC16:47
*** derekh_ has quit IRC16:51
*** mgoddard has quit IRC16:56
lucasagomesnagyz, ? vm image/16:56
*** mgoddard has joined #openstack-ironic16:56
lucasagomesright so you can deploy full disk, just do not specify any image kernel or ramdisk in glance16:56
lucasagomesin the image metadata16:56
*** Nisha has joined #openstack-ironic16:57
*** mgoddard has quit IRC16:58
*** davideagnello has joined #openstack-ironic16:58
lucasagomesmeeting?17:00
*** davideagnello has quit IRC17:00
NobodyCamlucasagomes: meeting is tonight17:00
openstackgerritJohn L. Villalovos proposed openstack/python-ironicclient: WIP: Refactor resource_fields  https://review.openstack.org/18198517:00
*** davideagnello has joined #openstack-ironic17:00
lucasagomesNobodyCam, ohhhhh I confused, thought it was 1700 UTC on Monday17:01
*** mgoddard has joined #openstack-ironic17:01
lucasagomesit's 05 UTC the next :-(17:01
lucasagomesit's very very early here17:01
NobodyCam:-p17:01
NobodyCamya and very very late here17:02
NobodyCam:-p17:02
lucasagomesyeah17:03
jlvillallucasagomes: This might of interest to you in the future: https://review.openstack.org/18198517:04
* lucasagomes looks17:05
jlvillallucasagomes: I think you were the one who did the resource_fields.py work initially.17:05
jlvillallucasagomes: Still a lot to do though.17:05
devanandaanyone opposed to cancelling tonght's irc meeting, since we're all going to be craming and talking about the summit anyway?17:05
lucasagomesjlvillal, ohhhh that's good... kan has a patch touching those17:05
*** mgoddard1 has joined #openstack-ironic17:06
jlvillallucasagomes: Yes, I was inspired by Clif's comments. :)17:06
*** mgoddard has quit IRC17:06
lucasagomesdevananda, I'm ok if you guys have that meeting, but I won't be able to make it because it's a bit early here17:06
NobodyCamdevananda: I would okay with that.. there are some items on the agenda for the meeting17:06
*** Marga_ has joined #openstack-ironic17:06
lucasagomesdevananda, but that said, I'd like if we could call some attention to the bugs in Nova17:07
lucasagomesthere are 20+ bugs tagged with Ironic, most are scheduling issues or some small fixes17:07
lucasagomesbut would good to have more eyes looking at those bugs17:07
lucasagomesjlvillal, gotcha!17:07
lucasagomesawesome!17:07
lucasagomesjlvillal, perhaps you could rebase on top of Kan's patch?17:09
lucasagomesso that they don't conflict?17:09
jlvillallucasagomes: Good idea.17:09
jlvillalThanks17:09
NobodyCamjlvillal: are you up for the nova liason position17:10
*** saripurigopi has quit IRC17:11
jlvillalNobodyCam: I believe so.  Help would be appreciated on what you would like me to communicate to the Nova team from Ironic.17:11
NobodyCamofc :)17:11
*** ramineni has joined #openstack-ironic17:11
NobodyCamand awesome :)17:11
jlvillalNobodyCam: I will be attending their meeting this week.17:11
NobodyCamw00t!17:11
*** mgoddard1 has quit IRC17:12
*** mgoddard has joined #openstack-ironic17:13
*** e0ne has quit IRC17:13
lucasagomesfolks since there's no meeting I will call it a day17:14
lucasagomeshave a great night everyone!17:14
NobodyCamhave a great night lucasagomes17:14
*** lucasagomes is now known as lucas-dinner17:14
NobodyCam:)17:14
jlvillallucasagomes: Good night.  Meeting is in 12 hours17:14
lucas-dinnerjlvillal, yeah I will probably be sleeping17:14
jlvillallucas-dinner: I think.  If it doesn't get cancelled.17:14
lucas-dinneryeah I mean, there's no meeting now :-) sorry17:15
jlvillallucas-dinner: Good plan :)17:15
lucas-dinnercause I confused the times17:15
*** jcoufal has quit IRC17:15
lucas-dinnerI thought it was 17:00 UTC today17:15
lucas-dinnerI will read the logs of the meeting at 5:00 UTC later17:15
*** Marga_ has quit IRC17:16
*** trown|lunch is now known as trown17:17
openstackgerritMerged openstack/ironic: Fix and enhance "Exercising the Services Locally" docs  https://review.openstack.org/18016317:21
NobodyCamoh /me like the size meter on the reviews page17:22
*** krtaylor has quit IRC17:24
*** Marga_ has joined #openstack-ironic17:25
rlooNobodyCam: I haven't quite figured out the size meter thing. Do I just look at green? :-)17:31
*** vipul has quit IRC17:31
NobodyCam????17:35
