jroll | do you have logs or anything? | 00:00 |
---|---|---|
Pradip | yes i have | 00:00 |
jroll | cirros should work | 00:00 |
Pradip | but the baremetal image created by devstack is also cirros | 00:01 |
jroll | yes | 00:01 |
*** smoriya has joined #openstack-ironic | 00:01 | |
jroll | so console logs would be great | 00:01 |
Pradip | the controller computer's network interface is getting the dhcp msgs | 00:02 |
Pradip | i checked using tcp dump | 00:02 |
Pradip | but i think it is not oing through | 00:02 |
Pradip | what logs do you need? | 00:02 |
jroll | hm | 00:03 |
jroll | are you seeing it send a dhcp reservation back? | 00:03 |
jroll | and getting a DHCPACK? | 00:03 |
Pradip | nooe | 00:03 |
Pradip | nope | 00:03 |
Pradip | that's the issue i am facing | 00:03 |
jroll | so you're seeing DHCPDISCOVER on the controller | 00:04 |
jroll | but no DHCPOFFER on the controller? | 00:04 |
jroll | are you seeing a message about "no free leases" or any other errors there? | 00:04 |
jroll | I'm not sure what neutron's dhcp logs look like, btw, I'm just shooting in the dark here | 00:05 |
jroll | but yes, dhcp logs from the controller would be helpful | 00:05 |
jroll | (or tcpdump) | 00:05 |
*** ijw has joined #openstack-ironic | 00:05 | |
Pradip | ok. give me a moment | 00:05 |
jroll | no problem | 00:05 |
*** naohirot has joined #openstack-ironic | 00:06 | |
Pradip | There is no DHCP messages in the log. Only that DHCP has started | 00:07 |
Pradip | basically I used tcp dump on the physical interface. there I can see packets | 00:07 |
Pradip | but nowhere inside the logs | 00:08 |
*** chlong has quit IRC | 00:08 | |
Pradip | jroll: So basically I think the DHCP module is not receiving any request | 00:09 |
jroll | but you do see packets on the controller node? | 00:10 |
*** Sukhdev has quit IRC | 00:10 | |
Pradip | only on the physical interface | 00:10 |
Pradip | i think it is being dropped at the interface | 00:11 |
jroll | the physical interface for the controller or for the node you are deploying to? | 00:11 |
Pradip | for the controller | 00:11 |
jroll | is the dhcp service listening on the correct interface? | 00:11 |
Pradip | how to check that | 00:11 |
jroll | netstat -tulnp | grep dnsmasq | 00:12 |
Pradip | jroll: I didn't get any output for that command | 00:13 |
jroll | this is using neutron dhcp, right? | 00:13 |
Pradip | yes | 00:13 |
*** r-daneel has quit IRC | 00:13 | |
jroll | which should be using dnsmasq if I'm not mistaken | 00:13 |
jroll | but it sounds like that isn't running | 00:13 |
Pradip | yes. I think so too. | 00:13 |
*** kkoski has quit IRC | 00:14 | |
Pradip | Do you have any sample configuration for the devstack so that I can check what am i missing? | 00:15 |
jroll | for which part? neutron? | 00:15 |
Pradip | yes | 00:15 |
jroll | sure, one sec | 00:15 |
JoshNang | Pradip: did you run that netstat as root or with sudo? i'm not sure it'll work correctly without that | 00:16 |
Pradip | yes suing sudo | 00:16 |
Pradip | using | 00:16 |
jroll | hm, it returned nothing for me too | 00:16 |
jroll | I guess check: ps aux | grep dnsmasq | 00:16 |
Pradip | yes. i checked that already | 00:17 |
Pradip | it is there | 00:17 |
jroll | aha | 00:17 |
jroll | sudo netstat -tulnp | grep dhclient | 00:17 |
jroll | though that doesn't sound like a dhcp server. | 00:17 |
Pradip | yes that is there too | 00:17 |
jroll | ignore that | 00:18 |
Pradip | ok | 00:18 |
jroll | I don't see anything special about dhcp in my neutron.conf | 00:19 |
Pradip | oh | 00:20 |
Pradip | maybe I am missing something in the local.conf file in devstack | 00:20 |
Pradip | btw I didn't set the private and public network. I let it choose the default one. can it be the problem? | 00:21 |
*** cing has joined #openstack-ironic | 00:21 | |
jroll | default should be fine | 00:23 |
jroll | my localrc has this, maybe related: | 00:24 |
jroll | NETWORK_GATEWAY=10.1.0.1 | 00:24 |
jroll | FIXED_RANGE=10.1.0.0/24 | 00:24 |
jroll | FIXED_NETWORK_SIZE=256 | 00:24 |
Pradip | that shouldn't be the problem | 00:26 |
Pradip | by default it choose 10.0.0.1 | 00:26 |
jroll | yeah, agree | 00:26 |
Pradip | and 10.0.0.0/24 to be the network | 00:26 |
jroll | but those are the only things in my localrc that look relevant | 00:26 |
Pradip | what services do you have enable? | 00:26 |
jroll | I'm using this localrc http://docs.openstack.org/developer/ironic/dev/dev-quickstart.html#deploying-ironic-with-devstack | 00:27 |
Pradip | ok. I am following the same one. I don't know what's wrong. anyways thanks for all the help. i will try to look into this | 00:28 |
jroll | you're welcome, good luck | 00:29 |
jroll | maybe make sure your switches are forwading dhcp etc | 00:29 |
jroll | I guess they are if it gets to the controller, never mind | 00:29 |
Pradip | thanks. Yes. for that I am not using any switch. I directly connected them using a hub | 00:30 |
Pradip | yes | 00:30 |
rloo | jroll: wrt network-provider vs network-driver. Does one of them describe it better than the other? I prefer provider only cuz there are so many different drivers around :) | 00:34 |
jroll | rloo: yeah, I think I like provider, I updated accordingly | 00:35 |
rloo | jroll: thx. I'll look at the spec tomorrow. going to try to sign off soon! | 00:35 |
jroll | same here, have a good night :) | 00:35 |
rloo | jroll: you too! | 00:35 |
openstackgerrit | Josh Gachnang proposed openstack/ironic-python-agent: Allow hardware managers to override clean step priority https://review.openstack.org/189969 | 00:50 |
*** cing has quit IRC | 00:54 | |
*** amotoki has joined #openstack-ironic | 01:02 | |
*** zhenguo has joined #openstack-ironic | 01:03 | |
*** ijw has quit IRC | 01:12 | |
*** achanda has joined #openstack-ironic | 01:13 | |
*** rloo has quit IRC | 01:15 | |
*** mtanino has quit IRC | 01:22 | |
openstackgerrit | Stephanie Miller proposed openstack/bifrost: Add libvirt package dependency to install step https://review.openstack.org/198932 | 01:27 |
*** cing has joined #openstack-ironic | 01:28 | |
*** chlong has joined #openstack-ironic | 01:30 | |
*** Haomeng has joined #openstack-ironic | 01:38 | |
*** kan__ has joined #openstack-ironic | 01:49 | |
*** Pradip has quit IRC | 01:54 | |
*** puranamr has quit IRC | 02:09 | |
openstackgerrit | Yuiko Takada proposed openstack/ironic-inspector: Migrate to oslo_db https://review.openstack.org/181905 | 02:28 |
*** pal has joined #openstack-ironic | 02:34 | |
*** oomichi has joined #openstack-ironic | 02:37 | |
openstackgerrit | Stephanie Miller proposed openstack/bifrost: Fix check for >= Ubuntu 14.10 https://review.openstack.org/198943 | 02:47 |
*** pal has quit IRC | 02:51 | |
openstackgerrit | Stephanie Miller proposed openstack/bifrost: Add libvirt package dependency to install step https://review.openstack.org/198932 | 02:52 |
*** hakimo has quit IRC | 02:54 | |
*** hakimo has joined #openstack-ironic | 02:55 | |
openstackgerrit | Yuiko Takada proposed openstack/ironic-inspector: Stop checking the status of node if nova boot fails https://review.openstack.org/198946 | 03:11 |
*** [1]shalq has joined #openstack-ironic | 03:20 | |
*** shalq has quit IRC | 03:22 | |
*** [1]shalq is now known as shalq | 03:22 | |
*** yuikotakada has joined #openstack-ironic | 03:54 | |
*** yuikotakada has quit IRC | 03:54 | |
*** Nisha has joined #openstack-ironic | 03:59 | |
*** pal has joined #openstack-ironic | 04:17 | |
*** Marga_ has joined #openstack-ironic | 04:32 | |
*** Marga_ has quit IRC | 04:32 | |
*** Marga_ has joined #openstack-ironic | 04:33 | |
*** pal_ has joined #openstack-ironic | 04:36 | |
openstackgerrit | Ramakrishnan G proposed stackforge/proliantutils: IPA: Update proliant hardware manager https://review.openstack.org/198603 | 04:37 |
*** Sukhdev has joined #openstack-ironic | 04:38 | |
*** pal has quit IRC | 04:39 | |
*** rameshg87 has joined #openstack-ironic | 04:50 | |
*** bizarrochristy has joined #openstack-ironic | 05:02 | |
mrda | No-one coming to the Ironic meetiong on now? | 05:03 |
jroll | apparently not | 05:04 |
*** devanand_ has joined #openstack-ironic | 05:05 | |
*** bizarrochristy has quit IRC | 05:07 | |
*** [1]shalq has joined #openstack-ironic | 05:11 | |
*** shalq has quit IRC | 05:13 | |
*** [1]shalq is now known as shalq | 05:13 | |
openstackgerrit | Merged openstack/ironic: Allow deleting even associated and active node in maintenance mode https://review.openstack.org/191736 | 05:15 |
*** Marga_ has quit IRC | 05:18 | |
*** Marga_ has joined #openstack-ironic | 05:19 | |
*** e0ne has joined #openstack-ironic | 05:22 | |
*** achanda has quit IRC | 05:23 | |
naohirot | rameshg87: jroll: I have a problem when IPA core os starts, console says "Failed to start user-configdrive.service." | 05:29 |
naohirot | rameshg87: jroll: do you have any idea what is wrong? This started when I recreated ipa iso yesterday | 05:30 |
rameshg87 | naohirot: something goes wrong because of that ? | 05:31 |
rameshg87 | naohirot: there is an option to prevent that | 05:31 |
rameshg87 | naohirot: https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/agent_config.template#L5 | 05:32 |
rameshg87 | naohirot: check the end of the line - coreos.configdrive=0 | 05:32 |
rameshg87 | naohirot: you might need to build an iso with these arguments in place | 05:32 |
*** lazy_prince has joined #openstack-ironic | 05:32 | |
rameshg87 | naohirot: we don't know because we generally use fedora based agent ramdisk for our testing with ilo drivers | 05:33 |
naohirot | rameshg87: let me check the URL | 05:33 |
rameshg87 | coreos doesn't boot properly on some of proliant hardware | 05:33 |
lazy_prince | naohirot: rameshg87: I put up a patch for IPA support in debian and ubuntu.. | 05:34 |
lazy_prince | https://review.openstack.org/#/c/198771/ | 05:34 |
naohirot | rameshg87: I'm having strange problem in my lab setting. | 05:36 |
naohirot | rameshg87: I'm not sure that this configdrive error happened before. | 05:37 |
rameshg87 | lazy_prince: oh great :) | 05:37 |
rameshg87 | naohirot: I think it could be because of that | 05:37 |
rameshg87 | naohirot: try adding that option in your iso as a kernel command line argument | 05:37 |
naohirot | rameshg87: but when I investigate my problem, I noticed that console says the configdrive error. | 05:37 |
rameshg87 | naohirot: okay | 05:37 |
naohirot | rameshg87: I wondered if it is related to my problem or not. | 05:38 |
naohirot | rameshg87: so you mean that | 05:38 |
rameshg87 | naohirot: I think it is | 05:38 |
