openstackgerrit | Julia Kreger proposed openstack/ironic-tempest-plugin master: Tear down raid sets https://review.opendev.org/757141 | 00:05 |
---|---|---|
guilhermesp | TheJulia: hey! nope... that's how i usually do for one of my regions `openstack baremetal node create --conductor-group <ironic_conductor_groups> --driver ipmi --driver-info ipmi_address=IP --driver-info ipmi_password=PW --driver-info ipmi_username=admin --name mellanox-node-1` | 00:07 |
guilhermesp | my understanding is that power operations as well as boot device are controlled through restful interface accordingly to this spec https://specs.openstack.org/openstack/ironic-specs/specs/backlog/openbmc-driver.html | 00:09 |
openstackgerrit | Julia Kreger proposed openstack/ironic-python-agent master: Support partition image efi contents https://review.opendev.org/748049 | 00:24 |
openstackgerrit | Julia Kreger proposed openstack/ironic-python-agent master: Add fstab pointer to EFI partition https://review.opendev.org/748260 | 00:25 |
TheJulia | guilhermesp: Okay, Yeah, I don't think anyplace in the code actually invokes getting the boot device so it may still work with ipmi. Is redfish an option at all. I remember there was talk of implementing redfish support in openbmc? | 00:27 |
*** k_mouza has joined #openstack-ironic | 00:27 | |
*** k_mouza has quit IRC | 00:31 | |
guilhermesp | I can try using redfish tho, TheJulia ( never used before, gonna read it around ). I have plan to enroll nodes with kvm over ip and the plan would be using redfish so, yeah it is an option | 00:33 |
*** k_mouza has joined #openstack-ironic | 00:36 | |
*** k_mouza has quit IRC | 00:40 | |
*** gyee has quit IRC | 00:43 | |
*** k_mouza has joined #openstack-ironic | 00:45 | |
openstackgerrit | Merged openstack/ironic master: Refactoring: split away continue_node_deploy/clean https://review.opendev.org/756343 | 00:49 |
*** k_mouza has quit IRC | 00:49 | |
*** k_mouza has joined #openstack-ironic | 00:54 | |
*** zzzeek has quit IRC | 00:58 | |
*** k_mouza has quit IRC | 00:59 | |
*** zzzeek has joined #openstack-ironic | 00:59 | |
*** k_mouza has joined #openstack-ironic | 01:02 | |
TheJulia | guilhermesp: okay, well let us know how it goes! | 01:03 |
guilhermesp | thank u TheJulia ! | 01:04 |
*** k_mouza has quit IRC | 01:07 | |
*** k_mouza has joined #openstack-ironic | 01:11 | |
*** k_mouza has quit IRC | 01:15 | |
*** Qianbiao has joined #openstack-ironic | 01:21 | |
*** xinliang has joined #openstack-ironic | 01:22 | |
*** k_mouza has joined #openstack-ironic | 01:27 | |
*** k_mouza has quit IRC | 01:31 | |
*** k_mouza has joined #openstack-ironic | 01:35 | |
*** k_mouza has quit IRC | 01:39 | |
*** k_mouza has joined #openstack-ironic | 01:58 | |
*** k_mouza has quit IRC | 02:03 | |
*** k_mouza has joined #openstack-ironic | 02:07 | |
*** k_mouza has quit IRC | 02:11 | |
*** stevebaker has quit IRC | 02:14 | |
*** k_mouza has joined #openstack-ironic | 02:32 | |
*** k_mouza has quit IRC | 02:37 | |
*** stevebaker has joined #openstack-ironic | 02:44 | |
*** rcernin has quit IRC | 02:48 | |
*** rcernin has joined #openstack-ironic | 03:04 | |
*** rcernin has quit IRC | 03:06 | |
*** rcernin has joined #openstack-ironic | 03:06 | |
*** stevebaker has quit IRC | 03:54 | |
*** xinliang has quit IRC | 04:04 | |
*** Qianbiao has quit IRC | 04:11 | |
*** Qianbiao has joined #openstack-ironic | 04:48 | |
*** tzumainn has quit IRC | 04:56 | |
*** Qianbiao has quit IRC | 05:29 | |
*** ociuhandu has joined #openstack-ironic | 05:45 | |
*** ociuhandu has quit IRC | 05:49 | |
*** zzzeek has quit IRC | 05:50 | |
*** zzzeek has joined #openstack-ironic | 05:52 | |
arne_wiebalck | Good morning, ironic! | 06:29 |