*** ukalifon1 has quit IRC17:35
*** krtaylor has joined #openstack-ironic17:36
rlooNobodyCam: the size meter on the reviews page. Do you see green for the (I think) smaller patches?17:36
-openstackstatus- NOTICE: We have discovered post-upgrade issues with Gerrit affecting nova (and potentially other projects). Some changes will not appear and some actions, such as queries, may return an error. We are continuing to investigate.17:38
*** ChanServ changes topic to "We have discovered post-upgrade issues with Gerrit affecting nova (and potentially other projects). Some changes will not appear and some actions, such as queries, may return an error. We are continuing to investigate."17:38
* NobodyCam pops out for quick smoke17:44
*** fresh has joined #openstack-ironic17:46
openstackgerritJohn L. Villalovos proposed openstack/python-ironicclient: WIP: Refactor resource_fields  https://review.openstack.org/18198517:48
*** edcragg has quit IRC17:51
*** gridinv_ has joined #openstack-ironic17:53
*** zer0c00l has quit IRC17:53
*** zer0c00l has joined #openstack-ironic17:54
*** achanda has joined #openstack-ironic17:54
*** gridinv_ has quit IRC17:57
*** pelix has quit IRC17:58
jlvillalrloo: Yes, green is smaller from what I can tell.18:00
jlvillalrloo: Seems to go green, yellow, red depending on how big the patch is.18:01
rloojlvillal: I have some concerns about that but maybe I'm pessimistic. Eg, who wants to spend lots of time reviewing. So "I'm only going to review the green patches"...18:01
*** dprince has quit IRC18:03
*** ijw has joined #openstack-ironic18:05
*** ijw has quit IRC18:05
*** ijw has joined #openstack-ironic18:05
jlvillalrloo: Hey that was my idea to only review green patches.  I thought I patented it ;)18:08
jlvillalrloo: Though maybe it can be used to blast through a whole bunch of small patches to review?  I think it is probably going to hurt the big patches some.  I think you are right on that.18:09
rloojlvillal: they could at least have used green for BIG and red for SMALL :)18:11
jlvillal:)18:11
rlooor no colour to be fair to those that are colour-blind.18:12
rlooor maybe the stackcounterthingy can give more weight to bigger patches. Hey, i like that idea.18:12
jlvillalrloo: That would be good!18:13
jlvillalNobodyCam: devananda: What was decided on tonight's meeting?  Is it a go or no-go?18:13
BadCubokay folks. I have changed up the dinner section a bit and Proposed Monday night at 630. Just need folks to +1/-1 it on pad and indicate venue preference18:15
*** achanda has quit IRC18:18
NobodyCamjlvillal: I have heard several other project have canceled their meeting this week18:22
*** hemna_ has joined #openstack-ironic18:24
*** bnemec has quit IRC18:25
*** lsmola_ has quit IRC18:25
*** Guest64787 has quit IRC18:25
*** smallbig has quit IRC18:25
*** morganfainberg has quit IRC18:25
*** clett has quit IRC18:25
*** mrda has quit IRC18:25
*** rsFF has quit IRC18:25
BadCubtwo new restaurants added to the list too. They have availability for Mon @73018:26
*** dmellado has quit IRC18:26
*** hemna has quit IRC18:26
*** slagle has quit IRC18:26
*** mrmoje has quit IRC18:26
*** Shrews has quit IRC18:26
*** svasheka has quit IRC18:26
*** bkero has quit IRC18:26
*** mrmoje has joined #openstack-ironic18:26
*** Nisha has quit IRC18:26
*** slagle has joined #openstack-ironic18:27
*** bnemec has joined #openstack-ironic18:27
*** lsmola_ has joined #openstack-ironic18:27
*** smallbig has joined #openstack-ironic18:27
*** morganfainberg has joined #openstack-ironic18:27
*** clett has joined #openstack-ironic18:27
*** mrda has joined #openstack-ironic18:27
*** rsFF has joined #openstack-ironic18:27
*** dmellado has joined #openstack-ironic18:27
*** svasheka has joined #openstack-ironic18:27
*** slagle has quit IRC18:27
*** slagle has joined #openstack-ironic18:27
*** Shrews has joined #openstack-ironic18:28
*** athomas has quit IRC18:28
*** dan| has joined #openstack-ironic18:29
*** ramineni has quit IRC18:29
*** bkero has joined #openstack-ironic18:30
*** bkero has quit IRC18:30
*** bkero has joined #openstack-ironic18:30
jlvillalNobodyCam: Okay.  Just trying to figure out if I should try to stay awake tonight for the meeting :)18:32