rameshg87 | naohirot: worth a try | 05:38 |
rameshg87 | naohirot: add the kernel command line argument coreos.configdrive=0 in your iso | 05:39 |
rameshg87 | naohirot: and try to boot it up | 05:39 |
rameshg87 | naohirot: because configdrive shouldn't be loaded when booting in the ipa environment | 05:39 |
naohirot | rameshg87: https://github.com/openstack/ironic-python-agent/blob/master/imagebuild/coreos/oem/cloud-config.yml#L5-L17 | 05:39 |
naohirot | rameshg87: the line 7 to 17 are not necessary? | 05:39 |
*** yuanying_ has joined #openstack-ironic | 05:40 | |
rameshg87 | naohirot: I think it means to not start these services, right ? | 05:40 |
naohirot | rameshg87: Aha, I see | 05:40 |
rameshg87 | naohirot: so user-configdrive.service shouldn't have started infact | 05:40 |
* rameshg87 doesn't have much experience with coreos | 05:41 | |
rameshg87 | naohirot: all are my guess. I haven't worked much on coreos | 05:41 |
naohirot | rameshg87: but conductor log says conductor sends configdrive data to IPA when conductor does heart beat. | 05:41 |
rameshg87 | naohirot: that's for the user image | 05:42 |
naohirot | rameshg87: I see | 05:42 |
rameshg87 | naohirot: the ipa environment as such should load the config drive | 05:42 |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Add RAIDInterface for RAID configuration https://review.openstack.org/196003 | 05:43 |
*** yuanying has quit IRC | 05:44 | |
naohirot | rameshg87: Okay, I'll check my environment again. | 05:44 |
*** yuanying_ has quit IRC | 05:44 | |
rameshg87 | ack | 05:44 |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Add RPCAPIs for RAID configuration https://review.openstack.org/196006 | 05:44 |
*** e0ne has quit IRC | 05:44 | |
naohirot | rameshg87: thanks1 | 05:44 |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Add APIs for RAID configuration https://review.openstack.org/196007 | 05:44 |
*** itamarl has joined #openstack-ironic | 05:51 | |
*** yuanying has joined #openstack-ironic | 05:51 | |
*** ig0r_ has joined #openstack-ironic | 05:52 | |
*** dasm has quit IRC | 05:53 | |
naohirot | rameshg87: >you might need to build an iso with these arguments in place | 05:53 |
naohirot | rameshg87: how do we do the above? | 05:53 |
rameshg87 | naohirot: how did you build ipa iso ? | 05:55 |
*** dasm has joined #openstack-ironic | 05:55 | |
naohirot | rameshg87: just "make iso" in IPA tree is not enough? | 05:55 |
*** ig0r__ has quit IRC | 05:55 | |
naohirot | rameshg87: Yes. >how did you build ipa iso ? | 05:55 |
naohirot | rameshg87: I'm sorry, you asked me the question. | 05:56 |
naohirot | rameshg87: I did "make iso" in ironic-python-agent/imagebuild/coreos# | 05:56 |
*** ukalifon1 has joined #openstack-ironic | 05:58 | |
*** yuanying has quit IRC | 06:04 | |
*** praneshp has quit IRC | 06:06 | |
naohirot | lazy_prince: I just looked at your patch | 06:07 |
naohirot | lazy_prince: when do we use ironic-agent element? | 06:07 |
naohirot | lazy_prince: I think it is for ilo_scsi right? | 06:08 |
naohirot | lazy_prince: it's not for iscsi_ilo, but for agent_ilo, I believe. | 06:09 |
naohirot | lazy_prince: it's reversed | 06:10 |
*** yuanying has joined #openstack-ironic | 06:10 | |
naohirot | lazy_prince: it's not for agent_ilo, but for iscsi_ilo, I believe. | 06:10 |
zhenguo | hello rameshg87, can you help to review https://review.openstack.org/#/c/177642/, it last for a long time. | 06:11 |
*** yuanying has quit IRC | 06:14 | |
*** yuanying has joined #openstack-ironic | 06:16 | |
rameshg87 | zhenguo: sure. I will do that today itself. | 06:19 |
zhenguo | rameshg87: thanks :) | 06:20 |
lazy_prince | naohirot: ironic-agent element is used to build a ramdisk image (deploy image) for use with ironic. | 06:24 |
naohirot | lazy_prince: hi | 06:25 |
lazy_prince | naohirot: hey.. | 06:25 |
naohirot | lazy_prince: so do you mean for both iscsi_ilo and agent_ilo? | 06:25 |
lazy_prince | naohirot: the deploy images built using ironic-agent will have agent embedded in it.. its just an image.. now, iLO is more for the way of power management or booting. | 06:26 |
lazy_prince | naohirot: so you could use the image with all agent_* or *_agent drivers.. | 06:27 |
naohirot | lazy_prince: I believe the DIB and the ironic-agent element for iscsi_ilo. | 06:27 |
*** radek__ has joined #openstack-ironic | 06:28 | |
lazy_prince | naohirot: i think there used to be a ironic-deploy element for iscsi al iscsi_* drivers.. | 06:28 |
lazy_prince | naohirot: let me check if its still there.. | 06:28 |
*** dtantsur|afk is now known as dtantsur | 06:28 | |
dtantsur | Morning Ironic | 06:29 |
naohirot | lazy_prince: really? do you mean that IPA is not used for agent_ilo? https://github.com/openstack/ironic-python-agent | 06:29 |
lazy_prince | naohirot: IPA is used for agent_iLO.. but not for iscsi_ilo.. | 06:30 |
lazy_prince | naohirot: https://github.com/openstack/diskimage-builder/tree/master/elements/deploy-ironic | 06:30 |
naohirot | lazy_prince: that makes sense. | 06:30 |
*** pal_ has quit IRC | 06:30 | |
lazy_prince | naohirot: https://github.com/openstack/diskimage-builder/tree/master/elements/ironic-agent | 06:31 |
naohirot | lazy_prince: but DIB with ironic-agent element works for both? | 06:31 |
lazy_prince | naohirot: you can use DIB to build images with multiple elements. a combination od deploy-ironic and ironic agent could yield an image that can work for iscsi and agent... | 06:32 |
lazy_prince | s/od/of/ | 06:32 |
naohirot | lazy_prince: oh really? I thought that Ironic is going to shift from DIB to IPA. | 06:33 |
*** Sukhdev has quit IRC | 06:34 | |
naohirot | lazy_prince: If it is true, as if we are going to shift from IPA to DIB. | 06:34 |
lazy_prince | naohirot: DIB is just a tool to generate deploy images.. be it iscsi deploy or IPA based deploy... what I know is IPA will be the way forward.. but it does not mean that IPA images will not be generated by using DIB... | 06:34 |
naohirot | lazy_prince: I see, we continue to maintain both DIB and IPA so that both can generate core os deploy image. Am I right? | 06:36 |
lazy_prince | naohirot: not sure if DIB IPA element supports generating IPA image based on core os. but its good to have support extended to Fedora/Ubuntu/Debian OSes too.. | 06:38 |
naohirot | lazy_prince: It seems I misunderstood that DIB with the ironic-agent element generate core os image. | 06:40 |
lazy_prince | naohirot: better late than never.. :) | 06:40 |
rameshg87 | naohirot: sorry, I seem to have missed your chat | 06:42 |
rameshg87 | naohirot: anyway I think you got reply from lazy_prince | 06:42 |
naohirot | lazy_prince: what do we call the image which DIB with the ironic-agent elemenet generates? | 06:42 |
naohirot | lazy_prince: If we call the image IPA, it's very confusing. | 06:43 |
rameshg87 | naohirot: for coreos iso-image-create script is used when you do a make iso | 06:43 |
rameshg87 | naohirot: https://github.com/openstack/ironic-python-agent/blob/master/imagebuild/coreos/iso-image-create#L126 | 06:43 |
lazy_prince | naohirot: maybe "fedora IPA image" | 06:43 |
rameshg87 | naohirot: youcan add the kernel cmdline I mentioned earlier on this line and give a try | 06:44 |
*** rameshg87 is now known as rameshg87-away | 06:45 | |
naohirot | rameshg87: Okay, I'll try again. | 06:45 |
naohirot | rameshg87-away: Yesterday, I tried to add coreos.autologin into the line. It didn't work. Maybe I missed something. | 06:46 |
naohirot | lazy_prince: that makes sense:-) >maybe "fedora IPA image" | 06:49 |
*** e0ne has joined #openstack-ironic | 06:57 | |
*** saripurigopi has joined #openstack-ironic | 07:00 | |
*** Marga__ has joined #openstack-ironic | 07:03 | |
*** e0ne has quit IRC | 07:04 | |
*** Marga_ has quit IRC | 07:06 | |
*** ifarkas has joined #openstack-ironic | 07:10 | |
*** yog__ has joined #openstack-ironic | 07:14 | |
*** oomichi has quit IRC | 07:22 | |
*** e0ne has joined #openstack-ironic | 07:36 | |
saripurigopi | hello Ironic | 07:37 |
*** jistr has joined #openstack-ironic | 07:37 | |
*** devanand_ has quit IRC | 07:41 | |
*** bizarrochristy has joined #openstack-ironic | 07:46 | |
*** max_lobur has joined #openstack-ironic | 07:59 | |
*** e0ne is now known as e0ne_ | 08:01 | |
*** ricky2 has joined #openstack-ironic | 08:02 | |
ricky2 | heya | 08:02 |
*** lucasagomes has joined #openstack-ironic | 08:02 | |
*** ricky2 is now known as rcarrillocruz | 08:02 | |
rcarrillocruz | any bifrost user right now in the room? | 08:02 |
rcarrillocruz | i'm getting some error with my baremetal.json, presumed ipmi_address missing (it's there) | 08:02 |
rcarrillocruz | anyone could share their baremetal.json (with passwords, ips, etc removed) ? | 08:03 |
*** e0ne_ is now known as e0ne | 08:03 | |
*** chlong has quit IRC | 08:13 | |
Haomeng | rcarrillocruz: did you run with *ipmi* driver, so ipmi_address is required I think, right? | 08:15 |
*** derekh has joined #openstack-ironic | 08:17 | |
Haomeng | rcarrillocruz: for ipmi driver, we have to input ipmi_address and ipmi_username, ipmi_password are required for ipmi connection | 08:18 |
Haomeng | rcarrillocruz: so make sure your *driver_info* are enough for these fields | 08:18 |
rcarrillocruz | Haomeng: http://paste.openstack.org/show/351210/ | 08:19 |
rcarrillocruz | when i run the deploy playbook, I get "msg: RPC do_node_deploy failed to validate deploy or power info. Error: Missing the following IPMI credentials in node's driver_info: ['ipmi_address']. (HTTP 500) | 08:19 |
rcarrillocruz | " | 08:19 |
rcarrillocruz | but ipmi_address is there | 08:19 |
rcarrillocruz | so is username and password | 08:19 |
Haomeng | rcarrillocruz: let me check playbook code | 08:21 |
*** jcoufal has joined #openstack-ironic | 08:22 | |
lucasagomes | rcarrillocruz, I don't know much bifrotst, but you can update the node using the ironic api directly right? | 08:25 |
rcarrillocruz | yeah, i can. It seems to me a JSON parsing issue, if I use the CSV style inventory it works, but it fails with JSON | 08:27 |
Haomeng | rcarrillocruz: yes, for the error message, for ironic node properties, they are not passed by json file as well | 08:28 |
Haomeng | lucasagomes: good morning:) | 08:28 |
lucasagomes | Haomeng, good morning! | 08:29 |
Haomeng | lucasagomes: :) | 08:29 |
Haomeng | rcarrillocruz: can you paste both the csv and the json which converted via csv, we can have difference to check which fields are wrong convered | 08:30 |