*** zzzeek has quit IRC | 06:38 | |
*** Qianbiao has joined #openstack-ironic | 06:39 | |
*** zzzeek has joined #openstack-ironic | 06:43 | |
*** k_mouza has joined #openstack-ironic | 06:52 | |
iurygregory | good morning arne_wiebalck and Ironic! | 06:55 |
arne_wiebalck | hey iurygregory o/ | 06:55 |
openstackgerrit | Merged openstack/ironic master: Do not pass BOOTIF=None if no BOOTIF can be guessed https://review.opendev.org/757859 | 06:56 |
openstackgerrit | Verification of a change to openstack/ironic failed: Log the ports we bind in flat https://review.opendev.org/757115 | 06:56 |
*** k_mouza has quit IRC | 06:57 | |
*** jtomasek has joined #openstack-ironic | 07:00 | |
*** anuradha1904 has joined #openstack-ironic | 07:07 | |
*** dsneddon has quit IRC | 07:24 | |
*** sshnaidm|afk is now known as sshnaidm | 07:36 | |
*** tosky has joined #openstack-ironic | 07:43 | |
openstackgerrit | Iury Gregory Melo Ferreira proposed openstack/ironic-prometheus-exporter master: Add Python3 wallaby unit tests https://review.opendev.org/755612 | 07:44 |
*** rcernin has quit IRC | 07:50 | |
*** rpittau|afk is now known as rpittau | 07:52 | |
rpittau | good morning ironic! o/ | 07:52 |
iurygregory | good morning rpittau o/ | 07:53 |
rpittau | hey iurygregory :) | 07:53 |
openstackgerrit | Riccardo Pittau proposed openstack/ironic master: Use centos as base element for dib images https://review.opendev.org/757811 | 07:56 |
*** Qianbiao has quit IRC | 08:10 | |
*** ociuhandu has joined #openstack-ironic | 08:11 | |
*** lucasagomes has joined #openstack-ironic | 08:19 | |
*** rcernin has joined #openstack-ironic | 08:26 | |
*** kafilat has joined #openstack-ironic | 08:32 | |
*** derekh has joined #openstack-ironic | 08:43 | |
janders | good morning arne_wiebalck iurygregory rpittau and Ironic o/ | 08:46 |
iurygregory | hey janders o/ | 08:47 |
rpittau | hey janders :) | 08:54 |
*** rcernin has quit IRC | 08:59 | |
*** ociuhandu_ has joined #openstack-ironic | 09:01 | |
*** Qianbiao has joined #openstack-ironic | 09:02 | |
*** ociuhandu has quit IRC | 09:04 | |
openstackgerrit | Zihao Wang proposed openstack/bifrost master: Bump hacking min version to 3.0.1 https://review.opendev.org/758042 | 09:12 |
*** ociuhandu_ has quit IRC | 09:15 | |
*** ociuhandu has joined #openstack-ironic | 09:16 | |
*** zzzeek has quit IRC | 09:17 | |
*** zzzeek has joined #openstack-ironic | 09:19 | |
arne_wiebalck | hey janders o/ | 09:20 |
openstackgerrit | Dmitry Tantsur proposed openstack/ironic stable/victoria: Do not pass BOOTIF=None if no BOOTIF can be guessed https://review.opendev.org/758050 | 09:34 |
*** jtomasek has quit IRC | 09:37 | |
*** zigo has joined #openstack-ironic | 10:12 | |
iurygregory | if any core has some time https://review.opendev.org/#/c/757800/ fix to avoid kernel panic in stable/train for IPE | 10:15 |
patchbot | patch 757800 - ironic-prometheus-exporter (stable/train) - Increase memory for stable/train - 1 patch set | 10:15 |
*** dougsz has joined #openstack-ironic | 10:19 | |
*** rcernin has joined #openstack-ironic | 10:24 | |
*** k_mouza has joined #openstack-ironic | 10:30 | |
openstackgerrit | Merged openstack/ironic master: Log the ports we bind in flat https://review.opendev.org/757115 | 10:37 |
*** kafilat has quit IRC | 10:51 | |
*** jtomasek has joined #openstack-ironic | 10:58 | |
*** dougsz has quit IRC | 11:07 | |
*** k_mouza has quit IRC | 11:08 | |
*** k_mouza has joined #openstack-ironic | 11:10 | |
*** dougsz has joined #openstack-ironic | 11:20 | |
openstackgerrit | Vladyslav Drok proposed openstack/ironic-inspector master: Add possibility to configure application root https://review.opendev.org/757536 | 11:20 |