pshigeBadCub: thank you!18:32
BadCubpshige: you're welcome! Thank you too! :)18:33
jlvillalrloo: On your comments on the sort field patch.  You might want to look at my follow-on patch: https://review.openstack.org/#/c/181985/   I think it will address some (but not all) of your concerns.18:33
jlvillalrloo: But still WIP18:33
rloojlvillal: I think the handling of chassis_uuid should be done in the same patch18:34
rloojlvillal: the handling of that is diff from dealing with duplication I think18:34
*** e0ne has joined #openstack-ironic18:35
jlvillalrloo: Okay.18:35
*** Sukhdev has quit IRC18:35
rloojlvillal: and the more i think about it, the more i'm not convinced we should even land that patch. unless we think it is wrong to sort on those keys that are being removed.18:35
jlvillalrloo: That I'm not sure about.  Either for or against.18:36
openstackgerritJohn L. Villalovos proposed openstack/python-ironicclient: WIP: Refactor resource_fields  https://review.openstack.org/18198518:36
rloojlvillal: i'll look at your patch after we've dealt with this patch :-)18:36
openstackgerritJohn L. Villalovos proposed openstack/python-ironicclient: WIP: Refactor resource_fields  https://review.openstack.org/18198518:37
jrollrloo: jlvillal: I mean, you can't reasonably sort json18:37
nagyzback.18:37
jrolls/reasonably/meaningfully/18:37
nagyzis there a way to get a node out of a stuck "cleaning" state?18:37
jlvillalrloo: Good plan :)18:37
nagyzI can't seem to be able to change it to anything.18:37
rloojroll: yeah, so either we allow or we don't allow. and if we don't allow, we should do it in client and REST API?18:37
*** zhenguo has quit IRC18:37
jrollrloo: imo we should drop it from both18:37
rloojroll: ok, i'm fine with that. just don't want to do a half-a** job and have to revisit it later.18:38
jrollindeed :)18:38
devanandajlvillal: nothing decided - i just asked, and like 2 folks said 'well i wont be there anyway...'18:41
jlvillaldevananda: Thanks.18:41
devanandajlvillal: I'll show up, but if there's not many folks there, I'll just close it right away18:41
jlvillaldevananda: Works for me.  I will plan on being there.18:42
* BadCub will be snoring at 10PM tonight18:42
nagyzso I've got a node stuck in "cleaning"; I tried to delete it, I cannot as it's powered on. I tried to power it off, I can't as it's cleaning...18:42
nagyzis there a --force I'm missing somewhere?18:42
nagyz:/18:42
devanandaspeaking of which - i've put up a poll about our meeting times - http://goo.gl/forms/nvbWdOZMFY18:42
nagyz(cleaning failed since it can't even boot - it was stuck at "select a boot device to boot from" before I did a nova delete)18:43
devanandasee the ML for more, or don't, but please provide feedback18:43
pshigenow I remember some Ironic t-shirts. what happened after that? :)18:43
BadCubdevananda: done and done18:44
devanandaty18:45
Shrewsdevananda: fyi, the second question doesn't leave room for "i attended half, was able to attend before Kilo, and found it productive"18:45
devanandaShrews: isn't that the first choice?18:46
devanandaoh wiat, no18:47
Shrewsdevananda: nope18:47
BadCubdarn.. does that mean we have to go back and take the test again? LOL18:47
devanandaShrews: so you mean an option for "I attended half as many meetings as I used to, and found that personally helpful" ?18:47
devanandaBadCub: no - i can edit the form18:47
Shrewsdevananda: yeah18:48
devanandaShrews: huh. I hadn't considered that. it's almost like 1hr/wk is too much meeting?18:48
devanandaShrews: there is always the "other" field :)18:48
Shrewsdevananda: well, i could not attend the Tuesday meeting b/c of the time, but I still found attending Monday helpful (but 1/2 as helpful as attending both, if that makes sense)18:49
Shrewsdevananda: oh! i scanned over the Other part18:49
*** dan| has quit IRC18:49
devanandaShrews: "half as helpful" is what i meant by #4, but I see the wording isn't good18:51
* devananda changes it18:51
*** harlowja has quit IRC18:52
*** harlowja has joined #openstack-ironic18:52
*** dan_ has joined #openstack-ironic18:54
devanandaShrews: updated18:55