*** bizarrochristy has quit IRC | 08:31 | |
*** bizarrochristy has joined #openstack-ironic | 08:32 | |
openstackgerrit | Yuiko Takada proposed openstack/ironic-inspector: Specify server id clearly on the gate test https://review.openstack.org/199012 | 08:32 |
Haomeng | rcarrillocruz: and share your ironic node driver properties also | 08:33 |
*** yuikotakada has joined #openstack-ironic | 08:38 | |
yuikotakada | Hi, Ironic | 08:39 |
*** pelix has joined #openstack-ironic | 08:39 | |
lucasagomes | yuikotakada, good morning | 08:40 |
lucasagomes | ugt* | 08:40 |
yuikotakada | lucasagomes, good ugt morning :) | 08:40 |
*** romcheg has joined #openstack-ironic | 08:40 | |
rameshg87-away | zhenguo: hi | 08:42 |
dtantsur | yuikotakada, lucasagomes, o/ | 08:43 |
lucasagomes | dtantsur, hi there, welcome back | 08:43 |
lucasagomes | how was the holiday? | 08:43 |
*** sambetts has quit IRC | 08:44 | |
zhenguo | rameshg87: hi | 08:44 |
dtantsur | pretty good, we went to Vienna for 2 nights :) | 08:44 |
yuikotakada | dtantsur, zdrastviche :) | 08:44 |
lucasagomes | dtantsur, ah nice one | 08:44 |
dtantsur | :) | 08:44 |
lucasagomes | I just passed through vienna once by car, never actually stopped | 08:44 |
lucasagomes | but looks beautiful | 08:44 |
dtantsur | yeah | 08:46 |
*** sambetts has joined #openstack-ironic | 08:46 | |
dtantsur | though it could be better, were it not for +35C outside :D | 08:47 |
lucasagomes | omg 35C man | 08:48 |
* lucasagomes likes the Irish summer, 16C | 08:49 | |
lucasagomes | actually in the afternoon it may go up to 20ish | 08:49 |
lucasagomes | dtantsur, tho if you had a swimming pool 35C may be great | 08:49 |
dtantsur | heh sounds nice :) | 08:49 |
dtantsur | lucasagomes, well, we didn't it | 08:50 |
Nisha | dtantsur, hi | 08:50 |
dtantsur | Nisha, hi | 08:50 |
* dtantsur relocates to the office now, sorry | 08:50 | |
Nisha | i was starting inspector in devstack....getting this error | 08:51 |
Nisha | Another app is currently holding the xtables lock | 08:51 |
Nisha | dtantsur, 5 min plz | 08:51 |
rcarrillocruz | k, i figured | 08:51 |
rcarrillocruz | so, i changed the baremetal.json | 08:51 |
rcarrillocruz | after doing the enroll | 08:51 |
dtantsur | Nisha, that seems to be a bug in the current devstack | 08:51 |
rcarrillocruz | and must had the missing ipmi params when i enrolled it | 08:51 |
Nisha | inspector doesnt start bec of that | 08:51 |
rcarrillocruz | enrolling the node again with the baremetal.json fixed that | 08:51 |
dtantsur | Nisha, I can't build my environment right now doe to that.. | 08:51 |
dtantsur | Nisha, mind reporting to devstack launchpad? | 08:52 |
Nisha | how to overcome this error? | 08:52 |
dtantsur | if I only knew... | 08:52 |
Nisha | previous builds of devstack worked fine with inspector? | 08:52 |
dtantsur | strangely enough, our gate seems to work | 08:52 |
Nisha | all the other services came up | 08:53 |
Nisha | it just stuck in inspector with this error | 08:53 |
dtantsur | Nisha, hmm, that maybe another problem | 08:53 |
dtantsur | Nisha, what's the real error? | 08:53 |
dtantsur | I mean, this xtables thingy is just failure to clean up | 08:53 |
* Nisha waiting for paste to finish | 08:55 | |
Nisha | http://paste.openstack.org/show/351334/ | 08:55 |
Nisha | no not in clean up | 08:56 |
dtantsur | Nisha, "There is a duplicated hook in your elements: ramdisk-base/post-install.d/01-ensure-binaries" <-- seems like DIB problem | 08:58 |
dtantsur | Nisha, hmm maybe try updating it: $ sudo pip install -U diskimage-builder | 08:59 |
Nisha | dtantsur, hmmm ok | 08:59 |
*** dtantsur is now known as dtantsur|brb | 09:00 | |
*** athomas has joined #openstack-ironic | 09:01 | |
*** e0ne is now known as e0ne_ | 09:11 | |
sambetts | Morning all o/ | 09:13 |
*** naohirot has quit IRC | 09:16 | |
*** boris-42 has quit IRC | 09:22 | |
*** e0ne_ is now known as e0ne | 09:24 | |
openstackgerrit | Merged openstack/ironic: Replace is_hostname_safe with a better check https://review.openstack.org/193587 | 09:33 |
*** killer_prince has joined #openstack-ironic | 09:34 | |
*** lazy_prince has quit IRC | 09:37 | |
*** lazy_prince has joined #openstack-ironic | 09:39 | |
*** killer_prince has quit IRC | 09:42 | |
*** smoriya has quit IRC | 09:53 | |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Add RAIDInterface for RAID configuration https://review.openstack.org/196003 | 09:56 |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Add RPCAPIs for RAID configuration https://review.openstack.org/196006 | 09:56 |
*** yuanying_ has joined #openstack-ironic | 09:57 | |
*** yuanying has quit IRC | 10:00 | |
*** Nisha has quit IRC | 10:00 | |
*** yuanying_ has quit IRC | 10:07 | |
*** yuanying has joined #openstack-ironic | 10:07 | |
*** e0ne is now known as e0ne_ | 10:07 | |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Add APIs for RAID configuration https://review.openstack.org/196007 | 10:11 |
rameshg87-away | lucasagomes: hi | 10:13 |
*** e0ne_ is now known as e0ne | 10:21 | |
lucasagomes | rameshg87-away, hi there | 10:22 |
TheJulia | Good morning | 10:22 |
*** dtantsur|brb is now known as dtantsur | 10:25 | |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Add support for inband raid configuration agent ramdisk https://review.openstack.org/198238 | 10:25 |
*** rameshg87-away is now known as rameshg87 | 10:25 | |
dtantsur | sambetts, TheJulia, morning | 10:25 |
rameshg87 | lucasagomes: hi | 10:25 |
rameshg87 | lucasagomes: in https://review.openstack.org/#/c/197141/4/ironic/conductor/manager.py | 10:25 |
sambetts | dtantsur: Hey :) | 10:25 |
rameshg87 | lucasagomes: one small optimization if it's possible | 10:26 |
*** athomas has quit IRC | 10:26 | |
lucasagomes | rameshg87, right | 10:26 |
rameshg87 | lucasagomes: we can check in the db itself if reservation in [c1, c2...] | 10:26 |
*** killer_prince has joined #openstack-ironic | 10:26 | |
rameshg87 | lucasagomes: but our dbapi doesn't seem to support it | 10:26 |
lucasagomes | yeah :-( I dunno how to do it with our current filters api | 10:27 |
rameshg87 | lucasagomes: yeah may be we can leave a todo note there and look at it later. right ? | 10:27 |
lucasagomes | rameshg87, we can open a bug about improving it | 10:27 |
lucasagomes | so we keep track | 10:27 |
rameshg87 | lucasagomes: yeah, +1 | 10:27 |
lucasagomes | cause I think this could be used in other places too | 10:27 |
lucasagomes | for example when we look at provision state for other stuff we could check if node.provision_state in [] | 10:28 |
rameshg87 | lucasagomes: oh yeah, it would be a meaningful enhancement too | 10:28 |
* rameshg87 is interested | 10:28 | |
rameshg87 | lucasagomes: I will file a bug and try to get some work on it this weekend | 10:29 |
lucasagomes | rameshg87, ack! that's good stuff | 10:29 |
lucasagomes | rameshg87, I could add a note for that patch if you want to | 10:29 |
*** lazy_prince has quit IRC | 10:30 | |
lucasagomes | but not sure if necessary, since it's more broad | 10:30 |
*** lazy_prince has joined #openstack-ironic | 10:30 | |
rameshg87 | lucasagomes: I will leave a comment there with the link to opened bug | 10:30 |
lucasagomes | rameshg87, ack, thanks! | 10:30 |
*** viktors|afk is now known as viktors | 10:30 | |
*** killer_prince has quit IRC | 10:31 | |
*** athomas has joined #openstack-ironic | 10:34 | |
*** romcheg has quit IRC | 10:38 | |
*** yuikotakada has quit IRC | 10:39 | |
*** romcheg has joined #openstack-ironic | 10:45 | |
*** bizarrochristy has quit IRC | 10:46 | |
*** max_lobur has quit IRC | 10:49 | |
*** bizarrochristy has joined #openstack-ironic | 10:49 | |
openstackgerrit | Dmitry Tantsur proposed openstack/ironic: Start using new ENROLL state https://review.openstack.org/194722 | 10:56 |
*** Marga__ has quit IRC | 11:00 | |
*** lucasagomes is now known as lucas-hungry | 11:01 | |
*** Marga_ has joined #openstack-ironic | 11:02 | |
*** davideagnello has quit IRC | 11:04 | |
rcarrillocruz | hey folks | 11:09 |
rcarrillocruz | to delete a node, it seems i can only do from manage/available | 11:10 |
rcarrillocruz | now | 11:10 |
rcarrillocruz | once a node is set to active | 11:10 |
rcarrillocruz | how can I move that state to something else? | 11:10 |
rameshg87 | rcarrillocruz: you need to tear down the node | 11:11 |
rameshg87 | rcarrillocruz: ironic node-set-provision-state <node-uuid> deleted | 11:11 |
*** max_lobur has joined #openstack-ironic | 11:11 | |
rcarrillocruz | hmm, ok, now cycling the server | 11:13 |
rcarrillocruz | went to cleaning state | 11:14 |
rcarrillocruz | thx, i should've put clean mode in ironic to off really | 11:14 |
*** Marga_ has quit IRC | 11:15 | |
*** Marga_ has joined #openstack-ironic | 11:17 | |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Clear ilo_boot_iso before deploy for glance images https://review.openstack.org/193436 | 11:21 |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Clear ilo_boot_iso before deploy for glance images https://review.openstack.org/193436 | 11:24 |
rameshg87 | rcarrillocruz: oh yeah | 11:25 |
rameshg87 | rcarrillocruz: you can set clean_nodes = false in ironic.conf | 11:25 |
* rameshg87 goes home | 11:25 | |
*** rameshg87 has quit IRC | 11:25 | |
*** e0ne is now known as e0ne_ | 11:25 | |
*** romcheg has quit IRC | 11:28 | |
*** cing has quit IRC | 11:33 | |
*** romcheg has joined #openstack-ironic | 11:41 | |
*** e0ne_ is now known as e0ne | 11:45 | |
openstackgerrit | Zhenguo Niu proposed openstack/ironic: When boot option is not persisted, set boot on next power on https://review.openstack.org/177642 | 11:46 |
*** derekh has quit IRC | 12:01 | |
*** trown|outttypeww is now known as trown | 12:01 | |
TheJulia | rcarrillocruz: I've pondered defaulting bifrost to cleaning off and putting explicit instructions in place for users to turn it on. | 12:01 |
*** derekh has joined #openstack-ironic | 12:02 | |
rcarrillocruz | yeah, ran ironic install play now with -e cleaning=false yadayada | 12:03 |
rcarrillocruz | otoh, i thought i could set arbitrary names for uuid | 12:03 |
rcarrillocruz | doesn't bifrost pull latest ironic? | 12:03 |
rcarrillocruz | i think i read we can have arbitrary uuids now in kilo? | 12:03 |
openstackgerrit | Julia Kreger proposed openstack/bifrost: WIP: Ansible 2.0 compatability **DO NOT MERGE** https://review.openstack.org/196832 | 12:04 |
*** chlong has joined #openstack-ironic | 12:06 | |
*** lucas-hungry is now known as lucasagomes | 12:07 | |
lucasagomes | rcarrillocruz, you can set hostname-like names for "name" | 12:08 |
lucasagomes | for the node resource | 12:08 |
rcarrillocruz | so, hostnames work | 12:08 |
