TheJulia | JayF: if memory serves, it is just a matter of powering the machine on the same network | 01:18 |
---|---|---|
TheJulia | since it will shunt to introspection | 01:18 |
TheJulia | at which point, autodiscovery takes over, and the normal inspection node creation takes over | 01:20 |
JayF | cc: awb | 01:44 |
JayF | so it sounds like there is no docs because ... it just works? | 01:44 |
TheJulia | Indeed | 01:45 |
JayF | awb: https://docs.openstack.org/ironic-inspector/2023.1/user/usage.html#discovery seems like normal ironic-inspectr docs would apply then | 01:46 |
JayF | that should help | 01:46 |
awb | Thanks for the mention! I have a stack of about ten machines that are only on the IPMI network right now and will be wired up later. It should be possible to use Ironic to get a Debian image on those, right? | 03:35 |
TheJulia | awb: you will need the BMC and some other network interface wired up. The machine will boot over the other network interface and we will try to extract the BMC address. You will need to set a username/password for the BMC for ironic to be able to manage the nodes | 04:14 |
TheJulia | awb: that being said, if the BMCs are set for dhcp, you may get 0.0.0.0 out of the BMCs in terms of the driver_info/ipmi_address which gets set. Depends on the BMC, really. | 04:15 |
* TheJulia resumes vacation | 04:25 | |
awb | I have DHCP already for the BMC/IPMI interfaces and can get a remote console. One of my machines is wired up for the internal network, the other ones are not yet. These are SuperMicros with dual on-board NICs and a separate BMC port. | 05:39 |
TheJulia | So I think they return 0.0.0.0 if they are not statically assigned on super micro gear | 05:40 |
TheJulia | Ugh wife is now glaring, closing irc | 05:40 |
awb | Interesting, thanks! And yes, back to vacation :D | 05:40 |
dtantsur | JayF, awb, I think discovery must be enabled on bifrost | 06:19 |
awb | I would be happy if I can just get one machine installed with hard coded values. I have never used any of the OpenStack tools and I now got to "baremetal node validate", which errors out for me. | 06:34 |
opendevreview | Merged openstack/ironic master: Fix typo in deploy_templates docs https://review.opendev.org/c/openstack/ironic/+/890163 | 07:05 |
rpittau | good morning ironic! o/ | 07:19 |
dtantsur | awb: some validation errors will only go away when EVERYTHING is set up for deployment | 07:23 |
dtantsur | validation API is quite unfortunate because it's not designed to fully pass other than right before the deployment | 07:25 |
kubajj | Good morning Ironic o/ | 07:44 |
opendevreview | Jacob Anders proposed openstack/ironic master: [WIP] Wait for secure boot state change if it's not immediate https://review.opendev.org/c/openstack/ironic/+/863999 | 08:31 |
rpittau | mmm I think I fixed the issue with tinycore, now it fails on legacy boot :D | 08:32 |
rpittau | ok, it was expected | 08:32 |
opendevreview | Riccardo Pittau proposed openstack/ironic-python-agent-builder master: [WIP] Build tinyipa with tinycore 14.x https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/887754 | 08:45 |
opendevreview | Jacob Anders proposed openstack/ironic master: [WIP] Wait for secure boot state change if it's not immediate https://review.opendev.org/c/openstack/ironic/+/863999 | 08:46 |
*** dmellado17042088 is now known as dmellado | 09:51 | |
iurygregory | good morning Ironic | 11:18 |
kubajj | Morning iurygregory | 11:24 |
mohammed | Database is locked issue is raising again on metal3ci and I see also on Zuul build https://zuul.opendev.org/t/openstack/build/db4a8e328c434fbe8aa8b2ac622f369c though after TheJulia fixes it was green for sometime. Did anyone check that ? | 11:35 |
iurygregory | mohammed, at least according to the history of the job the last 2 failures are in patches that are WIP and not merged yet, in metal3ci you have been constantly reproducing the issue with database is locked? | 11:41 |