*** dan_ is now known as Guest2558018:55
devanandalmk if that captures it18:55
Shrewsdevananda: yeah, but already responded before the change18:55
devanandanp18:56
*** mrmoje has quit IRC18:57
*** ijw has quit IRC18:58
*** ijw has joined #openstack-ironic18:59
*** mrmoje has joined #openstack-ironic18:59
*** ijw has quit IRC19:03
NobodyCamdid we ever find the documation (I thought) we had on devstack and real hardware?19:05
*** ukalifon1 has joined #openstack-ironic19:05
*** ukalifon1 has quit IRC19:06
nagyzI can't believe I have to fight python stacktraces and do manual mysql updates...19:10
*** Marga_ has quit IRC19:11
*** dguerri is now known as _dguerri19:12
*** e0ne has quit IRC19:22
nagyzinterestingly enough I see a tftp request (using iPXE) for "http://myserver/boot.ipxe"19:28
nagyzI thought iPXE is going to fetch that url however it's goingto be loaded first? :)19:28
nagyzthe request is happening before iPXE is loaded19:28
*** subscope has joined #openstack-ironic19:29
rloohi NobodyCam or anyone that knows the answer. If you use a package that isn't in global requirements, it needs to be there first before we add it to python-ironicclient, right? https://review.openstack.org/#/c/173674/9/requirements.txt19:29
devanandarloo: yes19:30
rloodevananda: thx. by the way, for 'features' that refer to a spec (that didn't discuss the feature) do we need a bug (or updated spec)? that ^^ is adding a cache for the client.19:31
openstackgerritMerged openstack/ironic: ironic/tests/drivers: Add autospec=True and spec_set=  https://review.openstack.org/17704119:36
devanandarloo: good question. if the code is supposedly implementing a spec, but adding something NOT in that spec, it raises the question, was the spec incomplete?19:37
devanandarloo: also, I think it bears further investigatoin as to whether other openstack projects are doing any similar client-side caching19:37
devanandamordred may know -- I seem to recall shade doing something with client caching19:38
mordredyes19:38
mordred2 things - we are using XDG values - so appdirs will give you good locations19:38
*** _dguerri is now known as dguerri19:38
mordredwe're using dogpile.cache for teh caching itself19:38
rloodevananda: two good questions :-)19:38
rloomordred: i notice that appdirs isn't in global-requirements (yet?)19:39
mordredand os-client-config contains support for configuring such a cache19:39
devanandacool. mrda's proposal already includes references to XDG and uses appdirs19:39
mordredrloo: I didn't  hear of appdirs before just now19:39
rloomordred: ok good. mrda is on the right path! :)19:39
mordredwell, getting there ....19:40
mordredI'm not crazy about the custom written caching code19:40
rloodevananda: I don't know if we want specs to be revised (my preference) or a bug opened? it needs *something*...19:40
mordredand I strongly suggest looking at using dogpile.cache- I believe keystoneclient is using it as well now19:41
rloomordred: yeah, would be nice to have a library with that caching code19:41
rloomordred: ahh good suggestion.19:41
*** Marga_ has joined #openstack-ironic19:42
mordredI just -1'd the review with taht suggestion in it19:43
rloomordred: thx!19:43
devanandajust read dogpile's docs -- looks reasonable19:44
mordreddevananda: it has the benefit of being able to use persistent disk caches, memory caches, or even shared thigns like redis/memcached. now, obviously that's crazypants for _this_ ... but still19:45
devanandayup19:45
mordredalso: https://github.com/openstack/os-client-config#cache-settings19:46
*** mrmoje has quit IRC19:46
devanandaand handles locking around those, if necessary, which might actually be helpful when python-ironicclient is called from, say, some multithreaded application19:46
mordredyah19:46
*** Marga_ has quit IRC19:46
devanandamordred: any relation between ^ and openstackclient?19:47
mordredyup19:47
mordredopenstackclient consumes os-client-config19:47
devanandahttps://review.openstack.org/#/c/171672/1119:47
mordredneat19:47
*** mrmoje has joined #openstack-ironic19:47
devanandaso yah, what ever API versoin caching mrda is implementing should be done so it works for the osc plugins here, too19:48