lucasagomes | ironic node-update <uuid> add name=<name> | 12:08 |
rcarrillocruz | but fqdn won't ? | 12:08 |
rcarrillocruz | i tried fqdn | 12:08 |
rcarrillocruz | and got a complaint from the pertinent create rest api call | 12:08 |
lucasagomes | I think it supporter fqdn, but I'm not sure the checks where correctly | 12:09 |
lucasagomes | I approved a patch today fixing some stuff on that area | 12:09 |
lucasagomes | lemme find the link | 12:09 |
lucasagomes | https://review.openstack.org/#/c/193587/ | 12:09 |
lucasagomes | this also allows you to use a free form name for the API version >= 1.10 | 12:10 |
rcarrillocruz | hmm | 12:11 |
rcarrillocruz | i must have some versioning mismatch then | 12:11 |
rcarrillocruz | not even non-fqdn works for me now | 12:11 |
rcarrillocruz | Error registering machine with Ironic: Expected a uuid but received blah | 12:11 |
lucasagomes | it's not uuid | 12:12 |
lucasagomes | it's name | 12:12 |
lucasagomes | how you're setting it? | 12:12 |
rcarrillocruz | ah, ok, so you are telling me it's a different param if i want to set name, not uuid | 12:13 |
rcarrillocruz | that explains | 12:13 |
lucasagomes | yes | 12:13 |
rcarrillocruz | the playbook uses by default the uuid param | 12:13 |
rcarrillocruz | thus it will always complain | 12:13 |
lucasagomes | the field is called "name" | 12:13 |
rcarrillocruz | makes sense now | 12:13 |
rcarrillocruz | i thought it would do some uuid parsing, if it's not a uuid then assume it was a name | 12:13 |
lucasagomes | the uuid is the canonical identifier of the resource | 12:13 |
lucasagomes | and the name is just a human-friendly way to access it | 12:14 |
lucasagomes | so you can do GET v1/nodes/<name> | 12:14 |
lucasagomes | e.g | 12:14 |
rcarrillocruz | ++ | 12:15 |
openstackgerrit | Julia Kreger proposed openstack/bifrost: Attempt to upgrade iPXE if no mac found https://review.openstack.org/193385 | 12:15 |
*** dprince has joined #openstack-ironic | 12:15 | |
lazy_prince | lucasagomes: is it possible to build a deploy image which can work with both agent_ipmitool and pxe_ipmitool..? | 12:16 |
lucasagomes | lazy_prince, a deploy ramdisk? If you use the IPA ramdisk it works with both drivers already | 12:16 |
lazy_prince | lucasagomes: aha.. you mean the core os one or the one built using DIB ironic-agent..? | 12:17 |
lucasagomes | lazy_prince, both | 12:17 |
lucasagomes | tho I never tested the ironic-agent | 12:17 |
lazy_prince | lucasagomes: aha.. k.. will give it a try then... | 12:18 |
lucasagomes | lazy_prince, cool! | 12:18 |
rcarrillocruz | thankfully i can grab the uuid with ipmitool | 12:21 |
rcarrillocruz | yay | 12:21 |
*** bizarrochristy has quit IRC | 12:24 | |
lucasagomes | :-) | 12:27 |
*** bizarrochristy has joined #openstack-ironic | 12:30 | |
openstackgerrit | Julia Kreger proposed openstack/bifrost: Vagrantfile and vagrant.yml for testing https://review.openstack.org/191110 | 12:31 |
* TheJulia needs more coffee | 12:31 | |
*** ralfv has joined #openstack-ironic | 12:32 | |
rcarrillocruz | folks, one issue i have is that my servers the provisioning NIC is the second one. Per my searching, it seems it's not possible to set up the NIC boot order with ipmi, is it? | 12:32 |
TheJulia | I don't think it is. The chassises I've watched generally try all their NICs | 12:34 |
TheJulia | that are enabled | 12:34 |
rcarrillocruz | i must check what's going on | 12:35 |
rcarrillocruz | cos for some reason | 12:35 |
rcarrillocruz | whenever i deploy with ironic | 12:35 |
rcarrillocruz | it tries to pxe boot with first nic | 12:35 |
rcarrillocruz | then the hard disk is the second option in the boot sequence | 12:35 |
rcarrillocruz | then the prov nic | 12:35 |
TheJulia | That sounds like it is bios settings | 12:35 |
rcarrillocruz | thus, it always boot from hard disk, doesn't reach to the second nic | 12:35 |
rcarrillocruz | :/ | 12:35 |
rcarrillocruz | yep | 12:35 |
rcarrillocruz | off to iLO | 12:35 |
TheJulia | My supermicro chassis does both nics, but I had to enable the second nick. The moonshot, it just worked. | 12:36 |
TheJulia | Enjoy! | 12:36 |
openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Mock the file creation for the GetConfigdriveTestCase tests https://review.openstack.org/199078 | 12:38 |
lucasagomes | TheJulia, good morning! | 12:38 |
*** dguerri` is now known as dguerri | 12:39 | |
lucasagomes | rcarrillocruz, yeah I've seem systems that only reliably boot from the first NIC | 12:39 |
*** Shrews has joined #openstack-ironic | 12:40 | |
lucasagomes | some may fallback to other nics as TheJulia pointed out | 12:40 |
dguerri | hello guys, quick question. Is there a way to override/skip the erase_devices (run by IPA) during the clean phase, without disabling cleaning in Ironic? | 12:40 |
TheJulia | Shrews: Good morning | 12:41 |
Shrews | TheJulia: o/ | 12:41 |
TheJulia | dguerri: Hmmmm | 12:42 |
dguerri | I am trying to define a custom hardware manager with a method with the same name (erase_devices) and higher priority, but it seems bot versions (the one in genericHadware manager and mine HM) get executed | 12:42 |
dguerri | TheJulia: hey there :) | 12:42 |
dguerri | TheJulia: do you think it is possible? | 12:42 |
openstackgerrit | Sergey Vilgelm proposed openstack/ironic: Update ironic.conf.sample https://review.openstack.org/196002 | 12:42 |
TheJulia | dguerri: most of the IPA folks won't be up for a couple hours. I'm looking at the code now | 12:44 |
dguerri | TheJulia: thanks! | 12:44 |
dguerri | TheJulia: at first glance, it is not possible... | 12:45 |
TheJulia | That is the feeling I'm getting looking at it | 12:47 |
TheJulia | hardware.py line 144 | 12:49 |
TheJulia | So your hardware manager could define it's own get_clean_steps | 12:50 |
*** ukalifon1 has quit IRC | 12:53 | |
dguerri | TheJulia: yup, I did that. As in https://github.com/rackerlabs/onmetal-ironic-hardware-manager/blob/master/onmetal_ironic_hardware_manager/__init__.py#L64 | 12:54 |
dguerri | TheJulia: but the GenericHardwaremanager steps (actually just 1 step) still get executed | 12:54 |
TheJulia | That is a bug then | 12:55 |
TheJulia | because I would think that the presence of the hardware manager should override the generic manager, but I don't see anything to that effect glancing at the code | 12:56 |
* TheJulia may have not had enough coffee yet | 12:56 | |
*** rloo has joined #openstack-ironic | 13:08 | |
openstackgerrit | Davide Guerri proposed openstack/ironic-python-agent: Disable local disks shredding if agent_erase_devices_iterations is 0 https://review.openstack.org/199088 | 13:09 |
dguerri | TheJulia: it makes sense | 13:10 |
dguerri | TheJulia: but it seems a "logic" issue rather than a real bug: https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/hardware.py#L181 | 13:11 |
dguerri | so it works as it is supposed to :) | 13:11 |
dguerri | I proposed a patch ^ for my use case | 13:11 |
dguerri | I hope that the rationale makes sense | 13:11 |
*** blakec has joined #openstack-ironic | 13:12 | |
*** cing has joined #openstack-ironic | 13:16 | |
*** thrash|g0ne is now known as thrash | 13:17 | |
TheJulia | dguerri: split the difference, defect :) | 13:22 |
*** kkoski has joined #openstack-ironic | 13:22 | |
TheJulia | dguerri: It make sense, I just think there needs to be better doc around all of that :) | 13:22 |
*** tiagogomes_ has joined #openstack-ironic | 13:26 | |
tiagogomes_ | Hi, I though I could use logical names in Ironic starting from Kilo, but I got this error ` error: unrecognized arguments: -n devhw7` when I tried it | 13:26 |
tiagogomes_ | Ironic version is 0.4.1 | 13:27 |
jroll | tiagogomes_: update your ironic client | 13:29 |
ralfv | does anyone on here know of attempts to control Sparc T4/T5 serves using ipmitool ? | 13:29 |
jroll | dguerri: you shouldn't need to disable shredding that way, http://docs.openstack.org/developer/ironic-python-agent/#how-are-methods-executed-on-hardwaremanagers | 13:29 |
jroll | dguerri: can I see your hw manager code? | 13:30 |
ralfv | specifically set bootdev to pxe | 13:30 |
ralfv | ipmitool chassis bootdev pxe | 13:30 |
*** bizarrochristy has quit IRC | 13:30 | |
ralfv | and ipmitool chassis bootparam set bootflag force_pxe | 13:30 |
ralfv | seem to have no effectr | 13:30 |
*** mtanino has joined #openstack-ironic | 13:31 | |
tiagogomes_ | jroll I am using a release made from the 2015.1.0. Perhaps the documentation could be improved saying which version or Ironic supports logical names | 13:33 |
jroll | tiagogomes_: just need to update the client, not the server, I think | 13:33 |
*** amotoki has quit IRC | 13:33 | |
*** boris-42 has joined #openstack-ironic | 13:33 | |
dguerri | jroll: unfortunately, it doesn't say anything about the overriding of a cleaning step :( | 13:35 |
jroll | dguerri: oh, ha, this didn't merge yet | 13:35 |
jroll | I was going to say we just fixed this | 13:35 |
jroll | https://review.openstack.org/#/c/189969/ | 13:35 |
dguerri | yay!! | 13:36 |
jroll | :) | 13:36 |
dguerri | cool | 13:36 |
dguerri | jroll: that's exactly what we need! | 13:36 |
jroll | dguerri: indeed, we just updated onmetal hw manager to work with latest upstream and hit that | 13:37 |
*** e0ne is now known as e0ne_ | 13:37 | |
dguerri | jroll: brilliant. So i am going to abandon my patch... but I have noticed that using 0 as the number of shredding iterations, doesn't do what you might expect | 13:38 |
dguerri | this is more about the shred tool, tho | 13:38 |
dguerri | jroll: thank you very much, btw | 13:38 |
jroll | dguerri: yeah, your patch may also be useful, let me think about it. you can already disable erase_devices so maybe not. idk. | 13:39 |
jroll | and you're welcome :) | 13:39 |
jroll | we have that patch in production so I'm 99% sure it works :P | 13:39 |
*** e0ne_ is now known as e0ne | 13:39 | |
dguerri | great ;) | 13:40 |
jroll | maybe I should review it :P | 13:40 |
jroll | bbiab | 13:40 |
openstackgerrit | Dmitry Tantsur proposed openstack/ironic-inspector: Specify server id clearly on the gate test https://review.openstack.org/199012 | 13:42 |
nagyz_ | will the kilo ironic version out of the box work with a vlan provider network (instead of a flat one)? | 13:47 |
nagyz_ | I'd assume all network orchestration is via neutron | 13:47 |
*** bizarrochristy has joined #openstack-ironic | 13:52 | |
lazy_prince | nagyz_: kilo may not but liberty will.. | 13:54 |
lazy_prince | nagyz_: Look at https://review.openstack.org/#/c/187829/ and https://review.openstack.org/#/c/188528/ | 13:55 |
