| opendevreview | Steve Baker proposed openstack/ironic master: Allocation candidates prefer matching name https://review.opendev.org/c/openstack/ironic/+/845856 | 03:31 |
|---|---|---|
| arne_wiebalck | Good morning, Ironic! | 06:25 |
| waleedm | TheJulia, Can you please review this https://review.opendev.org/c/openstack/ironic-python-agent/+/845583 | 07:21 |
| ajya | iurygregory, janders is this about 730 or 740? Or both? | 07:21 |
| rpittau | good morning ironic! o/ | 07:33 |
| dmellado | hiya rpittau ;) | 07:47 |
| rpittau | hey dmellado :) | 07:48 |
| dmellado | humidity 72% and hitting 30ยบ today, let's see xD | 07:48 |
| rpittau | kind of in the same situation here :/ | 07:49 |
| * dmellado wonders how July would be | 07:49 | |
| rpittau | :( | 07:50 |
| ajya | iurygregory, janders I checked - in 13G (R730) there is no support for these management functions, in 14G (R740) there is. Generally, I wouldn't use Redfish with 13G, try WS-Man. cc rpioso | 07:53 |
| ajya | while in 13G there is some Redfish support, it does not receive new features and sometimes bugs are not backported too. 13G is still WS-Man world. | 07:54 |
| ajya | e.g., Redfish RAID interface is not working in 13G too. | 07:55 |
| janders | Thank you ajya | 09:02 |
| janders | ajya I wonder - should we enhance clear_job_queue (and other clean/verify steps specific to iDRAC) so that they check the generation and don't run on 13G (or older hardware in general)? | 09:02 |
| janders | they won't run on non-Dells, but once we have a node with iDRAC driver Ironic will always try running them AFAIK | 09:03 |
| janders | also - do we need to install any specific packages to use iDRAC specific stuff ( clean/verify steps, using idrac-redfish-virtualmedia boot interface )? | 09:06 |
| ajya | janders: there is idrac-wsman implementation for clear_job_queue. Or do you want to keep using idrac-redfish on 13G? | 09:08 |
| janders | I can try (to get to the bottom of it) but the problem is that downstream things tend to be streamlined and we don't have as much control as in Ironic | 09:08 |
| ajya | about packages - like sushy-oem-idrac or python-dracclient? | 09:08 |
| janders | yeah this kind of stuff | 09:09 |
| janders | (for the context - downstream = metal3/ocp, not OSP - OSP is very much like upstream in this sense - more tweakable) | 09:09 |
| ajya | looking at the errors, you already have sushy-oem-idrac installed otherwise wouldn't get so far. Not sure how metal3 installs these, but can check if they are there out of the box | 09:14 |
| ajya | janders: metal3 has option for idrac-wsman (with idrac://), but wsman has no support for virtual media boot. | 09:16 |
| ajya | It's possible to mix and match idrac-wsman and idrac-redfish but as you said, metal3 does not have that flexibility | 09:16 |
| rpittau | ajya: the sushy-oem-idrac is installed in the ironic container in metal3 | 09:17 |
| rpittau | we take the pkgs from tripleo repos | 09:18 |
| ajya | whether to enhance idrac-redfish steps to check for 13G, it could be doable, but still better to use wsman with 13G | 09:19 |
| ajya | maybe it is not documented enough that this is preferred way, there is some initial redfish support, but it can't keep up with what's possible in Ironic now | 09:20 |
| ajya | thanks rpittau | 09:20 |
| janders | ajya if we check for support for idrac-redfish steps, what approach would you recommend? check for server generation? check for presence of the RedFish URLs required for things to work? | 09:22 |
| rpittau | I'm actually wondering if an issue I'm investigating can be due to "old" hardware not supporting virtualmedia | 09:23 |
| janders | I may put a patch together | 09:23 |
| janders | if I do I will request a review from you :) | 09:23 |
| ajya | janders: checking for presence would be easier, otherwise have to start parsing model names and model number conventions to determine generation or have a list of all generation models. | 09:30 |
| janders | ajya ACK, will have a look into that | 09:30 |
| ajya | janders: maybe even check that in sushy-oem-idrac https://opendev.org/openstack/sushy-oem-idrac/src/branch/master/sushy_oem_idrac/resources/manager/manager.py#L166 | 09:30 |
| janders | hmmm that may be a good idea | 09:31 |