mohammed | iurygregory: on metal3 ci we are using the latest and we have seen this once at least https://github.com/metal3-io/cluster-api-provider-metal3/issues/1082#issuecomment-1660128548 | 11:51 |
mohammed | So it might not the wip changes I guess | 11:52 |
iurygregory | gotcha =( | 11:58 |
iurygregory | mohammed, what is the easy way to check the ironic.conf for the jenkis build? | 11:58 |
mohammed | iurygregory: unfortunately we are not collecting this file on the build not sure if we can get it when running the same ironic image ? | 12:07 |
iurygregory | right, it should be the same from ironic-image | 12:08 |
iurygregory | good catch (I'm still waking up, need more coffee) | 12:08 |
opendevreview | Riccardo Pittau proposed openstack/ironic-python-agent-builder master: [WIP] Build tinyipa with tinycore 14.x https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/887754 | 12:09 |
jfargen | How does bifrost manage images? | 14:06 |
dtantsur | jfargen: there is no much management. Images are stored in the nginx public location, you can also use file:/// links if you access Ironic yourself. | 14:20 |
jfargen | I did not see anything with regards to uploading an ipa image or overcloud images as happens with traditional tripleo deployments. | 14:21 |
dtantsur | jfargen: bifrost will get you an IPA image and configure Ironic to use it. It will also provide you with the default instance image (I think we default to building a debian image). | 14:23 |
jfargen | Okay, I ask because during the ironic-enroll-dynamic task I get this error, dropping it in pastebin as not to spam the channel. | 14:25 |
jfargen | https://pastebin.com/D3x09Y4j | 14:25 |
jfargen | For this use case, trying to use redfish virtual media and have added these properties. | 14:27 |
jfargen | "driver": "redfish", | 14:27 |
jfargen | "boot_interface": "redfish-virtual-media", | 14:27 |
jfargen | for the nodes instackenv.json. | 14:27 |
dtantsur | jfargen: start with checking your iDRAC firmware version. Old versions have "funny" bugs sometimes. Then make sure your iDRAC (!) can actually reach 192.168.67.49:8080 (firewall, networking, etc). | 14:30 |
jfargen | It is latest. | 14:30 |
jfargen | The iDRAC firmware is latest. | 14:30 |
dtantsur | Also check if any traces of the request are in local logs (not just nginx itself, but also audit, especially if you have selinux) | 14:30 |
opendevreview | Mahnoor Asghar proposed openstack/ironic master: WIP: Add inspection (processing) hooks https://review.opendev.org/c/openstack/ironic/+/887554 | 14:34 |
jfargen | The iDRAC has connectivity back to the host running bifrost/ironic. | 14:34 |
jfargen | acadm>>ping 192.168.67.49 | 14:34 |
jfargen | PING 192.168.67.49 (192.168.67.49) 56(84) bytes of data. | 14:34 |
jfargen | 64 bytes from 192.168.67.49: icmp_seq=1 ttl=63 time=0.813 ms | 14:34 |
jfargen | --- 192.168.67.49 ping statistics --- | 14:34 |
jfargen | 1 packets transmitted, 1 received, 0% packet loss, time 0ms | 14:34 |
jfargen | rtt min/avg/max/mdev = 0.813/0.813/0.813/0.000 ms | 14:34 |
jfargen | I guess, bifrost doesn't manage the hosts firewall config. | 14:39 |
jfargen | Added port 8080/tcp to firewalld config. | 14:41 |
jfargen | dtantsur; Thanks, this is really exciting! | 14:43 |
jfargen | The node is trying to introspect, there are a ton of errors look like network, with redfish virtual media is there an assumption that there will be dhcpv4 on the network? | 14:50 |
jfargen | And is there a way to statically assign and ip for introspection? | 14:56 |
rpittau | and it'a green \o/ https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/887754 | 15:14 |
rpittau | time to clean it up a bit | 15:14 |
opendevreview | Riccardo Pittau proposed openstack/ironic-python-agent-builder master: Build tinyipa with tinycore 14.x https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/887754 | 15:17 |
rpittau | there you go, ready for reviews :) | 15:17 |
JayF | jfargen: so you still use DHCP in default config w/virtual media | 15:18 |