*** r-daneel has joined #openstack-ironic | 13:56 | |
*** kan__ has quit IRC | 13:57 | |
*** bizarrochristy has quit IRC | 13:58 | |
*** bizarrochristy has joined #openstack-ironic | 13:58 | |
*** cing has quit IRC | 14:00 | |
nagyz_ | right but I can't just simply create a flat network in our current infrastructure :/ | 14:00 |
nagyz_ | based on the review this is only about setting up the ToR/access switches? | 14:01 |
nagyz_ | I could manually go and set all access vlans on the ports to the same VLAN I have the vlan provided network on in neutron | 14:01 |
*** ijw has joined #openstack-ironic | 14:01 | |
nagyz_ | would that work? | 14:01 |
lazy_prince | nagyz_: yes.. but if used with SDN controllers, it could configure everything for you.. | 14:01 |
lazy_prince | nagyz_: yes.. that will do.. | 14:02 |
*** Kennan2 has joined #openstack-ironic | 14:03 | |
*** zhenguo has quit IRC | 14:03 | |
*** Kennan has quit IRC | 14:03 | |
lazy_prince | nagyz_: a vlan provider network should work fine with kilo ironic as long as the vlan provider network has proper routes set for the management network so that I ronic can deploy images.. | 14:04 |
nagyz_ | right, it has glance access | 14:05 |
nagyz_ | using ML2 in L will be better, but for the time being I can live with this ;-) | 14:05 |
nagyz_ | any news about the deployment network separation blueprint? (I know I could search...) | 14:05 |
lazy_prince | nagyz_: true | 14:05 |
nagyz_ | ah that's the same review you've linked (I'm reading the comments) | 14:06 |
lazy_prince | nagyz_: those two review ids are for deployment network sepration along with tenant network isolation.. | 14:06 |
nagyz_ | is it safe to have the same network for baremetal instances and VMs? will the cleaning work nicely? :-) | 14:09 |
openstackgerrit | Merged openstack/ironic: Image service should not be set in ImageCache constructor https://review.openstack.org/194752 | 14:10 |
openstackgerrit | Merged openstack/ironic: Add iRMC Virtual Media Deploy module for iRMC Driver https://review.openstack.org/151958 | 14:11 |
jroll | cleaning shouldn't have any problems, just needs access to ironic etc same as provisioning | 14:12 |
lazy_prince | nagyz_: well.. that is on us to ensure when we implement it.. :) | 14:12 |
nagyz_ | I meant in the current situtation - in K. | 14:13 |
nagyz_ | I already have a neutron network that has a couple hundred VMs but now it would be convenient to also do baremetal in the same network | 14:14 |
jroll | nagyz_: manual vlan setup should work fine for your use case, will need routes to ironic and the rest of the control plane, you'll also need to make sure your images are set up to use that vlan | 14:14 |
jroll | yeah, there's no issues with that | 14:14 |
lazy_prince | nagyz_: and for VMs and baremetal instances to be able to talk to each other, we may have to have them same network.. | 14:14 |
nagyz_ | they are | 14:14 |
nagyz_ | in the same VLAN. | 14:14 |
nagyz_ | I have one VLAN with a /20 IP block | 14:14 |
NobodyCam | good morning Ironicers | 14:14 |
jroll | nagyz_: should just work | 14:15 |
jroll | morning NobodyCam :) | 14:15 |
nagyz_ | jroll, we'll see. :-) | 14:15 |
NobodyCam | morning jroll nagyz_ :) | 14:15 |
jroll | nagyz_: the only requirements are that the metal can access the control plane at provisioning time | 14:16 |
jroll | and cleaning time | 14:16 |
jroll | so if your routes and acls allow for that you should be good. | 14:16 |
lazy_prince | jroll: can you please review https://review.openstack.org/198720 | 14:16 |
jroll | maybe :) | 14:16 |
openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Mock the file creation for the GetConfigdriveTestCase tests https://review.openstack.org/199078 | 14:16 |
jroll | +1 lazy_prince | 14:17 |
lazy_prince | jroll: and probably https://review.openstack.org/#/c/198771/ too.. | 14:18 |
jroll | heh | 14:18 |
jroll | now you're making me work! | 14:18 |
lazy_prince | jroll: :) | 14:18 |
openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Mock the file creation for the GetConfigdriveTestCase tests https://review.openstack.org/199078 | 14:19 |
jroll | lazy_prince: I'll come back to that one, tagged myself. I have a meeting shortly | 14:19 |
lazy_prince | jroll: np.. | 14:19 |
lazy_prince | TheJulia: can you also review https://review.openstack.org/#/c/198771/ This is about enabling IPA for ubuntu and debian | 14:20 |
NobodyCam | morning lazy_prince | 14:21 |
TheJulia | lazy_prince: will do, thank you! | 14:21 |
lazy_prince | morning NobodyCam | 14:21 |
*** itamarl has quit IRC | 14:23 | |
lazy_prince | lucasagomes: had some issue with network in the lab so have not tested it yet.. but my devstack has come up and will be testing it shortly.. | 14:23 |
rcarrillocruz | TheJulia: is it normal that after booting off the CoreOS image I get a user prompt? It then takes 10-20sec, and machine reboots again | 14:23 |
rcarrillocruz | dunno if it's expected or i need to tweak things on my side | 14:23 |
*** ralfv has quit IRC | 14:23 | |
TheJulia | rcarrillocruz: Thats fairly normal | 14:23 |
lucasagomes | lazy_prince, cool, yeah I would say mark the patch as WIP then | 14:23 |
lucasagomes | until you get it sorted | 14:23 |
NobodyCam | morning lucasagomes :) | 14:24 |
rcarrillocruz | k, i need to fix why it doesn't get the default route from configdrive now | 14:24 |
lucasagomes | NobodyCam, jroll good ugt morning! | 14:24 |
jroll | morning lucasagomes! | 14:24 |
rcarrillocruz | i sorted the NIC order issue by disabling network boot on first nic straightaway | 14:24 |
lazy_prince | lucasagomes: marked it as WIP.. | 14:24 |
jroll | and morning TheJulia rcarrillocruz and anyone else I haven't spoke to yet :) | 14:24 |
lucasagomes | lazy_prince, thanks! | 14:24 |
rcarrillocruz | halo jroll, i'm pretty new in this ironic land :-) | 14:25 |
jroll | I know, good morning to you anyway! :) | 14:25 |
jroll | \o/ for new folks | 14:25 |
*** chlong has quit IRC | 14:25 | |
dtantsur | morning jroll, NobodyCam, and the others :) | 14:27 |
jroll | ohai2u dtantsur | 14:27 |
NobodyCam | morning dtantsur and rcarrillocruz :) | 14:27 |
rcarrillocruz | morning :-) | 14:27 |
*** vandorss has joined #openstack-ironic | 14:28 | |
*** vandorss has quit IRC | 14:29 | |
rcarrillocruz | quick question: what are the bare minimum elements for a dib on baremetal? | 14:31 |
rcarrillocruz | by default i see on bifrost playbooks there's vm, cloudinit-datasources and serial-console | 14:31 |
rcarrillocruz | i assume i should only have cloud-init/simple-init ? | 14:31 |
*** ralfvand0 has joined #openstack-ironic | 14:32 | |
lazy_prince | rcarrillocruz: if you do not use vm element then, you may have to use baremetal element | 14:34 |
rcarrillocruz | k, i'll experiment with baremetal+simple-init | 14:35 |
lazy_prince | so while we are on it, does partition image still need kernel and ramdisk to do local boot from disk.. ? | 14:35 |
lazy_prince | i mean does ironic still need them to be uploaded on glance..? | 14:36 |
*** BadCub-away is now known as BadCub | 14:36 | |
jroll | lazy_prince: yes | 14:40 |
jroll | but we also support whole disk images | 14:40 |
lazy_prince | jroll: cool... | 14:40 |
lucasagomes | :-( seriously that tempest passes | 14:42 |
lucasagomes | Body: {"address": "fa:16:3e:12:c2:ca", "extra": {}, "node_uuid": "6e84ebf8-1430-427b-9531-3540ac25205f", "uuid": null} | 14:42 |
lucasagomes | "uuid": null | 14:42 |
jroll | what's that? | 14:42 |
lucasagomes | I will have to make the uuid field nullable for RPC objects | 14:42 |
lucasagomes | jroll, https://review.openstack.org/#/c/198755/ | 14:42 |
jroll | oh! yay | 14:43 |
jroll | ... but tempest isn't passing? | 14:43 |
lucasagomes | yeah due ^ | 14:43 |
lucasagomes | trying to figure now | 14:43 |
jroll | oh, I thought you were sad that tempest passed | 14:43 |
openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Migrate RPC objects to oslo.versionedobjects Fields https://review.openstack.org/198755 | 14:43 |
lucasagomes | no I was sad cause it didn't pass | 14:44 |
jroll | oh hm, port-create right? | 14:44 |
jroll | it doesn't have a uuid until it gets to the conductor | 14:44 |
lucasagomes | before the fields for our RPC objects where very flexible, like str_or_none | 14:44 |
lucasagomes | now we can add checks and stuff | 14:44 |
jroll | right | 14:44 |
lucasagomes | jroll, yeah. So I will make it nullable | 14:44 |
lucasagomes | it was chekcing null is not a valid uuid heh | 14:44 |
jroll | yeah, not ideal though :/ | 14:44 |
lucasagomes | yeah | 14:45 |
lucasagomes | but fair, that's how it was before | 14:45 |
lucasagomes | I just made it a bit more restrict | 14:45 |
jroll | gotcha | 14:46 |
lucasagomes | we are way behing the oslo versioned objects in Ironic, we didn't sync the code while it was in nova | 14:46 |
openstackgerrit | Jarrod Johnson proposed stackforge/pyghmi: Extend and document user_delete in ipmi.command https://review.openstack.org/199135 | 14:46 |
lucasagomes | for quite a while | 14:46 |
jroll | +A on the first patch in that series btw | 14:46 |
jroll | good stuff | 14:46 |
lucasagomes | thanks | 14:46 |
jroll | lucasagomes: on the second in fields.py | 14:47 |
jroll | I wonder if you could use a mixin so you aren't duplicating __call__ everywhere | 14:47 |
jroll | or a metaclass or whatever | 14:47 |
jroll | I should say, I know you can, wondering if it's worth it | 14:48 |
lucasagomes | jroll, ++ totally | 14:48 |
openstackgerrit | Jarrod Johnson proposed stackforge/pyghmi: Extend and document user_delete in ipmi.command https://review.openstack.org/199135 | 14:48 |
openstackgerrit | Merged stackforge/pyghmi: Have session.py work better in Windows https://review.openstack.org/193108 | 14:48 |
jroll | ok, will comment | 14:48 |
lucasagomes | I'm just waiting tempest to pass and also I'm testing it locally to see if everything is fine | 14:48 |
lucasagomes | then I will change it | 14:48 |
lucasagomes | I made the objects callable as a temporary thing, so we can first move to objects | 14:49 |
lucasagomes | then fix the interfaces and remove that callable | 14:49 |
lucasagomes | otherwise you have to do everything at once and the patch will be gigantic | 14:49 |
jroll | yeah sounds good | 14:50 |
lucasagomes | cool thanks for the comment | 14:52 |
jroll | seeing raid_config reminds me | 14:54 |
jroll | we were able to do that migration fully online \o/ | 14:54 |
jroll | I'd love for all migrations to be that way | 14:54 |