*** zzzeek has quit IRC | 11:29 | |
*** zzzeek has joined #openstack-ironic | 11:32 | |
*** JamesBenson has joined #openstack-ironic | 11:32 | |
*** Qianbiao has quit IRC | 11:34 | |
openstackgerrit | Vladyslav Drok proposed openstack/ironic-inspector master: Add possibility to configure application root https://review.opendev.org/757536 | 11:34 |
*** priteau has joined #openstack-ironic | 11:39 | |
*** Qianbiao has joined #openstack-ironic | 11:45 | |
guilhermesp | mornings! TheJulia hey, openbmc nodes are being powered controlled by redfish! Thanks for the heads up! | 12:49 |
*** Goneri has joined #openstack-ironic | 13:04 | |
Qianbiao | hello, does ipab support dhcp from infiniband port? | 13:06 |
*** rloo has joined #openstack-ironic | 13:12 | |
rpittau | Qianbiao: it does, but I think it's limited to Mellanox cards | 13:13 |
openstackgerrit | Pranjali Srivastava proposed openstack/ironic master: WIP: Create 'redfish' driver Redfish Interop Profile https://review.opendev.org/754061 | 13:13 |
*** Goneri has quit IRC | 13:18 | |
TheJulia | guilhermesp: that is good to know, Just let us know if you run into anything odd with that since I don't know of anyone who has used openbmc + redfish before now | 13:18 |
guilhermesp | i have an odd already lol when we first add a node with redfish, it doesnt "read" the current power state. I.e if the node is on, when is enrolled is presented as "power off" even t he node being on. I needed to turn on and off the node to reflect the actual power state | 13:20 |
guilhermesp | also, i needed to `--driver-info redfish_verify_ca=false` to all openbmc nodes | 13:21 |
guilhermesp | i will try to provision instances today so, if i have some other comments to add, will do later today probably | 13:21 |
openstackgerrit | Merged openstack/bifrost master: Support redfish-virtual-media with legacy boot https://review.opendev.org/757848 | 13:33 |
Qianbiao | rpittau ty | 13:35 |
openstackgerrit | Julia Kreger proposed openstack/ironic-tempest-plugin master: Tear down raid sets https://review.opendev.org/757141 | 13:37 |
iurygregory | good morning TheJulia =) | 13:43 |
*** Goneri has joined #openstack-ironic | 13:48 | |
*** tzumainn has joined #openstack-ironic | 13:52 | |
*** rcernin has quit IRC | 14:01 | |
TheJulia | good morning iurygregory | 14:08 |
TheJulia | guilhermesp: I wonder if the bmc is slighly out of sync then becasue we just record what the BMC advertises the state to be. | 14:08 |
TheJulia | guilhermesp: not verifying the CA certificate makes sense because you'd need to have TLS infrastucture setup and in testing, you likely don't have that :) | 14:09 |
*** priteau has quit IRC | 14:09 | |
*** priteau has joined #openstack-ironic | 14:19 | |
guilhermesp | yep TheJulia that makes sense | 14:20 |
*** Goneri has quit IRC | 14:31 | |
*** ianychoi_ has quit IRC | 14:33 | |
tzumainn | hi! is someone here familiar with the workflow for booting from a volume? I'm following the steps, and encountering something I think is weird - the node is attached to the provisioning network, and ironic will delete the neutron port resulting in errors - but I don't know if that's something that's supposed to happen... ? | 14:34 |
TheJulia | o/ tzumainn | 14:36 |
tzumainn | hi! | 14:36 |