mordredso - perhaps we could add something to the openstackclient interface that would allow for getting at the cache setting in the openstackclient o-c-c object19:48
mordredand then if you're in your own code, making your own o-c-c object and grabbing the cache setting from it19:49
* mordred waves hands, but dreams of a glorious future19:49
devananda++19:49
devanandafor now, let's at least use the same caching libs :)19:49
* mordred thinks he may add appdata to o-c-c19:49
mordredso that rather than having XDG things baked in, it can refer to appdata and work as expected on osx and windows too19:50
mordredI'm assuming mrda will be submitting a patch to global requirements ... unless someone wants me to19:50
devanandai'm sure mrda will19:50
devanandahe's probably just waking up right now :)19:51
*** fresh has quit IRC19:55
rloodevananda: wrt features in "completed" specs or whatever. My preference is to have everything related to a feature, documented together. but our specs are release-based. So I'm concerned about having links in patches in release X+1 to specs that were 'completed' in release X. But I don't think it makes sense not to link a patch to something (bug/BP) unless it is internal code being cleaned up.19:57
rloodevananda: or maybe these things fall into the 'small spec' category. I don't know if we've had any guinea pigs there yet?19:58
mordredmrda: hey - I'm actually about to hack on appdata support for os-client-config - so I'll send up the reqs patch20:03
mordreds/appdata/appdirs/20:03
*** kkoski has quit IRC20:08
mordredhttps://review.openstack.org/18206120:11
*** kkoski has joined #openstack-ironic20:13
devanandarloo: oh hmm. so we haven't really formalized this yet, but I think that a spec which is only partially completed in one cycle should a) be updated to indicate that, and b) be re-proposed and fast tracked for the next cycle20:17
*** drwho has joined #openstack-ironic20:17
devanandarloo: I agree re: non-cleanup-patches need a link to some artifact (bug, bp, etc)20:18
rloodevananda: is that something to discuss at the summit or email list? cuz folks might wonder what 'partially completed' means. wrt patches getting merged, features "turned on". eg, should cleaning NOT have been turned on?20:19
rloodevananda: and we only got some of the api microversioning working. guess it depends on how much the spec takes on/describes, vs adding stuff in a new spec.20:20
devanandayah...20:21
devanandaso both the state machine and microversion specs were marked "informational"20:21
devanandawe've never taken on tasks where the design spanned >1 cycle of implementation before20:22
devanandawe didn't plan on doing it that way, but we did. for the state machine, we've got follow-on specs that implement one step (eg, zapping) and reference the 'informational' spec20:23
rloodevananda: oh, right. hmm. and we have separate specs for state-machine-related tasks (except for all the plumbing you put in, i think that was tied to the orig spec)20:23
devanandaright20:23
devanandathe "design" was done, along with the internal plumbing, but each new state got its own spec20:23
devanandadoes that work for microversions too?20:23
* devananda thinks20:24
rlooand even the microversion spec might not be complete (we borrowed stuff from nova I think but not sure we spelled it out in our spec). need to look at that spec again. sec.20:24
nagyzwhen using iPXE, shouldn't it do a PXE -> iPXE -> ipxe.boot conf boot sequence? right now I'm seeing PXE going directly to fetch ipxe.boot by actually passing on the http url as is to tftp...20:25
nagyzwith all openstack projects I feel like more time is spent adding new features than improving error handling and documentation :)20:25
rloodevananda: our microversion spec does mention that the client should cache, but doesn't go into any more detail. I think it makes sense to have a (small) spec for the caching -- 'using XDG. dogpile.cache...'.20:29
devanandarloo: ++20:29
*** Marga_ has joined #openstack-ironic20:29
rloodevananda: so 'informational' could either 1. ONLY be informational; or 2. include plumbing patches + informational for further work.20:29