*** bradjones|away is now known as bradjones | 14:56 | |
*** yog__ has quit IRC | 14:56 | |
*** achanda has joined #openstack-ironic | 14:56 | |
openstackgerrit | Merged openstack/bifrost: Add README and Troubleshooting updates for SSH keys https://review.openstack.org/198135 | 14:57 |
openstackgerrit | Merged openstack/bifrost: Update VM testing steps https://review.openstack.org/197694 | 14:57 |
lucasagomes | nice! | 14:58 |
*** mgoddard has quit IRC | 14:59 | |
*** ifarkas has quit IRC | 15:00 | |
*** mgoddard has joined #openstack-ironic | 15:00 | |
*** ifarkas has joined #openstack-ironic | 15:00 | |
*** achanda has quit IRC | 15:02 | |
openstackgerrit | Merged openstack/bifrost: Add libvirt package dependency to install step https://review.openstack.org/198932 | 15:04 |
*** ijw has quit IRC | 15:06 | |
*** puranamr has joined #openstack-ironic | 15:07 | |
*** dasm has quit IRC | 15:15 | |
*** e0ne is now known as e0ne_ | 15:16 | |
*** ralfvand0_ has joined #openstack-ironic | 15:16 | |
*** romcheg has quit IRC | 15:19 | |
*** ralfvand0 has quit IRC | 15:20 | |
*** ralfvand0_ has quit IRC | 15:23 | |
*** rameshg87 has joined #openstack-ironic | 15:23 | |
*** ralfvand0_ has joined #openstack-ironic | 15:23 | |
rameshg87 | rloo: hi | 15:25 |
rloo | hi rameshg87, i was looking for you | 15:25 |
rameshg87 | oh :) | 15:25 |
rloo | rameshg87: did you see my comments. on some review i just did... | 15:25 |
rameshg87 | rloo: wanted to check regarding https://review.openstack.org/#/c/193436/7 | 15:25 |
rameshg87 | rloo: yeah, I was about to ping that only | 15:25 |
rameshg87 | rloo: I agree backward compatibility is at risk here | 15:26 |
*** bizarrochristy has quit IRC | 15:26 | |
rloo | rameshg87: yeah, that one. | 15:26 |
rameshg87 | rloo: deploy with older proliantutils will fail if no virtual media is attached | 15:26 |
rloo | rameshg87: do you know if we/you need to add code to handle older proliantutils, or are you good with it barfing/failing | 15:26 |
*** e0ne_ is now known as e0ne | 15:26 | |
NobodyCam | morning rloo rameshg87 :) | 15:26 |
rameshg87 | morning NobodyCam | 15:26 |
NobodyCam | :) | 15:26 |
rloo | morning NobodyCam | 15:26 |
NobodyCam | :) | 15:26 |
*** rameshg871 has joined #openstack-ironic | 15:27 | |
*** rameshg87 has quit IRC | 15:27 | |
*** shalq has quit IRC | 15:27 | |
*** bizarrochristy has joined #openstack-ironic | 15:27 | |
*** shalq has joined #openstack-ironic | 15:28 | |
rameshg871 | rloo: but otherwise I am sure I will annoy people with that error msg :( | 15:28 |
rloo | rameshg87: honestly, i'm fine with it barfing but i can't recall if we have a 'way' of handling those | 15:28 |
*** rameshg871 is now known as rameshg87 | 15:28 | |
rameshg87 | rloo: the earlier proliantutils will throw exception if virtual media is not attached | 15:29 |
rameshg87 | rloo: so in newer code, we make sure we eject virtual media before beginning deploy | 15:29 |
rameshg87 | rloo: so that itself will fail | 15:29 |
*** romcheg has joined #openstack-ironic | 15:29 | |
* rameshg87 is confused | 15:29 | |
rloo | rameshg87: so with newer code, when you try to eject, the old proliantutils will throw exception? | 15:29 |
*** rameshg87 has quit IRC | 15:30 | |
*** ijw has joined #openstack-ironic | 15:31 | |
*** rameshg871 has joined #openstack-ironic | 15:32 | |
* rameshg871 is frustrated with connectivity issue | 15:32 | |
rameshg871 | rloo: but if/when people update ironic, I guess they would be caring to update it's dependencies as well | 15:32 |
rloo | rameshg871: so you're ok with the current / new behavior, when using older proliantutils? | 15:32 |
rameshg871 | rloo: yes .. | 15:33 |
rloo | rameshg871: ok. so you want to add a followon patch to address my other comments? | 15:33 |
rameshg871 | rloo: yeah, I will do right away .. | 15:34 |
rloo | rameshg871: ok | 15:34 |
rameshg871 | rloo: oh soon I am planning to do one more change that will break anyway .. | 15:34 |
rameshg871 | rloo: 2.1.1 supports giving cacert as well, so planning to add a new argument here - https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/ilo/common.py#L155-L159 | 15:35 |
rameshg871 | rloo: cacert=xxxx | 15:35 |
rameshg871 | rloo: so that will break it anyway if they don't update to 2.1.1 | 15:35 |
rloo | rameshg871: break away | 15:35 |
rameshg871 | :) | 15:36 |
*** rameshg871 is now known as rameshg87 | 15:36 | |
*** davidlenwell has quit IRC | 15:38 | |
*** saripurigopi has quit IRC | 15:40 | |
openstackgerrit | Julia Kreger proposed openstack/bifrost: Update README for simple-init/glean https://review.openstack.org/198142 | 15:42 |
*** praneshp has joined #openstack-ironic | 15:43 | |
*** Nisha has joined #openstack-ironic | 15:46 | |
*** praneshp_ has joined #openstack-ironic | 15:48 | |
openstackgerrit | Ramakrishnan G proposed openstack/ironic: Address follow-up comments https://review.openstack.org/199163 | 15:48 |
rameshg87 | rloo: ^^ | 15:48 |
*** praneshp has quit IRC | 15:49 | |
*** praneshp_ is now known as praneshp | 15:49 | |
rloo | thx rameshg87 | 15:50 |
openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Mock the file creation for the GetConfigdriveTestCase tests https://review.openstack.org/199078 | 15:52 |
lucasagomes | rloo, good ugt morning | 15:52 |
* lucasagomes gotta love python 3 which has no file() builtin type | 15:52 | |
rloo | hi lucasagomes | 15:52 |
*** ijw has quit IRC | 15:53 | |
*** davidlenwell has joined #openstack-ironic | 15:53 | |
*** davidlenwell is now known as daviddavid | 15:54 | |
*** daviddavid is now known as davidlenwell | 15:54 | |
*** UForgotten has quit IRC | 15:55 | |
*** davidlenwell has quit IRC | 15:58 | |
*** davidlenwell has joined #openstack-ironic | 15:58 | |
*** jistr has quit IRC | 16:00 | |
*** mgarza_ has joined #openstack-ironic | 16:00 | |
*** romcheg1 has joined #openstack-ironic | 16:03 | |
*** romcheg has quit IRC | 16:04 | |
*** dtantsur is now known as dtantsur|afk | 16:04 | |
*** jistr has joined #openstack-ironic | 16:07 | |
*** davidlenwell has quit IRC | 16:08 | |
*** davidlenwell has joined #openstack-ironic | 16:08 | |
*** davidlenwell is now known as omnibus7 | 16:12 | |
*** omnibus7 is now known as davidlenwell | 16:12 | |
*** davidlenwell is now known as david_enwell | 16:12 | |
tiagogomes_ | Hi, I am trying to setup nova to schedule baremetal servers using Ironic, but I am getting a "No valid host was found" error. The flavor matches the node extra properties. On the compute log there is an "No free port available for instance for instance XYZ" | 16:14 |
tiagogomes_ | Anyone clue about this? Everytime I try to use Nova with Ironic I run into trouble :P | 16:14 |
*** absubram has joined #openstack-ironic | 16:16 | |
jroll | sounds like it couldn't allocate a port in neutron | 16:16 |
* rameshg87 goes to sleep | 16:17 | |
rameshg87 | goodnight folks | 16:18 |
*** rameshg87 has quit IRC | 16:18 | |
*** romcheg1 has quit IRC | 16:18 | |
tiagogomes_ | creating it manually works | 16:19 |
*** david_enwell is now known as davidlenwell | 16:19 | |
*** davidlenwell has quit IRC | 16:20 | |
openstackgerrit | Sam Betts proposed openstack/ironic-inspector: Allow addition of hooks without overriding defaults https://review.openstack.org/199179 | 16:20 |
*** davidlenwell has joined #openstack-ironic | 16:20 | |
sambetts | night all | 16:23 |
*** davidlenwell has quit IRC | 16:23 | |
*** davidlenwell has joined #openstack-ironic | 16:24 | |
*** davidlenwell has quit IRC | 16:27 | |
*** davidlenwell has joined #openstack-ironic | 16:28 | |
NobodyCam | night sambetts | 16:28 |
rcarrillocruz | bleh | 16:28 |
*** jcoufal has quit IRC | 16:28 | |
tiagogomes_ | is there a particular filter that I should use in Nova? | 16:29 |
rcarrillocruz | can't get network set up properly | 16:29 |
rcarrillocruz | brb | 16:29 |
tiagogomes_ | this is my configuration: http://paste.openstack.org/show/352438/ | 16:31 |
*** davidlenwell has quit IRC | 16:31 | |
*** e0ne has quit IRC | 16:33 | |
*** jistr has quit IRC | 16:35 | |
*** davidlenwell has joined #openstack-ironic | 16:35 | |
TheJulia | rcarrillocruz: how so? | 16:35 |
*** davidlenwell is now known as davidlenwell_ | 16:36 | |
*** davidlenwell_ is now known as davidlenwell | 16:38 | |
openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Migrate RPC objects to oslo.versionedobjects Fields https://review.openstack.org/198755 | 16:39 |
*** romcheg has joined #openstack-ironic | 16:42 | |
*** itamarl has joined #openstack-ironic | 16:46 | |
rcarrillocruz | TheJulia: when I restart from coreOS, the cloud-init interfaces info shows the eth0/eth1 as down | 16:47 |
rcarrillocruz | thus they don't get the network config | 16:47 |
rcarrillocruz | not sure if it's a race condition where it can't be set up cos they NICs are not up by the time or what | 16:48 |
TheJulia | rcarrillocruz: try adding simple-init to your list of diskimage-builder elements | 16:49 |
TheJulia | and rm -r -f /tftpboot/master_images | 16:49 |
* lazy_prince goes to bed | 16:49 | |
TheJulia | (although I think that fix landed already) | 16:49 |
lazy_prince | bye everyone.. | 16:50 |
TheJulia | goodnight lazy_prince | 16:50 |
rcarrillocruz | master images huh | 16:50 |
rcarrillocruz | ok | 16:50 |
rcarrillocruz | i already gave it a try to simple-init | 16:50 |
rcarrillocruz | i removed cloud-init for it: | 16:50 |
rcarrillocruz | shell: disk-image-create -a amd64 -o "{{http_boot_folder}}/{{deploy_image_filename}}" -t qcow2 "{{dib_os_element}}" vm simple-init "{{ extra_dib_elements}}" | 16:50 |
rcarrillocruz | that from the create_dib_image.yml file | 16:50 |
TheJulia | burn in devuser as well so you can take a look at what it's doing. Keep in mind you'll need to set the environment variable for a password for devuser | 16:51 |
TheJulia | There may just be something funky going on that we can't account for or know about | 16:51 |
TheJulia | fwiw... networking has caused the most headaches since it extremely variable | 16:51 |
lucasagomes | folks I will call it a day | 16:52 |
TheJulia | goodnight lucasagomes | 16:52 |
lucasagomes | have a good night everyone! | 16:52 |
lucasagomes | TheJulia, you too! g'nigh | 16:52 |
*** lucasagomes is now known as lucas-dinner | 16:52 | |
rcarrillocruz | TheJulia: devuser = a system user with a normal password I can set via envvar ? | 16:53 |
TheJulia | rcarrillocruz: yes | 16:53 |
rcarrillocruz | excellent! | 16:53 |
TheJulia | hold on | 16:53 |
rcarrillocruz | i was looking for that :D, since i can't use ssh keys without network | 16:53 |