*** Goneri has joined #openstack-ironic | 14:36 | |
TheJulia | tzumainn: what _should_ be occuring is in this case, the node will forever need to be able to reach the endpoint the nodes connect to for getting the boot config (that is, unless, the bmc supports iscsi booting directly in the driver) | 14:37 |
TheJulia | so basically the port should get torn down in the provisioning network, the final port should get created and attached via ml2 and updated with the required dhcp configuration including to network boot | 14:38 |
tzumainn | TheJulia, what defines the final port? | 14:39 |
TheJulia | the vif attachment | 14:44 |
TheJulia | so whatever is the final vif | 14:44 |
tzumainn | gotcha | 14:44 |
TheJulia | oh, you guys are running flat networking + neutron plugin right? | 14:44 |
tzumainn | ml2-ansible networking | 14:45 |
TheJulia | then the right thing should be taking place | 14:45 |
arne_wiebalck | TheJulia: Scaling: seems like the timeouts (in nova or ha proxy) come from the conductor not handling request quickly enough, but the conductor does not crash. I could increase the timeouts, but it seems scaling out the conductors solves this timeout problem and I don't want to go overboard with trying to create an OOM situation. From what I see, the instances go into ERROR with a timeout while the | 14:45 |
arne_wiebalck | conductor stays up ... which is different (and much better) from what I have seen before. | 14:45 |
* arne_wiebalck tries to say he will stop scaling tests for now | 14:45 | |
TheJulia | arne_wiebalck: stop scale testing, post comments in gerrit! | 14:46 |
TheJulia | I need to retool that patch a little to make it backportable | 14:46 |
TheJulia | and then we go from there! | 14:46 |
arne_wiebalck | TheJulia: ++ | 14:46 |
TheJulia | *insert evil laugh here* | 14:46 |
arne_wiebalck | TheJulia: we also got some good spin off, like the additional RAID hardening patches. | 14:47 |
arne_wiebalck | TheJulia: These were worth it: from 200 instances, 190 went into active. | 14:47 |
TheJulia | on a single conductor? | 14:47 |
arne_wiebalck | TheJulia: No, 3 conductors. | 14:47 |
TheJulia | not bad | 14:48 |
arne_wiebalck | TheJulia: With 1 conductor, 150 timed out. | 14:48 |
TheJulia | wow | 14:48 |
TheJulia | what is the threadpool set to ? | 14:48 |
arne_wiebalck | Scaling the conductors seems to be the correct approach. | 14:48 |
TheJulia | in your conductors | 14:48 |
arne_wiebalck | Conductors or APIs? | 14:49 |
arne_wiebalck | The thread pool size? | 14:49 |
tzumainn | TheJulia, the endpoint for getting the boot config - is that just the ironic endpoint? | 14:50 |
*** ianychoi has joined #openstack-ironic | 14:51 | |
arne_wiebalck | TheJulia: Seems that we currently run with the defaults. | 14:52 |
* arne_wiebalck is sure we changed values in the past | 14:52 | |
TheJulia | arne_wiebalck: conductor | 14:55 |
TheJulia | so basically at 200 concurrently you maxed out your threadpool sizea nyway | 14:55 |
TheJulia | size | 14:55 |
TheJulia | tzumainn: that is the endpoint for ipxe | 14:55 |
tzumainn | TheJulia, gotcha - okay, thanks! I'm seeing some weird behavior given what you said, but I'll dig a bit deeper and see if I can figure it out first | 14:57 |
arne_wiebalck | TheJulia: the threadpool size is 64 ? | 15:00 |
*** belmoreira has joined #openstack-ironic | 15:01 | |
arne_wiebalck | TheJulia: 3x 64 would be close to 200 and 1x 64 is around the number which worked for 1 controller. | 15:01 |
TheJulia | arne_wiebalck: 80 I think, but ?6? or ?8? + 1 get reserved | 15:01 |
*** JamesBenson has quit IRC | 15:01 | |
*** JamesBenson has joined #openstack-ironic | 15:01 | |
arne_wiebalck | TheJulia: Ok. Take away message I think: I cannot break the conductor (train) and get good results by scaling it out. | 15:02 |
