| openstackgerrit | Julia Kreger proposed openstack/ironic master: Block port deletions where vif is present https://review.opendev.org/665835 | 00:04 |
|---|---|---|
| TheJulia | rloo: mgoddard: drastically changed, thoughts would be appreciated ^ | 00:04 |
| TheJulia | stevebaker: did you ever obtain consensus on a forward path re data validations? | 00:19 |
| openstackgerrit | Julia Kreger proposed openstack/ironic master: Extend power sync timeout for Ericsson SDI https://review.opendev.org/712592 | 00:22 |
| *** TxGirlGeek has quit IRC | 00:22 | |
| *** ijw has joined #openstack-ironic | 00:26 | |
| *** ijw has quit IRC | 00:30 | |
| *** ijw has joined #openstack-ironic | 00:30 | |
| *** openstackstatus has joined #openstack-ironic | 00:44 | |
| *** ChanServ sets mode: +v openstackstatus | 00:44 | |
| *** erbarr has quit IRC | 00:55 | |
| *** rloo has quit IRC | 01:17 | |
| *** diurnalist has joined #openstack-ironic | 01:33 | |
| *** diurnalist has quit IRC | 01:38 | |
| *** gyee has quit IRC | 01:51 | |
| stevebaker | TheJulia: I parked that work while I focused on the ansible baremetal provisioning. But when it was discussed in the first ironic meeting I took the results of that as enough of a consensus to move forward. But I will email the list before continuing | 02:13 |
| TheJulia | stevebaker: okay | 02:14 |
| TheJulia | stevebaker: thanks! | 02:15 |
| *** mkrai has joined #openstack-ironic | 03:04 | |
| *** ricolin_ has joined #openstack-ironic | 03:31 | |
| *** ricolin_ has quit IRC | 03:33 | |
| *** ricolin_ has joined #openstack-ironic | 03:34 | |
| *** yaawang has quit IRC | 03:44 | |
| *** yaawang has joined #openstack-ironic | 03:45 | |
| *** bdodd has joined #openstack-ironic | 04:00 | |
| openstackgerrit | Julia Kreger proposed openstack/ironic master: WIP ipv6 jobs https://review.opendev.org/613169 | 04:05 |
| *** mkrai has quit IRC | 04:21 | |
| *** mkrai has joined #openstack-ironic | 04:21 | |
| *** diurnalist has joined #openstack-ironic | 05:47 | |
| *** strigazi has quit IRC | 05:49 | |
| *** gudrutis2 has quit IRC | 05:50 | |
| *** strigazi has joined #openstack-ironic | 05:51 | |
| *** diurnalist has quit IRC | 05:52 | |
| arne_wiebalck | Good morning, ironic! | 06:38 |
| *** strigazi has quit IRC | 07:03 | |
| *** rpittau|afk is now known as rpittau | 07:17 | |
| rpittau | good morning ironic! o/ | 07:18 |
| iurygregory | good morning arne_wiebalck rpittau and Ironic o/ | 07:26 |
| arne_wiebalck | hey iurygregory rpittau o/ | 07:33 |
| rpittau | Hey iurygregory :) | 07:39 |
| iurygregory | arne_wiebalck, import six? O.o | 07:53 |
| arne_wiebalck | iurygregory: huh? | 07:53 |
| iurygregory | arne_wiebalck, https://review.opendev.org/#/c/696156/7/ironic_python_agent/utils.py | 07:53 |
| patchbot | patch 696156 - ironic-python-agent - Software RAID: Add UEFI support - 7 patch sets | 07:53 |
| iurygregory | I'm reviewing your patch and that was the first thing I saw =D | 07:53 |
| arne_wiebalck | don't punish the messenger! | 07:54 |
| rpittau | the reason is probably because that patch started in november when we still had six | 07:55 |
| arne_wiebalck | probably | 07:55 |
| iurygregory | rpittau, yeah | 07:55 |
| arne_wiebalck | and I have tested this on Stein on Python 2 | 07:55 |
| arne_wiebalck | tested with real hardware I mean | 07:55 |
| iurygregory | oh so this can be backported? | 07:55 |
| rpittau | no, in theory not | 07:56 |
| rpittau | at least not upstream | 07:56 |
| iurygregory | oh so no six :D | 07:56 |
| arne_wiebalck | it can, I have a Stein backport, but this should not go upstream, | 07:56 |
| arne_wiebalck | and we will use this on Train as well I guess | 07:56 |
| iurygregory | arne_wiebalck, gotcha, you will probably need for py2 in stein =) | 07:56 |
| arne_wiebalck | but also this will be downstream only | 07:56 |
| rpittau | arne_wiebalck: train should be fine | 07:56 |
| iurygregory | train stills supports py2 so in the backport you would need to "solve conflicts" adding six etc | 07:57 |
| * arne_wiebalck has still not moved to Train | 07:57 | |
| iurygregory | wait a bit to move to train, since the train can lose control | 07:58 |
| arne_wiebalck | we will need to do some cleanup on this area anyway, like all the uefi code that has been done meanwhile | 07:58 |
| openstackgerrit | Riccardo Pittau proposed openstack/bifrost master: Fix dhcp job https://review.opendev.org/712304 | 07:59 |