| ajya | and then raise missing error that Ironic chooses to handle as ignore | 09:31 |
| ajya | rpittau: how old? From what I saw for virtual media boot in 13G there is functionality for idrac-redfish-virtual-media, but we are not actively testing 13G. | 09:32 |
| rpittau | ajya: not sure, I need to double-check, the problem is that even if we use idrac-virtualmedia it does not get the correct bios_interface, at the moment I'm checking to see if it's a regression in one of the metal3 components | 09:34 |
| ajya | janders: same would be needed for idrac_card_service as that's where reset_idrac step uses | 09:36 |
| ajya | s/where/what | 09:37 |
| ajya | rpittau: I can check if deployment with latest 13G iDRAC succeeds using idrac-redfish in all places, minus verify steps | 09:38 |
| rpittau | ajya: thanks! | 09:48 |
| * TheJulia yawns | 10:24 | |
| janders | guys, is there an easy-enough way to deploy bifrost with Ironic from the time of a particular merged change? | 11:04 |
| iurygregory | thank you ajya, I will follow-up with folks here to see what we can do since is unsupported =) | 11:13 |
| opendevreview | Verification of a change to openstack/networking-generic-switch master failed: Trim spaces in ngs_trunk_ports and ngs_physical_networks https://review.opendev.org/c/openstack/networking-generic-switch/+/845774 | 11:26 |
| TheJulia | janders: replace the branch with a commit hash | 11:36 |
| TheJulia | for the assets used | 11:36 |
| rpittau | ah snap we don't checkout swiftclient from master, we need a new release :/ | 11:38 |
| dmellado | janders: should be here https://docs.openstack.org/bifrost/latest/install/index.html#clone-bifrost | 11:38 |
| dmellado | beware of the python3.6 issue, depending on your environment and so, though | 11:38 |
| TheJulia | rpittau: unfortunately | 11:39 |
| rpittau | I think it's time to remove support for py3.6 from bifrost, we're not testing it anymore | 11:41 |
| TheJulia | I cannot send a single line with enough +'s | 11:41 |
| rpittau | lol | 11:41 |
| * TheJulia goes back to reading a proposal | 11:41 | |
| opendevreview | Riccardo Pittau proposed openstack/bifrost master: Remove support for Python 3.6 and 3.7 https://review.opendev.org/c/openstack/bifrost/+/846154 | 11:54 |
| iurygregory | ajya, would it be a problem to add a check in clear_job_queue to check for "The attribute Links/Oem/Dell/DellJobService is missing from the resource /redfish/v1/Managers/iDRAC.Embedded.1']" in case is not present we would not fail. Does WS-Man supports virtual-media? (This is a requirement in the setup they are testing downstream) | 11:57 |
| opendevreview | Riccardo Pittau proposed openstack/bifrost master: Remove support for Python 3.6 and 3.7 https://review.opendev.org/c/openstack/bifrost/+/846154 | 12:04 |
| opendevreview | Riccardo Pittau proposed openstack/ironic master: Use bifrost on centos9 integration job https://review.opendev.org/c/openstack/ironic/+/846155 | 12:06 |
| opendevreview | Riccardo Pittau proposed openstack/ironic master: Use bifrost on centos9 integration job https://review.opendev.org/c/openstack/ironic/+/846155 | 12:08 |
| ajya | iurygregory: wouldn't be a problem, talked with janders about it earlier; | 12:09 |
| ajya | iurygregory: in Ironic there is no WS-Man option for virtual media, but now thinking if there were support at iDRAC for that. Not sure if this has come up, or ppl just went straight with redfish. | 12:11 |
| TheJulia | The biggest issue is boot mode management from my POV | 12:13 |
| TheJulia | that has repeatedly come up in the last few weeks with people trying to use the idrac hw type with wsman drivers | 12:14 |
| TheJulia | Of course, that skews my own perception | 12:14 |
| TheJulia | But I also think many went directly to redfish for virtual media | 12:14 |
| TheJulia | The drivers for irmc/ilo were much more high security deployment ops | 12:14 |
| TheJulia | aiui | 12:14 |
| ajya | TheJulia: yes, there is a little dance in WS-Man to get it set and it is slower, that's why use Redfish whenever possible. | 12:18 |
| dmellado | rpittau++ on the removal xD | 12:25 |
| rpittau | :) | 12:25 |
| dmellado | it was the first time ever that I had to go and debug pip... | 12:26 |
| dmellado | xD | 12:26 |