TheJulia | https://github.com/openstack/diskimage-builder/tree/master/elements/devuser | 16:53 |
rcarrillocruz | ++ | 16:53 |
*** Pradip has joined #openstack-ironic | 16:53 | |
TheJulia | In the environment variables that are passed in to disk-image-create, you'll want to add an entry for DIB_DEV_USER_PASSWORD | 16:54 |
* TheJulia ponders if there needs to be a "the computer does not like me mode" | 16:54 | |
jroll | TheJulia: do computers ever like us? :) | 16:55 |
TheJulia | jroll: not really | 16:55 |
*** lazy_prince has quit IRC | 16:55 | |
TheJulia | Hi, Virtualbox, just because I'm not interacting with the gui does not mean that you can put the VM to sleep | 16:55 |
cinerama | probably retribution for being asked to explain love and subsequently blowing up in all those star trek episodes | 16:55 |
*** davideagnello has joined #openstack-ironic | 16:56 | |
NobodyCam | brb | 16:57 |
jroll | heh | 16:57 |
*** trown is now known as trown|lunch | 17:00 | |
openstackgerrit | Harshada Mangesh Kakad proposed openstack/ironic: Implement get_boot_device for seamicro driver https://review.openstack.org/165788 | 17:04 |
*** e0ne has joined #openstack-ironic | 17:09 | |
*** derekh has quit IRC | 17:12 | |
*** ijw has joined #openstack-ironic | 17:15 | |
*** itamarl has quit IRC | 17:17 | |
*** e0ne is now known as e0ne_ | 17:18 | |
*** david-lyle has quit IRC | 17:19 | |
*** romcheg1 has joined #openstack-ironic | 17:19 | |
*** praneshp has quit IRC | 17:20 | |
*** romcheg has quit IRC | 17:21 | |
*** e0ne_ is now known as e0ne | 17:21 | |
*** praneshp has joined #openstack-ironic | 17:23 | |
*** romcheg1 has quit IRC | 17:23 | |
*** ijw has quit IRC | 17:24 | |
*** romcheg has joined #openstack-ironic | 17:24 | |
*** achanda has joined #openstack-ironic | 17:25 | |
*** e0ne is now known as e0ne_ | 17:26 | |
*** e0ne_ is now known as e0ne | 17:27 | |
openstackgerrit | Julia Kreger proposed openstack/bifrost: Variablize the full iPXE binary https://review.openstack.org/199190 | 17:28 |
*** e0ne is now known as e0ne_ | 17:33 | |
*** praneshp has quit IRC | 17:35 | |
*** e0ne_ is now known as e0ne | 17:39 | |
*** pcrews has quit IRC | 17:46 | |
*** dguerri is now known as dguerri` | 17:47 | |
openstackgerrit | Merged openstack/bifrost: Fix check for >= Ubuntu 14.10 https://review.openstack.org/198943 | 17:49 |
* TheJulia wonders if she is going to have merge conflicts in gerrit | 17:50 | |
*** blakec has quit IRC | 17:51 | |
*** athomas has quit IRC | 17:52 | |
*** max_lobur has quit IRC | 17:55 | |
openstackgerrit | Merged stackforge/pyghmi: Extend and document user_delete in ipmi.command https://review.openstack.org/199135 | 17:57 |
*** trown|lunch is now known as trown | 17:59 | |
*** mgarza_ has quit IRC | 18:00 | |
*** marzif_ has joined #openstack-ironic | 18:03 | |
*** coolsvap|away is now known as coolsvap | 18:04 | |
*** mgarza_ has joined #openstack-ironic | 18:06 | |
*** romcheg has quit IRC | 18:08 | |
*** praneshp has joined #openstack-ironic | 18:13 | |
jroll | bug #1461937 | 18:16 |
openstack | bug 1461937 in Ironic "Node stuck on DEPLOYING (potentially all *ING) state(s)" [High,In progress] https://launchpad.net/bugs/1461937 - Assigned to Lucas Alvares Gomes (lucasagomes) | 18:16 |
*** e0ne is now known as e0ne_ | 18:17 | |
jroll | soooo, question | 18:20 |
*** zz_natorious is now known as natorious | 18:20 | |
jroll | does anyone think it's wrong that nodes are in DEPLOYING while waiting for agents to complete write_image command or whatever? | 18:20 |
jroll | should those be in DEPLOYWAIT? | 18:20 |
*** bizarrochristy has quit IRC | 18:20 | |
*** marzif_ has quit IRC | 18:21 | |
jroll | the current status of them being in DEPLOYING without a lock kind of breaks some assumptions in our code | 18:21 |
jroll | devananda: ^ thoughts? | 18:21 |
rloo | jroll: you mean the actual agent, not the agent driver? yeah, that seems odd. | 18:21 |
jroll | rloo: I mean the node provision state | 18:21 |
jroll | "in our code" being in ironic | 18:21 |
jroll | for example https://review.openstack.org/#/c/194132/ fails any ongoing agent deploys whenever a conductor starts :( | 18:21 |
rloo | jroll: provision state == DEPLOYING, while waiting for agent (not agent driver) to do something, seems odd. | 18:22 |
*** bizarrochristy has joined #openstack-ironic | 18:22 | |
jroll | rloo: I guess so. idk. | 18:22 |
jroll | I think my real question is "should a node ever be in DEPLOYING without a lock" | 18:22 |
rloo | jroll: i don't think that's what lucas' patch means. I mean, i don't think in 'normal' case, DEPLOYIING doesn't have a lock | 18:24 |
jroll | rloo: it does. in the agent driver. | 18:24 |
jroll | that's how it's been since the first patch landed | 18:25 |
rloo | jroll: i think in that patch, the conductor died, a conductor takes over or raises from the dead, and we have code to remove locks when the conductor wakes up. | 18:25 |
rloo | jroll: oh, i suspect that is different than what that patch was meant to do. sec, need to look. | 18:25 |
jroll | rloo: the code to remove locks is already there. that patch assumes anything in DEPLOYING that has no lock was locked by that conductor and fails it. | 18:25 |
jroll | rloo: thinking if I just move that block above the "clear the locks" code, it should be fine | 18:26 |
rloo | jroll: looking at line 272: https://review.openstack.org/#/c/194132/13/ironic/conductor/manager.py | 18:26 |
rloo | jroll: you're saying that in the agentdriver code, if you fix, then that line 272 doesn't need to do anything? | 18:27 |
jroll | no | 18:27 |
jroll | rloo: so look at it this way | 18:27 |
jroll | rloo: this code fails any node in DEPLOYING that doesn't have a lock | 18:27 |
jroll | rloo: it does that after the conductor clears its own locks | 18:28 |
jroll | rloo: thus it assumes that anything in DEPLOYING must have a lock, and so only that conductor's deploys are failed. | 18:28 |
rloo | jroll: right | 18:28 |
jroll | is how I'm reading this, anyway | 18:28 |
jroll | so I'm thinking if we fail anything in DEPLOYING and locked by *this* conductor; then remove the locks; that will work for agent and iscsi | 18:29 |
jroll | but I'm not 100% sure. | 18:29 |
jroll | I do know it's a bug, though, because it just failed an ongoing deploy in my staging environment that should not have been failed :) | 18:29 |
rloo | jroll: hmm. so yeah, i think wrt lucas' patch, the logic/sense would be what you said, check for DEPLOYING and locked first, do the error handling, then clear all locks. | 18:30 |
jroll | rloo: yeah, that's what I'm thinking, I'll need to play with it a bit | 18:31 |
jroll | rloo: and then the philosophical question of if we should switch the node to be in DEPLOYWAIT while waiting for the agent to do a thing :) | 18:31 |
rloo | jroll: i think the assumption is that a node cannot be in DEPLOYING with no lock. Are you saying that assumption is incorrect? | 18:31 |
jroll | rloo: yes, it's very easy to demonstrate | 18:32 |
*** Nisha has quit IRC | 18:32 | |
*** natorious is now known as zz_natorious | 18:33 | |
jroll | I also love that this doesn't log at all :( | 18:33 |
*** Nisha has joined #openstack-ironic | 18:33 | |
rloo | jroll: if that's the case, then this patch isn't quite right. | 18:33 |
rloo | jroll: and/or we should enforce that assumption. | 18:33 |
jroll | rloo: right, agree. going to ponder it more | 18:33 |
*** Nisha has quit IRC | 18:34 | |
*** Nisha has joined #openstack-ironic | 18:34 | |
*** pcrews has joined #openstack-ironic | 18:34 | |
*** mgarza_ has quit IRC | 18:38 | |
*** e0ne_ is now known as e0ne | 18:39 | |
*** Nisha has quit IRC | 18:41 | |
*** Nisha has joined #openstack-ironic | 18:42 | |
*** mgarza_ has joined #openstack-ironic | 18:45 | |
*** Sukhdev has joined #openstack-ironic | 18:45 | |
*** pcrews has quit IRC | 18:50 | |
*** e0ne has quit IRC | 18:50 | |
*** achanda has quit IRC | 18:51 | |
*** romcheg has joined #openstack-ironic | 18:52 | |
*** zz_natorious is now known as natorious | 18:56 | |
*** natorious is now known as zz_natorious | 18:57 | |
*** harlowja_ has joined #openstack-ironic | 19:02 | |
*** harlowja has quit IRC | 19:03 | |
*** Nisha has quit IRC | 19:05 | |
*** Nisha has joined #openstack-ironic | 19:06 | |
*** max_lobur has joined #openstack-ironic | 19:08 | |
openstackgerrit | Julia Kreger proposed openstack/bifrost: Fixing deployment image default for install role https://review.openstack.org/199224 | 19:08 |
*** Nisha has quit IRC | 19:08 | |
*** Nisha has joined #openstack-ironic | 19:09 | |
*** puranamr has quit IRC | 19:10 | |
rloo | jroll: somewhat related to conductor/DEPLOYING. (I'm not going to look at this patch): https://review.openstack.org/#/c/197141/ | 19:12 |
*** Nisha has quit IRC | 19:13 | |
jroll | rloo: hmm, thanks | 19:13 |
*** Nisha has joined #openstack-ironic | 19:14 | |
rloo | NobodyCam, devananda, anyone that can update BPs etc: https://blueprints.launchpad.net/ironic/+spec/ipxe-dynamic-config | 19:14 |
*** mgarza_ has quit IRC | 19:14 | |
rloo | NobodyCam, devananda: that BP should be approved cuz the spec has been approved: http://specs.openstack.org/openstack/ironic-specs/specs/liberty/ipxe-dynamic-config.html | 19:14 |
rloo | NobodyCam, devananda: and please update the URL spec link in the BP. Thx | 19:15 |
rloo | BadCub: ^^ if you can chase someone down to do the above, or if you have the powers? :) | 19:15 |
* BadCub looks | 19:16 | |
BadCub | rloo: ^^ approved and set to L1 | 19:17 |
rloo | BadCub: thank you! But don't set it to L1. we aren't using those (and L1 has passed) | 19:17 |
BadCub | rlo | 19:17 |
BadCub | thanks for the reminder. Old habit lol | 19:17 |
BadCub | fixed | 19:18 |
rloo | BadCub: it doesn't really matter although it might confuse people if they think we are using those milestones. dunno. | 19:18 |
lucas-dinner | jroll, oh I thought it would be always locked when DPELOYING | 19:18 |
lucas-dinner | DEPLOYING* | 19:18 |
lucas-dinner | jroll, I mean, usually I see DEPLOYING == conductor is doing something | 19:19 |
BadCub | rloo: indeed it may | 19:19 |
lucas-dinner | DEPLOYWAIT == ramdisk doing something | 19:19 |
lucas-dinner | like when the ramdisk is installing the bootloader for localboot we move the state back to DEPLOYWAIT | 19:19 |
jroll | lucas-dinner: yeah, I see why someone might make that assumption but it's wrong for the agent driver | 19:19 |
lucas-dinner | may be a wrong assumption of my part then | 19:20 |
*** ndipanov has quit IRC | 19:20 | |
lucas-dinner | jroll, I see | 19:20 |