| arne_wiebalck | my main aim would be to get this in, as the current code seems to be working, and then cleanup ... yeah, I know :-D | 07:59 |
| arne_wiebalck | there was also some refactoring suggested by TheJulia and dtantsur|afk | 08:00 |
| * rpittau goes and finish breakfast | 08:00 | |
| arne_wiebalck | I am just not clear I will be able to test all this again (as this is quite work intensive, the UEFI node has meanwhile been taken from me, ...) | 08:01 |
| arne_wiebalck | anyway, please leave your comments, this will be helpful for the refactor/cleanup | 08:01 |
| arne_wiebalck | I'll try to keep track | 08:02 |
| *** sziviani_ has joined #openstack-ironic | 08:04 | |
| *** amoralej|off is now known as amoralej | 08:06 | |
| *** tesseract has joined #openstack-ironic | 08:08 | |
| iurygregory | arne_wiebalck, sure o/ | 08:11 |
| *** Algeroth has joined #openstack-ironic | 08:19 | |
| Algeroth | Hello, anyone faced a problem with UEFI baremetal deployment "image.install_bootloader, params: {u'efi_system_part_uuid': u'3520-DD51', u'prep_boot_part_uuid': None, u'root_uuid': u'28984a52-5dd9-41c8-ac33-4175fe44f5c3'}, status: FAILED, result: None." googling it already, but didn't found any solution | 08:31 |
| Algeroth | detailed error "ironic-python-agent[10011]: 2020-03-11 12:03:21.647 10011 ERROR root [-] Command execution error: invalid literal for int() with base 10: 'p1': ValueError: invalid literal for int() with base 10: 'p1'" | 08:32 |
| iurygregory | Algeroth, you are probably using an old version of the ironic-python-agent | 08:34 |
| *** k_mouza has joined #openstack-ironic | 08:44 | |
| *** k_mouza has quit IRC | 08:49 | |
| Algeroth | iurygregory, was thinking about it... will try to build imae again and check, but i'm afraid i did it already | 08:51 |
| *** rpittau is now known as rpittau|bbl | 08:52 | |
| iurygregory | Algeroth, I only saw this error when we were testing patches (before merging in the ironic-python-agent repo) | 08:52 |
| *** khansa has joined #openstack-ironic | 08:53 | |
| *** lucasagomes has joined #openstack-ironic | 09:12 | |
| *** alexmcleod has joined #openstack-ironic | 09:13 | |
| *** dtantsur|afk is now known as dtantsur | 09:13 | |
| dtantsur | morning ironic | 09:13 |
| iurygregory | morning dtantsur | 09:14 |
| iurygregory | dtantsur, if you have sometime https://storyboard.openstack.org/#!/story/2007416 does it look ok? | 09:15 |
| dtantsur | iurygregory: yep | 09:16 |
| iurygregory | dtantsur, tks o/ | 09:16 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic stable/train: Change force_raw_images to use sha256 if md5 is selected https://review.opendev.org/712627 | 09:24 |
| *** priteau has joined #openstack-ironic | 09:25 | |
| openstackgerrit | Mark Goddard proposed openstack/bifrost master: Fix Ubuntu keystone deployment https://review.opendev.org/712112 | 09:29 |
| *** k_mouza has joined #openstack-ironic | 09:34 | |
| *** sshnaidm|afk is now known as sshnaidm | 09:39 | |
| *** derekh has joined #openstack-ironic | 09:41 | |
| openstackgerrit | Merged openstack/ironic-python-agent master: Amending ntp time sync release note https://review.opendev.org/712552 | 09:41 |
| mgoddard | morning ironic | 09:42 |
| iurygregory | morning mgoddard | 09:44 |
| *** ociuhandu has joined #openstack-ironic | 09:59 | |
| openstackgerrit | khansa proposed openstack/sushy-cli master: Add system BIOS management command https://review.opendev.org/711541 | 10:02 |
| openstackgerrit | Mark Goddard proposed openstack/bifrost master: Fix Ubuntu keystone deployment https://review.opendev.org/712112 | 10:19 |
| *** ociuhandu has quit IRC | 10:25 | |
| *** ociuhandu has joined #openstack-ironic | 10:28 | |
| arne_wiebalck | dtantsur: would you have a moment for some advice concerning the UEFI RAID patch? | 10:38 |
| dtantsur | arne_wiebalck: I can try | 10:38 |
| arne_wiebalck | https://review.opendev.org/#/c/696156 | 10:39 |
| patchbot | patch 696156 - ironic-python-agent - Software RAID: Add UEFI support - 7 patch sets | 10:39 |
| arne_wiebalck | dtantsur: have a look at iurygregory's last comment and my reply, please | 10:39 |
| openstackgerrit | Verification of a change to openstack/bifrost failed: Ensure DIB is installed from source https://review.opendev.org/712120 | 10:40 |