TheJulia | at a glance, 14 I think get reserved | 15:02 |
TheJulia | or reserved or consumed automagically | 15:03 |
TheJulia | at least, that is on master | 15:03 |
arne_wiebalck | Ok | 15:03 |
TheJulia | I bet nova was likely seeing "NoFreeWorker" errors | 15:03 |
TheJulia | as well as some timeouts | 15:03 |
arne_wiebalck | nova gets 500s | 15:04 |
arne_wiebalck | which is probably ha proxy cutting the connectin | 15:04 |
arne_wiebalck | connection | 15:04 |
TheJulia | http 503 is what it would return | 15:05 |
TheJulia | NoFreeConductorWorker | 15:05 |
arne_wiebalck | I think we only get 500s ... | 15:05 |
arne_wiebalck | Nova colleagues claim Ironic is too slow. | 15:06 |
TheJulia | then yeah, likely haproxy | 15:06 |
TheJulia | lol | 15:06 |
arne_wiebalck | I claim Nova should be more gently. | 15:06 |
arne_wiebalck | gentle | 15:06 |
TheJulia | nova internally just backs the queue up | 15:06 |
TheJulia | I _wonder_ if the worker to respond with the no free worker is actually working as expected | 15:07 |
arne_wiebalck | I have not checked the API logs yet ... | 15:08 |
TheJulia | well, the API should take the reqeust | 15:08 |
TheJulia | the conductors should be going "err, no, can't work that, can't allocate free workers" | 15:09 |
TheJulia | but if that is not instantly coming back when it is out of workers, then we have an issue | 15:09 |
TheJulia | or coming back in a reasonable amount of time, which may be what your nova colleagues are observing | 15:09 |
Qianbiao | Hi, a quick question. what is the purpose to include "runtime-ssh-host-keys" element in IPAB. | 15:09 |
arne_wiebalck | TheJulia: I don't see errors in our logs, so it may be that our ha proxy timeout is too short ... from when to when would a thread be tied up? | 15:17 |
arne_wiebalck | Qianbiao: I think this is to include ssh keys to log into the IPA image (which may be helpful for debugging). | 15:18 |
*** rcernin has joined #openstack-ironic | 15:22 | |
*** rcernin has quit IRC | 15:27 | |
Qianbiao | arne_wiebalck hmm, if we want to login to IPA, should we update the authorized_keys. | 15:28 |
arne_wiebalck | yes | 15:28 |
arne_wiebalck | we do this | 15:28 |
arne_wiebalck | it is very handy to be able to log into the IPA | 15:29 |
Qianbiao | "runtime-ssh-host-keys" is not for this, it generated keys under /etc/ssh | 15:29 |
arne_wiebalck | Qianbiao: right, we use the local-config DIB for what I described | 15:33 |
Qianbiao | arne_wiebalck DIB has dynamic-login element can do what you described. | 15:34 |
Qianbiao | I just use devuser for debug :) | 15:35 |
Qianbiao | arne_wiebalck It seems IPA requires ssh-hostkey https://github.com/IamFive/diskimage-builder/commit/c67e7ed15ede7e3d387dca82c06845f0c0e332f9 | 15:39 |
Qianbiao | although i do not know where IPA uses it :) | 15:39 |
arne_wiebalck | ah ... not sure either :) | 15:40 |
*** gyee has joined #openstack-ironic | 15:41 | |
Qianbiao | arne_wiebalck you are right, https://bugs.launchpad.net/diskimage-builder/+bug/1556145 | 15:43 |
openstack | Launchpad bug 1556145 in diskimage-builder "ironic-agent ramdisk does not generate SSH hostkey on boot" [High,Fix released] - Assigned to Gregory Haynes (greghaynes) | 15:43 |
Qianbiao | But this is wierd, AFAIK, ssh should success even there are no ssh-hostkey generated. will test it later. | 15:43 |
*** Qianbiao has quit IRC | 15:50 | |
openstackgerrit | Riccardo Pittau proposed openstack/ironic master: Run bifrost integration job on focal https://review.opendev.org/758198 | 15:51 |