devanandarloo: i think in both cases, we landed some plumbing at the same time20:30
rloodevananda: or landed plumbing before the spec started :-)20:30
devanandarloo: touche :)20:30
rloodevananda: but that was an exception, not the rule. (I hope!)20:30
rloook, i think i have a good idea of informational specs. Dealing with 'partially complete' specs though might be a bit more delicate so I'll leave that for you to dictate ;)20:32
*** pcrews has quit IRC20:33
*** mtanino has quit IRC20:34
*** jjohnson2 has joined #openstack-ironic20:36
openstackgerritJarrod Johnson proposed stackforge/pyghmi: Flesh out and rework some sensor descriptions  https://review.openstack.org/18207420:39
openstackgerritRuby Loo proposed openstack/python-ironicclient: Revert fix that issues Unauthorized exception  https://review.openstack.org/18207520:39
*** openstackgerrit_ has joined #openstack-ironic20:44
*** achanda has joined #openstack-ironic20:48
*** Sukhdev has joined #openstack-ironic20:52
*** Marga_ has quit IRC20:58
*** Marga_ has joined #openstack-ironic20:59
nagyzso any ideas why it's not fetching the iPXE chainloaders instead of the ipxe config file?21:00
lucas-dinnernagyz, dhcp configuration? iPXE will only be chainloded if the request comes from a non iPXE client (see dhcp option 175)21:03
lucas-dinnernagyz, r u using neutron as ur dhcp provider?21:03
lucas-dinnerif so, did you set the ipxe tag in the configuration file? see http://docs.openstack.org/developer/ironic/deploy/install-guide.html#neutron-configuration21:04
*** trown is now known as trown|outttypeww21:04
*** Marga_ has quit IRC21:10
*** Marga_ has joined #openstack-ironic21:10
*** mtanino has joined #openstack-ironic21:10
*** jjohnson2 has quit IRC21:13
*** pcrews has joined #openstack-ironic21:14
openstackgerritMerged stackforge/pyghmi: Break out of FRU read if zero data returned  https://review.openstack.org/18101921:14
*** Marga_ has quit IRC21:17
*** Marga_ has joined #openstack-ironic21:18
*** absubram has quit IRC21:23
*** spandhe has joined #openstack-ironic21:32
*** kkoski has quit IRC21:38
*** achanda has quit IRC21:43
mrdaMorning Ironic21:46
nagyzlucas-dinner, yes, I'm using neutron, and it's set up21:47
lucas-dinnernagyz, right, so it's trying to fetch boot.ipxe before chainloading unionly.kpxe?21:47
nagyzyes21:48
lucas-dinnernagyz, but, r you sure the nic doesn't have ipxe by default?21:48
nagyzyes, it's a normal intel NIC.21:48
mrdathanks mordred21:48
lucas-dinnervery odd21:48
* lucas-dinner thinks21:49
nagyzand what's even weirder is that it did work initially, then I noticed I had a typo in the port (8080 instead of 80) for my http host, fixed that and restarted ironic21:49
* mrda reads scrollback as his ears are burning21:49
nagyzI however have not restarted the server to start "fresh" - shall I try?21:49
nagyzI issued a power reset, let's see21:49
lucas-dinnernagyz, shouldn't be needed... after you set that dnsmasq config in neutron. Did you restart the neutron-dhcp-agent service right?21:50
lucas-dinnerwhen you do a ps ax | grep dnsmasq21:50
lucas-dinnerdo you see dnsmasq with --config-file=<ur config file>21:50
lucas-dinner?21:50
nagyzright, I do. I even see the right config option file21:50
nagyzlet me do a pastebin21:50
lucas-dinnerright hmmm /me thinks21:51
nagyzhttp://pastebin.com/7sgdSz8K21:51
nagyzthis also looks good21:51
nagyzyes in the log:21:51
nagyzMay 11 23:51:37 kilo-controller in.tftpd[15951]: RRQ from 9.4.196.106 filename http://9.4.196.1//boot.ipxe remapped to /tftpboot/http://9.4.196.1//boot.ipxe21:51
nagyzMay 11 23:51:37 kilo-controller in.tftpd[15951]: sending NAK (1, File not found) to 9.4.196.10621:51
lucas-dinnerhttp://9.4.196.1//boot.ipxe no port here?21:52
lucas-dinnercoming from 80?21:52
nagyzshould be - I just put the files into the main apache dir.21:52
*** pcrews has quit IRC21:52
nagyzand it works from firefox :-)21:52
lucas-dinnergotcha21:52
lucas-dinneryeah21:52
lucas-dinnershould be fine21:52
nagyzI do see this in the option file:21:53
nagyztag:132b0f88-f86f-4576-a14e-0effe8623a30,tag:!ipxe,option:bootfile-name,undionly.kpxe21:53
nagyzbut seems like this is not working21:53
lucas-dinneryeah, that's making use of the ipxe tag that is created by that custom neutron configuration21:53