jroll | and maybe the agent driver is broken, but that's how it is right now | 19:20 |
*** pcrews has joined #openstack-ironic | 19:20 | |
lucas-dinner | jroll, gotcha, feel free to revert that patch then until we fix it | 19:20 |
lucas-dinner | jroll, or if u want tomorrow I can take a look at the agent and fix it | 19:20 |
jroll | lucas-dinner: I'll probably hack on things this afternoon and go from there | 19:20 |
jroll | I'll let you know before I leave | 19:20 |
lucas-dinner | to move to DEPLOYWAIT once the conductor is operating | 19:20 |
lucas-dinner | jroll, ack | 19:20 |
lucas-dinner | it's maybe good to put on DEPLOYWAIT because that's abortable as well | 19:21 |
lucas-dinner | via nova's api | 19:21 |
lucas-dinner | jroll, yeah update the bug with whatever you find, so we can fix this inconsistency | 19:21 |
jroll | lucas-dinner: yep will do | 19:22 |
lucas-dinner | cool thanks! | 19:22 |
jroll | lucas-dinner: off hand... | 19:22 |
jroll | do you know if DEPLOYWAIT -> DELETE goes through cleaning? | 19:22 |
jroll | I can dive into code if you don't know from the top of your head | 19:22 |
JoshNang | jroll: almost certainly would | 19:23 |
jroll | cool | 19:23 |
lucas-dinner | jroll, yes it should go to DEPLOYWAIT -> DELETING | 19:23 |
*** mgarza has joined #openstack-ironic | 19:24 | |
lucas-dinner | and then cleaning and so on | 19:24 |
jroll | lucas-dinner: awesome, ty sir. I'll let you know what I end up with | 19:24 |
jroll | stepping away for a few | 19:24 |
lucas-dinner | cool! | 19:25 |
*** Nisha has quit IRC | 19:25 | |
*** Nisha has joined #openstack-ironic | 19:25 | |
*** Nisha has quit IRC | 19:30 | |
*** bizarrochristy has quit IRC | 19:35 | |
*** achanda has joined #openstack-ironic | 19:37 | |
openstackgerrit | Merged openstack/ironic: Initial oslo.versionedobjects conversion https://review.openstack.org/198687 | 19:42 |
rcarrillocruz | oh my | 19:43 |
rcarrillocruz | TheJulia: i think my issue could be a wrong netmask | 19:43 |
rcarrillocruz | cos it's wronly set and the hosts portion is smaller as it should be, the machine can't reach the gateway, and that may be the reason why cloud-init does not set it up | 19:44 |
rcarrillocruz | tearing down everything and redeploying | 19:44 |
*** bizarrochristy has joined #openstack-ironic | 19:44 | |
TheJulia | interesting... | 19:45 |
TheJulia | That would do it :( You could do dhcp with reservations or something along those lines if your using an external dhcp server. | 19:47 |
*** david-lyle has joined #openstack-ironic | 19:50 | |
rcarrillocruz | sticking with dnsmasq | 19:53 |
*** e0ne has joined #openstack-ironic | 19:54 | |
*** max_lobur has quit IRC | 19:54 | |
rcarrillocruz | hmm, now it's not set even the IP of the management interface | 19:57 |
rcarrillocruz | just keeps the prov interface IP it got from dnsmasq | 19:57 |
rcarrillocruz | hmm | 19:57 |
rcarrillocruz | i hate lack of consistency :-) | 19:57 |
rcarrillocruz | oh | 19:58 |
rcarrillocruz | excellent , cloud-init logs available | 19:58 |
rcarrillocruz | let's see | 19:58 |
*** mgarza has quit IRC | 20:00 | |
*** mgarza_ has joined #openstack-ironic | 20:03 | |
*** puranamr has joined #openstack-ironic | 20:04 | |
openstackgerrit | Arun S A G proposed openstack/ironic-python-agent: Fix iso-image-create to work on RedHat distros https://review.openstack.org/199237 | 20:05 |
*** pelix has quit IRC | 20:05 | |
*** klaus__ has joined #openstack-ironic | 20:05 | |
*** chlong has joined #openstack-ironic | 20:06 | |
*** klaus__ has quit IRC | 20:06 | |
*** marzif_ has joined #openstack-ironic | 20:07 | |
*** lucas-dinner has quit IRC | 20:08 | |
*** puranamr has quit IRC | 20:15 | |
*** marzif__ has joined #openstack-ironic | 20:16 | |
openstackgerrit | Merged openstack/ironic: Clear ilo_boot_iso before deploy for glance images https://review.openstack.org/193436 | 20:18 |
TheJulia | rcarrillocruz: seems like we need to get you in a place where you can pass a robust static configuration in :) | 20:18 |
*** ijw has joined #openstack-ironic | 20:18 | |
*** marzif_ has quit IRC | 20:18 | |
rcarrillocruz | hah | 20:18 |
*** puranamr has joined #openstack-ironic | 20:19 | |
rcarrillocruz | huh | 20:20 |
rcarrillocruz | redeploy again | 20:20 |
rcarrillocruz | one blade with correct IP on em1 | 20:20 |
rcarrillocruz | the other one without it | 20:20 |
rcarrillocruz | and default gw going thru prov device | 20:20 |
rcarrillocruz | that btw is set to do so anywhere on my vars | 20:20 |
* rcarrillocruz shakes fist at cloud-init | 20:21 | |
rcarrillocruz | or simple-init in this case | 20:21 |
*** lucasagomes has joined #openstack-ironic | 20:21 | |
TheJulia | so simple-init is much more difficult to troubleshoot, there is a case where if the interface does not get carrier up in under a second, it won't mark the interface as connected. I think yesterday they pushed a release that changed that to 5 seconds | 20:22 |
*** penick has joined #openstack-ironic | 20:22 | |
TheJulia | rcarrillocruz: seems odd to get inconsistent results across two nodes from the same deploy... | 20:22 |
TheJulia | The whole idea is "I want to stamp out many machines... very quickly" | 20:23 |
TheJulia | s/is/was/ | 20:23 |
rcarrillocruz | driving me nuts yeah | 20:23 |
rcarrillocruz | so | 20:23 |
rcarrillocruz | baremetal.json | 20:23 |
rcarrillocruz | the nics section | 20:23 |
rcarrillocruz | i only have defined there the prov interface mac | 20:23 |
rcarrillocruz | should I set th emanagement interface as well? | 20:23 |
rcarrillocruz | how is this data used in the playbooks? | 20:24 |
* rcarrillocruz greps code | 20:24 | |
*** bizarrochristy has quit IRC | 20:24 | |
TheJulia | rcarrillocruz: You can give it a shot, I've not mucked with mulitple operating interfaces at the same time. My thoguht was somebody would use it to get things minimally online on whatever build network they were on and then use whatever configuration management tools they wanted to use to assert their full config into place. | 20:24 |
* TheJulia wonders where she put a spare dual port gigabit ethernet card.... | 20:25 | |
rcarrillocruz | yeah, i just want ironic to config my mangement interface, since it's what gives me outside world access | 20:25 |
rcarrillocruz | i'll change to it and inspect how it behaves | 20:25 |
*** marzif__ has quit IRC | 20:26 | |
*** lucasagomes has quit IRC | 20:27 | |
TheJulia | rcarrillocruz: whatever mac you define for the first port on the node is what ends up getting assigned a static ip... if thats what you want. There is a setting to turn off static assignment and just do dhcp | 20:28 |
rcarrillocruz | yeah, that's what i want | 20:31 |
rcarrillocruz | i'm double checking macs if they are right to what i want | 20:31 |
*** dprince has quit IRC | 20:34 | |
*** bizarrochristy has joined #openstack-ironic | 20:34 | |
*** coolsvap is now known as coolsvap|away | 20:39 | |
*** jamielennox is now known as jamielennox|away | 20:41 | |
*** e0ne has quit IRC | 20:44 | |
*** Sukhdev has quit IRC | 20:45 | |
*** jamielennox|away is now known as jamielennox | 20:52 | |
*** igordcard_ has joined #openstack-ironic | 20:56 | |
rcarrillocruz | o-k | 20:57 |
rcarrillocruz | i think this is the first time i get the two test blades deployed correctly with all net config | 20:57 |
rcarrillocruz | perfect time to go to dinner and bed | 20:57 |
rcarrillocruz | love leaving when i am winning! | 20:57 |
rcarrillocruz | have a good one TheJulia and all | 20:57 |
*** marzif__ has joined #openstack-ironic | 20:58 | |
TheJulia | goodnight :) | 20:59 |
*** puranamr has quit IRC | 21:01 | |
*** puranamr has joined #openstack-ironic | 21:03 | |
*** kkoski has quit IRC | 21:04 | |
*** trown is now known as trown|outttypeww | 21:12 | |
*** marzif_ has joined #openstack-ironic | 21:26 | |
*** marzif__ has quit IRC | 21:26 | |
*** marzif_ has quit IRC | 21:32 | |
openstackgerrit | Laura Moore proposed openstack/ironic-specs: Update of the Ironic ML2 Integration spec https://review.openstack.org/188528 | 21:40 |
*** max_lobur has joined #openstack-ironic | 21:45 | |
*** Pradip has quit IRC | 21:54 | |
*** bizarrochristy has quit IRC | 21:55 | |
*** radek__ has quit IRC | 21:55 | |
*** romcheg1 has joined #openstack-ironic | 21:56 | |
*** romcheg has quit IRC | 21:57 | |
*** mgarza_ has quit IRC | 21:57 | |
*** bizarrochristy has joined #openstack-ironic | 21:57 | |
*** mgarza has joined #openstack-ironic | 21:59 | |
*** thrash is now known as thrash|g0ne | 22:12 | |
mrda | Morning Ironic | 22:15 |
*** ijw_ has joined #openstack-ironic | 22:16 | |
*** ijw has quit IRC | 22:19 | |
NobodyCam | goot en morgan mrda | 22:20 |
mrda | Danke, guten morgen | 22:21 |
mrda | NobodyCam: ^ | 22:21 |
NobodyCam | :) | 22:21 |
openstackgerrit | Stephanie Miller proposed openstack/bifrost: Move diskimage-builder and bootable image creation to roles https://review.openstack.org/199329 | 22:24 |
NobodyCam | Ow ya goin mate | 22:25 |
openstackgerrit | Merged openstack/ironic: Add localboot support for iscsi_irmc driver https://review.openstack.org/163738 | 22:25 |
openstackgerrit | Merged openstack/ironic: Add whole disk image support for iscsi_irmc driver https://review.openstack.org/181771 | 22:25 |
*** puranamr has quit IRC | 22:28 | |
*** puranamr has joined #openstack-ironic | 22:29 | |
openstackgerrit | Stephanie Miller proposed openstack/bifrost: Move diskimage-builder and bootable image creation to roles https://review.openstack.org/199329 | 22:35 |
*** Sukhdev has joined #openstack-ironic | 22:45 | |
*** chlong has quit IRC | 22:46 | |
*** ijw_ has quit IRC | 22:56 | |
*** ijw has joined #openstack-ironic | 22:57 | |
*** bizarrochristy has quit IRC | 22:57 | |
*** r-daneel has quit IRC | 23:00 | |
*** max_lobur has quit IRC | 23:01 | |
*** bizarrochristy has joined #openstack-ironic | 23:04 | |
*** puranamr has quit IRC | 23:09 | |
*** absubram has quit IRC | 23:11 | |
*** zhenguo has joined #openstack-ironic | 23:14 | |
openstackgerrit | Ruby Loo proposed openstack/python-ironicclient: Expose node's clean_step and bump default version https://review.openstack.org/199340 | 23:14 |
*** praneshp has quit IRC | 23:26 | |
*** krtaylor has quit IRC | 23:29 | |
*** romcheg1 has quit IRC | 23:29 | |
*** ijw has quit IRC | 23:31 | |
*** ijw has joined #openstack-ironic | 23:33 | |
*** krtaylor has joined #openstack-ironic | 23:42 | |
*** yuanying_ has joined #openstack-ironic | 23:53 | |
*** yuanying has quit IRC | 23:56 | |
*** rloo has quit IRC | 23:56 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!