openstackgerrit | Riccardo Pittau proposed openstack/ironic-inspector master: Run bifrost integration job on focal https://review.opendev.org/758199 | 15:52 |
*** derekh has quit IRC | 15:54 | |
openstackgerrit | Riccardo Pittau proposed openstack/bifrost master: [WIP] Convert job to focal and remove bionic jobs https://review.opendev.org/758201 | 15:56 |
rpittau | good night! o/ | 16:05 |
*** rpittau is now known as rpittau|afk | 16:05 | |
openstackgerrit | Riccardo Pittau proposed openstack/bifrost master: [WIP] Convert job to focal and remove bionic jobs https://review.opendev.org/758201 | 16:09 |
*** lucasagomes has quit IRC | 16:09 | |
*** tosky has quit IRC | 16:21 | |
openstackgerrit | Pierre Riteau proposed openstack/bifrost master: Add missing space in error message https://review.opendev.org/758206 | 16:33 |
*** dougsz has quit IRC | 16:33 | |
*** iurygregory has quit IRC | 16:41 | |
*** iurygregory has joined #openstack-ironic | 16:44 | |
openstackgerrit | Julia Kreger proposed openstack/ironic-tempest-plugin master: Tear down raid sets https://review.opendev.org/757141 | 16:45 |
*** ociuhandu_ has joined #openstack-ironic | 16:50 | |
*** ociuhandu has quit IRC | 16:54 | |
*** ociuhandu_ has quit IRC | 16:55 | |
*** _mike_kelley has joined #openstack-ironic | 17:07 | |
_mike_kelley | Is this the correct channel for sushy? I haven't been able to locate a dedicated channel. | 17:08 |
arne_wiebalck | _mike_kelley: It is the correct channel :) | 17:19 |
* arne_wiebalck has to leave | 17:20 | |
arne_wiebalck | Bye everyone, see you all tmrw o/ | 17:20 |
_mike_kelley | Outstanding. I'm working with sushy-emulator to model a set of vms in a non-openstack libvirt host. Everything works well enough, but I need to be able to target a subset of the VMs on the host and leave the remaining VMs untouched. I can do that easily enough using System.name and Manager.name in a standalone script, but I need to be able to work it into an existing system that was designed for physical BMCs, | 17:27 |
_mike_kelley | umption that one IP address represents one installation target. So I'm investigating the chassis object in sushy-emulator, but it doesn't look like that object has a mechanism to assign specific VMs to a specific chassis yet. Am I correct on this? If so, is there another avenue I've missed? | 17:27 |
*** kafilat has joined #openstack-ironic | 17:28 | |
*** k_mouza has quit IRC | 17:30 | |
*** k_mouza has joined #openstack-ironic | 17:35 | |
*** _mike_kelley has quit IRC | 17:35 | |
TheJulia | goodnight arne_wiebalck | 17:40 |
*** k_mouza has quit IRC | 17:40 | |
*** Goneri has quit IRC | 17:45 | |
*** k_mouza has joined #openstack-ironic | 17:46 | |
*** Goneri has joined #openstack-ironic | 17:47 | |
*** dsneddon has joined #openstack-ironic | 17:50 | |
*** k_mouza has quit IRC | 17:50 | |
*** _mike_kelley has joined #openstack-ironic | 17:53 | |
*** k_mouza has joined #openstack-ironic | 18:01 | |
*** k_mouza has quit IRC | 18:06 | |
openstackgerrit | Merged openstack/bifrost master: Add missing space in error message https://review.opendev.org/758206 | 18:14 |