lucas-dinnernagyz, do a "ps ax | grep dnsmasq" please21:53
nagyzhttp://pastebin.com/4DSqvYvB21:54
nagyzeverything looks good21:56
lucas-dinneryeah conf file is there hmmm yeah21:57
jrollhas anyone ever thought about using ironic to provision networking gear like switches and firewalls?21:57
jrolldevananda: ^21:57
lucas-dinneroff the top of my head I've no idea. More weird is that as u said was working before21:57
lucas-dinnernagyz, looks good indeed21:57
lucas-dinnerjroll, from time to time I see people talking about it21:57
nagyzI can try to blow away the node and recreate21:58
jrolllucas-dinner: could be fun :D21:58
nagyzjroll, I'd really, really, really (can't emphasize this enough) prefer if the developers actually could focus on making it usable instead of adding more "features" in.21:58
nagyzfor example there seems to be absolutely no way to do ANYTHING with a node that's stuck in a cleaning phase. zero. zilch. nada.21:59
nagyzbesides manually going into the DB and setting it's power state to off and then deleting it.21:59
jrollnagyz: I understand your frustration; I run ironic in production booting thousands of instances per day. it's immature, and I feel the pain too.22:00
lucas-dinnerjroll, yeah, looks a bit tricky tho would require some coordination with something else to reconfigure the switch after it's reprovisioned, and some way to not lose connectivity when deploying the image on the switch too :-)22:00
nagyzlucas-dinner, I guess I'll go to sleep and in the morning blow the node away and retry with a fresh mind22:00
lucas-dinnernagyz, ack... lemme know it's quite late here too22:00
nagyzit's just turned midnight22:00
lucas-dinnernagyz, I can try to help you more tomorrow with it22:00
nagyzjroll, don't get me wrong I think it's super that there is something like this in openstack22:00
lucas-dinnernagyz, cool, yeah almost 00:00 here too22:00
lucas-dinnernagyz, (I'm UTC +1 btw)22:01
nagyzjroll, and if I do run into minor issues I'll submit a patch22:01
jrollnagyz: however, "different types of hardware" isn't really a feature, it's just another driver, the driver interactions work well and are well-understood :)22:01
nagyzlucas-dinner, I'm UTC+1 too :) (Zurich, Switzerland)22:01
jrolllucas-dinner: I'm thinking new switches, primarily, not doing the triple-o style thing with using images to upgrade software22:01
lucas-dinnernagyz, cool, yeah talk to u in the morning tomorrow then. I'm in Dublin, Ireland22:01
nagyzcheers22:02
nagyzhave a good night y'all22:02
lucas-dinnernagyz, g'night22:02
jrollnight nagyz :)22:02
*** Marga_ has quit IRC22:04
*** Marga_ has joined #openstack-ironic22:04
lucas-dinnerjroll, http://onie.org/ :-)22:07
jroll:)22:07
*** subscope has quit IRC22:09
devanandafor those curious about meeting attendance and activity ... lucas inspired me to crunch some numbers - http://paste.openstack.org/show/220218/22:10
*** uggla_ has joined #openstack-ironic22:10
lucas-dinnerdevananda, +1 looks good! I think it helps us to make the decision weather alternates meetings have been a good thing or not22:11
*** uggla has quit IRC22:11
jrolldevananda: anecdotal data point: I think tuesday meeting attendance seemed to drop a lot after freeze22:12
jlvillaldevananda: Also when the switch to daylight savings time happened it impacted me.  9-10pm was more doable for me than 10-11pm.22:15
*** subscope has joined #openstack-ironic22:18
devanandajlvillal: *nod*22:20
jrollditto22:21
*** jgrimm is now known as zz_jgrimm22:24
*** rwsu has quit IRC22:25
lucas-dinnerfolks have a good night!22:25
BadCubg'night lucas-dinner22:25
*** drwho has quit IRC22:26
lucas-dinnerBadCub, g'night :D22:26
*** lucas-dinner has quit IRC22:28
*** natorious is now known as zz_natorious22:30
*** pcrews has joined #openstack-ironic22:40
*** achanda has joined #openstack-ironic22:46
NobodyCambrb22:50
*** Sukhdev has quit IRC22:52
BadCubI am taking leave for the evening. Have a good night everyone!22:56
*** BadCub is now known as BadCub-Hiding22:56
mrdaNight BadCub-Hiding22:59
mrdadevananda: have you used dogpile.cache before?23:01
NobodyCammorning mrda23:04
mrdahey NobodyCam23:04