| opendevreview | Riccardo Pittau proposed openstack/ironic-python-agent master: lsblk: output all available columns https://review.opendev.org/c/openstack/ironic-python-agent/+/845785 | 12:38 |
| rpittau | dmellado: oh that's no fun, did it a lot of times and it's always messy | 12:39 |
| TheJulia | ajya: did we change the default load order yet, perhaps it is time? Meaning towards redfish. I'm finding internally some people don't grok/delineate mainly because there has not been a forced need nor have an easy knob, so we end up in some bad patterns/expectations. On a plus side, I think we save the interfaces and they become static after the fact on the object model, so only differing behavior with new nodes | 13:20 |
| TheJulia | iurygregory: any objection of a blog post w/r/t the baremetal sig session ? | 13:23 |
| TheJulia | asking because I'm unlikely to self approve content for ironicbaremetal.org :) | 13:23 |
| ajya | TheJulia: idrac-wsman is still default, but could look at changing that. Have to look at the gaps left and differences, they are small though, and might not be applicable to all. | 13:33 |
| TheJulia | I think it would be good to go ahead and do so, but definitely something that needs to be done with intention and some detailed research first :) | 13:52 |
| iurygregory | TheJulia, none =) | 14:10 |
| opendevreview | Derek Higgins proposed openstack/ironic-python-agent master: Gather details about bond interfaces if present https://review.opendev.org/c/openstack/ironic-python-agent/+/846177 | 14:10 |
| TheJulia | derekh: like auto-forming bonds ? | 14:15 |
| derekh | TheJulia: in the setup we hit this networking (including bonds) is setup with nmstate/NetworkManager , then once ready IPA gets started in a container | 14:17 |
| TheJulia | nice | 14:17 |
| TheJulia | was not aware it could do that! | 14:17 |
| derekh | its a recent change to how we run IPA in IPI but now people are passing bonding info in via nmstate and we're no longer collecting the IP address on the bonds... | 14:19 |
| TheJulia | ugh | 14:39 |
| ajya | rpittau: was able to deploy a node using idrac-redfish-virtual-media and other idrac-redfish interfaces successfully on R630. The only thing out of ordinary was that from grub menu to IPA it took almost 5 mins. | 15:00 |
| TheJulia | That is a while... | 15:11 |
| TheJulia | but, that is highly variable on the bmc/network connectivity more than anything else | 15:11 |
| ajya | not sure what's causing it, but on the same network 14G/15G systems are much faster when loading IPA | 15:15 |
| TheJulia | hmmm | 15:15 |
| TheJulia | still a ton of variables with a different model and all | 15:15 |
| rpittau | ajya: thanks! that's really useful info | 15:33 |
| opendevreview | Riccardo Pittau proposed openstack/bifrost master: Remove support for Python 3.6 and 3.7 https://review.opendev.org/c/openstack/bifrost/+/846154 | 15:37 |
| opendevreview | Riccardo Pittau proposed openstack/bifrost master: Remove old unused CentOS Stream 8 based jobs https://review.opendev.org/c/openstack/bifrost/+/846193 | 15:42 |
| rpittau | good night! o/ | 17:05 |
| JayF | I think IPA gate may be busted? There's a basically noop change failing the same jobs over and over -- ipa-tempest-bios-ipmi-direct-src and ipa-tempest-uefi-redfish-vmedia-src | 18:05 |
| JayF | Is there any known issue going on? Or is this fresh breakage | 18:05 |
| JayF | looks like something broken inside DIB | 18:06 |
| JayF | Hmm, CentOS 8. I wonder if this is related to the DIB py26 deprecation that had to be reverted... | 18:10 |
| * JayF has come full circle, troubleshooting a gate breakage in another project that a change he pushed caused | 18:13 | |
| TheJulia | Where did the day go | 18:58 |
| * TheJulia feels confuzzled | 18:58 | |
| * TheJulia calls it a day | 20:01 | |
| stevebaker[m] | good morning | 20:42 |
| TheJulia | good morning stevebaker[m] | 20:50 |
| stevebaker[m] | TheJulia: welcome back! | 20:50 |
| TheJulia | Thanks | 20:53 |
| TheJulia | Feeling slightly fried, but I started my day a little after 3 am this morning | 20:53 |
| stevebaker[m] | TheJulia: whelp that doesn't match with *any* timezone you've been in :p | 21:04 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!