*** k_mouza has joined #openstack-ironic | 18:16 | |
*** k_mouza has quit IRC | 18:21 | |
*** rcernin has joined #openstack-ironic | 18:22 | |
*** k_mouza has joined #openstack-ironic | 18:22 | |
*** k_mouza has quit IRC | 18:26 | |
*** k_mouza has joined #openstack-ironic | 18:29 | |
*** k_mouza has quit IRC | 18:33 | |
*** rcernin has quit IRC | 18:37 | |
*** k_mouza has joined #openstack-ironic | 18:37 | |
*** k_mouza has quit IRC | 18:42 | |
*** belmoreira has quit IRC | 18:51 | |
*** ociuhandu has joined #openstack-ironic | 19:00 | |
*** belmoreira has joined #openstack-ironic | 19:03 | |
*** ociuhandu has quit IRC | 19:05 | |
*** k_mouza has joined #openstack-ironic | 19:05 | |
*** dougsz has joined #openstack-ironic | 19:07 | |
*** k_mouza has quit IRC | 19:10 | |
*** rloo has quit IRC | 19:13 | |
*** k_mouza has joined #openstack-ironic | 19:13 | |
*** rloo has joined #openstack-ironic | 19:14 | |
*** rloo has quit IRC | 19:15 | |
*** k_mouza has quit IRC | 19:15 | |
*** rloo has joined #openstack-ironic | 19:16 | |
*** k_mouza has joined #openstack-ironic | 19:17 | |
*** dougsz has quit IRC | 19:17 | |
*** rcernin has joined #openstack-ironic | 19:19 | |
*** priteau has quit IRC | 19:20 | |
*** rcernin has quit IRC | 19:24 | |
*** stevebaker has joined #openstack-ironic | 19:24 | |
openstackgerrit | Steve Baker proposed openstack/ironic master: Fix ipmitool timing argument calculation https://review.opendev.org/756183 | 19:34 |
stevebaker | morning | 19:41 |
*** k_mouza has quit IRC | 19:42 | |
*** k_mouza has joined #openstack-ironic | 19:49 | |
*** tosky has joined #openstack-ironic | 19:53 | |
*** k_mouza has quit IRC | 19:54 | |
*** kafilat has quit IRC | 19:57 | |
TheJulia | good morning stevebaker | 19:58 |
*** k_mouza has joined #openstack-ironic | 19:59 | |
openstackgerrit | Julia Kreger proposed openstack/ironic-tempest-plugin master: Tear down raid sets https://review.opendev.org/757141 | 20:10 |
*** jtomasek has quit IRC | 20:15 | |
openstackgerrit | Julia Kreger proposed openstack/ironic master: IPMI: Handle vendor set boot device differences https://review.opendev.org/757198 | 20:17 |
dking | TheJulia: Thank you. I'm trying to understand exactly how it works to help me get an idea of what should be different. I'm trying to see where callback_url is actually set. But maybe it's simply set when a heartbeat request is received? And then it expects it to always be the same IP? | 20:23 |
TheJulia | it is exactly that | 20:25 |
TheJulia | it is a security mechanism that expects it not to change | 20:25 |
TheJulia | this was before we had agent token | 20:25 |
dking | I'm still trying to find where it's set, but is there a way for the node to check to see what the agent_url is, and attempt to reuse it for the callback_url? | 20:26 |
TheJulia | it is in the api code | 20:28 |
TheJulia | and that gets passed into the conductor as part of heartbeat and in the central heartbeat processinc code if memory serves | 20:28 |
TheJulia | so in the agent code in driver/modules/agent_base.py | 20:28 |
TheJulia | I think | 20:28 |
*** belmoreira has quit IRC | 20:36 | |
dking | Okay. I see it there, but I'm looking at the IPA client. I'm assuming it sets it somewhere. It looks like it does get the address there somewhere, but it's passed down a long way. I'm trying to backtrack. | 20:36 |
dking | Because the client needs to be sure to send the request from the right IP. | 20:36 |
*** k_mouza has quit IRC | 20:36 | |
dking | So far, I see that it's set from "self.agent.advertise_address", but that "agent" is what is passed in from a couple of parent classses, so I'm tryingto find it. | 20:37 |
dking | Ah, here: ironic_python_agent/agent.py IronicPythonAgent.set_agent_advertise_addr() | 20:39 |
dking | It looks to see if the address is already set (in the class), and if not, it looks next for the network interface (with a note from dtantsur that it should be deprecated) if it exists, and if not, then it parses the api_url. | 20:44 |