JayF | jfargen: we support some kind of file you can send to set static | 15:18 |
jfargen | Yes, if you can point me to the file, that would be super helpful. | 15:18 |
JayF | well I'm failing big time finding the doc | 15:19 |
JayF | I don't work much in bifrost so I don't know a lot of this off the top | 15:19 |
JayF | rpittau: ^ do you remember, perchance | 15:19 |
rpittau | I found this https://docs.openstack.org/bifrost/latest/user/dhcp.html | 15:21 |
rpittau | the first entry should be helpful | 15:21 |
JayF | that looks like configuring static ip<>mac mappings | 15:28 |
JayF | not configuring no dhcp needed | 15:28 |
rpittau | oh yeah, sorry, I think the only way to do that would be using configdrive then https://docs.openstack.org/bifrost/latest/user/howto.html#static-network-configuration | 15:51 |
JayF | hmm but we still require dhcp for provisioning, then | 15:52 |
JayF | I know you can configure Ironic to not need that; I guess we don't have it wired up for bifrost :( | 15:52 |
rpittau | mmm I remember there was an option to do that, but not which one :D | 16:10 |
rpittau | it's late, my brain can't compute correctly anymore, time to go, see ya tomorrow o/ | 16:11 |
dtantsur | jfargen, JayF, somewhat experimental, but https://docs.openstack.org/ironic/latest/admin/dhcp-less.html | 16:15 |
dtantsur | note that bifrost does NOT automate setting node.network_data | 16:16 |
dtantsur | (someone literally complained about it on another chat a few days ago) | 16:16 |
JayF | I knew it was documented somewhere :D | 16:16 |
* dtantsur dislikes mariadb... | 16:18 | |
opendevreview | Mahnoor Asghar proposed openstack/ironic master: Add inspection (processing) hooks https://review.opendev.org/c/openstack/ironic/+/887554 | 16:47 |
JayF | I'm going to note, generally that the EM->Unmaintained proposal up for vote on the TC would no longer permit Ironic non-SLURP branches from being supported beyond the 18 month maintained line | 18:39 |
JayF | If this is concerning or breaks a downstream workflow; please comment directly on the TC proposal about your use case | 18:39 |
dtantsur | I don't think our capacity allows us to support any branches, (SLURP, non-SLURP, whatever) beyond 18 months. | 18:58 |
JayF | With my upstream hat on, I agree and have been acting accordingly on the TC. | 19:00 |
JayF | With my reality hat on, I see you all backporting stuff to god knows how far back all the time | 19:00 |
dtantsur | Well, I'd love to have stuff supported forever :) I just don't think we can | 19:00 |
JayF | and so I'm letting you know now so when some boss in a mauve hat says "I need 2023.2 supported for eleventy years", you were warned ;) | 19:00 |
dtantsur | haha, true | 19:00 |
dtantsur | could be more important for TheJulia's team than for us | 19:01 |
* dtantsur realizes that it's 9pm and leaves | 19:01 | |
JayF | yeah the weird thing that's happening is just non-slurp branches are not eligible for unmaintained/ branch status (the EM replacement) | 19:01 |
JayF | so that means e.g. bobcat just goes away after 18 months, no real choice | 19:01 |
JayF | bugfix/ branches live outside that ecosystem and we can do whatever we want with them | 19:01 |
JayF | but the integrated branch, we'd need to abide by those rules | 19:01 |
dtantsur | I think that's does not make sense but I have neither a case nor a desire to argue | 19:02 |
dtantsur | I'll reserve my opinion on this slurp thing | 19:02 |
JayF | dtantsur: half as much stuff for our volunteer group numbering zero to not-maintained is better, somehow | 19:02 |
JayF | dtantsur: my real opinion on slurp: I hope it's the first step to only releasing once a year | 19:02 |
dtantsur | yep | 19:02 |
TheJulia | Coordinated once a year, perhaps. I’m split on the policy, I think it will create more downstream work, but that may be less overhead than fixing CI on stable branches. Or more, possibly. | 19:32 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!