NobodyCam:)23:04
*** yuanying has joined #openstack-ironic23:04
devanandamrda: i have not personally, no. but see os-client-config and shade, and the irc conversation earlier about these23:07
*** subscope has quit IRC23:10
mrdaYeah, been looking.  I think we just want a human readable file serialisation format, not dbm.  dogpile.cache might be overkill.23:10
* mrda keeps reading23:10
devanandawhy human readable?23:12
jrolldebugging maybe?23:13
mrdaA unix philosophy is good.  Make things easy for operators/other tooling.23:14
devanandasure, it's good to be able to use other tools, but human readabl file format isn't the only way to do that23:15
devanandasqlite isn't human readable, but i can work with it just the same23:15
devananda(as an example)23:15
mrdaI like what rloo suggested, something along the lines of - if I don't want the caches, I just want to be able to blow them away23:15
mrdaI also like the idea of one file per ironic server, each with the cached ironic server microversion23:16
devananda++ to both of those23:16
mrdabut I do agree with your statement that TMTOWTDI23:16
openstackgerritJohn L. Villalovos proposed openstack/python-ironicclient: Refactor resource_fields.py  https://review.openstack.org/18198523:17
mrdaI'll keep looking and see what I can come up with23:17
*** hemna_ is now known as hemnafk23:19
devanandamrda: I get the sense you object to dogpile, but I dont know why23:19
mrdadevananda: No, no objection here.  It's just targeted at a bigger problem than whgat we're trying to solve.23:21
mrdaOr should I say, the documentation seems to suggest that it is targeted at bigger problems23:21
mrdaif it's fit for purpose for what we want, and it's already in openstack usage, then sure, no problems here23:21
jrollwe basically want to cache an integer, right?23:22
mrdayep23:23
jrollheh23:23
mrdaone version per hostname:port combination23:23
mrda(ref https://review.openstack.org//#/c/173674/)23:23
mrdajroll: ^^23:23
devanandayes. but we conceivably also want to cache the session token23:24
mrdaoh, didn't know that.23:24
devanandaand, with openstackclient, want to be part of the same caching that is used for other client libs23:24
jrollI personally don't want my auth token on disk23:24
mrdathat's a nice idea devananda23:24
mrdalol, or I could be wrong :)23:24
devanandajroll: dogpile supports other (eg, in-mem) backends23:24
jrollcool23:25
mrdabut I don't think we're suggesting that you need memcache to run ironicclient?23:25
devanandamy point is that there's more to caching than just storing an int, and that other client projects are already adopting dogpile for doing this23:25
devanandamrda: certainly not :)23:25
devanandamrda: but IF you're using memcache, say, alongside nova, then perhaps your nova virt driver might cache the session in that memcache23:26
devanandamrda: rather than keep generating new auth for every request ;)23:26
mrdasure23:26
mrda(I've seen this problem before, remember working on that token caching bit)23:26
devanandathis isn't an attempt to implement all of that. but it doesn't hurt to use a tool that will, someday, let us do that -- rather than roll a completely new caching implementation23:26
*** mdbooth has quit IRC23:26
mrdaI'd like to keep it simple (and hence maintainable for operators)23:27
mrdabut happy to take direction too23:27
devanandaagreed23:27
mrda:)23:27
*** arif-ali has quit IRC23:28
*** arif-ali has joined #openstack-ironic23:29
mrdaafk, be back in a short while23:31
*** kkoski has joined #openstack-ironic23:32
*** mdbooth has joined #openstack-ironic23:32
*** kkoski has quit IRC23:33
NobodyCamokay need to head out a bit early today.23:34
*** kkoski has joined #openstack-ironic23:35
*** kkoski has quit IRC23:45
*** kkoski has joined #openstack-ironic23:46
*** mrmoje has quit IRC23:51
*** oomichi has joined #openstack-ironic23:52
-openstackstatus- NOTICE: Gerrit is going offline while we perform an emergency downgrade to version 2.8.23:53
*** ChanServ changes topic to "Gerrit is going offline while we perform an emergency downgrade to version 2.8."23:54
*** kkoski has quit IRC23:59

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!