*** k_mouza has joined #openstack-ironic | 20:47 | |
*** k_mouza has quit IRC | 20:59 | |
*** k_mouza has joined #openstack-ironic | 21:12 | |
*** k_mouza has quit IRC | 21:17 | |
TheJulia | dking: it determines it from the ip on the network it has a default route on | 21:17 |
TheJulia | so if the IP changes addressing is (a) unstable or (b) there are multipel default routes which is bad | 21:17 |
*** jtomasek has joined #openstack-ironic | 21:24 | |
dking | TheJulia: It's not having a problem while it's constantly running. The issue I've noticed is after a reboot after a firmware update. The firmware requires a reboot, and so that's set as part of the cleaning step. | 21:25 |
dking | Should it be very unexpected for NICs to get swapped during firmware updates or other potential cleaning steps? | 21:25 |
*** anuradha1904 has quit IRC | 21:27 | |
*** k_mouza has joined #openstack-ironic | 21:29 | |
*** k_mouza has quit IRC | 21:34 | |
*** k_mouza has joined #openstack-ironic | 21:37 | |
*** k_mouza has quit IRC | 21:41 | |
*** k_mouza has joined #openstack-ironic | 21:46 | |
*** jtomasek has quit IRC | 21:46 | |
*** k_mouza has quit IRC | 21:50 | |
*** rloo has quit IRC | 22:02 | |
*** vkmc has quit IRC | 22:10 | |
openstackgerrit | Julia Kreger proposed openstack/ironic-inspector master: WIP: Try2... power off the node before inspection https://review.opendev.org/756930 | 22:10 |
*** vkmc has joined #openstack-ironic | 22:10 | |
*** k_mouza has joined #openstack-ironic | 22:11 | |
*** erbarr has quit IRC | 22:13 | |
*** erbarr has joined #openstack-ironic | 22:15 | |
*** k_mouza has quit IRC | 22:15 | |
openstackgerrit | Julia Kreger proposed openstack/ironic-tempest-plugin master: Tear down raid sets https://review.opendev.org/757141 | 22:24 |
TheJulia | dking: not sure if we should expect that but it seems like there could be other things going on. Your welcome to file a bug or hack at it and upload a patch. If you do file a bug please include your `baremetal node show` output so we can see the driver_internal_info field and reconcile it against the state | 22:25 |
TheJulia | arne_wiebalck: fyi https://review.opendev.org/#/c/752308/ works and installs a centos8 image without issues which is a really good sigh. The following patch seems to have a legit failure and I'll look at that tomorrow | 22:30 |
patchbot | patch 752308 - ironic-python-agent - Don't run os-prober from grub2-mkconfig - 10 patch sets | 22:30 |
*** JamesBenson has quit IRC | 22:31 | |
*** JamesBenson has joined #openstack-ironic | 22:31 | |
*** rcernin has joined #openstack-ironic | 22:33 | |
*** rcernin has quit IRC | 22:36 | |
*** JamesBenson has quit IRC | 22:36 | |
*** rcernin has joined #openstack-ironic | 22:36 | |
*** stevebaker has quit IRC | 23:04 | |
*** JamesBenson has joined #openstack-ironic | 23:06 | |
*** JamesBenson has quit IRC | 23:12 | |
janders | good morning Ironic o/ | 23:18 |
dking | TheJulia: Thank you. I'm just disabling the reboot in the clean step for now, and hoping it doesn't cause a problem. I might run some tests to see if I could move or remove that network line, or maybe just waiting until I talk with dtantsur to see how he expected that line to be deprecated. | 23:20 |
TheJulia | Ok | 23:21 |
*** tosin has joined #openstack-ironic | 23:23 | |
*** tosky has quit IRC | 23:27 | |
*** JamesBenson has joined #openstack-ironic | 23:39 | |
*** JamesBenson has quit IRC | 23:44 | |
*** rcernin has quit IRC | 23:45 | |
*** rcernin has joined #openstack-ironic | 23:50 | |
*** tosin has quit IRC | 23:51 | |
*** Goneri has quit IRC | 23:59 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!