| dtantsur | arne_wiebalck: https://review.opendev.org/#/c/696156/7/ironic_python_agent/extensions/image.py@696 ? | 10:41 |
| patchbot | patch 696156 - ironic-python-agent - Software RAID: Add UEFI support - 7 patch sets | 10:41 |
| arne_wiebalck | dtantsur: yres | 10:42 |
| arne_wiebalck | dtantsur: software RAID needs the bootloader to be installed no matter what, but the logic added in between the original patch and the rebasetakes a shortcut | 10:43 |
| arne_wiebalck | dtantsur: for secure boot | 10:43 |
| dtantsur | arne_wiebalck: why do we need it so strictly? | 10:43 |
| arne_wiebalck | dtantsur: I guess you won't be able to boot if the bios does not find the efi partition (which is a partition inside a RAID device on top of a partition) | 10:44 |
| arne_wiebalck | dtantsur: that's why Raphael's code relocates the contents of the EFI partition | 10:45 |
| iurygregory | not only for secure boot, but for all deployments where the target is uefi.. | 10:48 |
| arne_wiebalck | it is needed for secure boot, but is now done for all UEFI deployments | 10:48 |
| iurygregory | yeah | 10:48 |
| *** ociuhandu has quit IRC | 10:49 | |
| iurygregory | the easy way would be have a flag that would identify software raid deployment so we would go directly to _install_grub2 =) | 10:50 |
| *** ociuhandu has joined #openstack-ironic | 10:50 | |
| arne_wiebalck | a "force" flag to install_bootloader, to enforce the installation of the bootloader | 10:51 |
| dtantsur | arne_wiebalck: I think we have a way to detect sw RAID in IPA already? | 10:51 |
| arne_wiebalck | which would be set by the conductor when we are handling a s/w RAID | 10:51 |
| dtantsur | even if we do relocate the EFI partition, we may still try efibootmgr? or not? | 10:52 |
| arne_wiebalck | dtantsur: there is some code to detect a s/w RAID, yes ... let me check | 10:53 |
| arne_wiebalck | dtantsur: but what about partition images? | 10:53 |
| arne_wiebalck | dtantsur: wouldn't these need this as well? | 10:53 |
| iurygregory | partition images will go to grub2 afaik | 10:53 |
| *** ociuhandu has quit IRC | 10:53 | |
| arne_wiebalck | how? | 10:53 |
| iurygregory | because it won't find the efibootmgr if I do remember | 10:54 |
| arne_wiebalck | doesn't this depend on the image? | 10:54 |
| arne_wiebalck | the code checks whether efibootmanager is available in the ramdisk from what I see | 10:56 |
| *** ociuhandu has joined #openstack-ironic | 10:58 | |
| arne_wiebalck | install_bootloader is called for s/w RAID and partition images and whole disk images with UEFI | 10:58 |
| arne_wiebalck | on the conductor side, I mean | 10:58 |
| iurygregory | humm | 10:59 |
| arne_wiebalck | https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/agent_base.py#L959 | 11:00 |
| *** mkrai has quit IRC | 11:02 | |
| *** mkrai_ has joined #openstack-ironic | 11:02 | |
| *** ociuhandu has quit IRC | 11:03 | |
| arne_wiebalck | here is the RAID detector: https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/hardware.py#L189 | 11:03 |
| arne_wiebalck | so we could extend the if-statement with "and not hardware.is_md_device(device)" | 11:06 |
| arne_wiebalck | this would not solve the issue for partition images (and secure boot would probably not work with s/w RAID either) | 11:08 |
| *** derekh has quit IRC | 11:10 | |
| iurygregory | looking | 11:11 |
| openstackgerrit | Mark Goddard proposed openstack/bifrost master: Fix Ubuntu keystone deployment https://review.opendev.org/712112 | 11:11 |
| *** sshnaidm has quit IRC | 11:18 | |
| *** derekh has joined #openstack-ironic | 11:20 | |
| *** ociuhandu has joined #openstack-ironic | 11:20 | |
| openstackgerrit | Ilya Etingof proposed openstack/sushy master: Ignore failing message registry download https://review.opendev.org/712655 | 11:22 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: Generalize clean step functions to support deploy steps https://review.opendev.org/711437 | 11:22 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: Remove compatibility with pre-deploy steps drivers https://review.opendev.org/711918 | 11:25 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: Generalize clean step functions to support deploy steps https://review.opendev.org/711437 | 11:25 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: WIP: In-band deploy steps https://review.opendev.org/698768 | 11:25 |
| *** ociuhandu has quit IRC | 11:27 | |
| *** ociuhandu has joined #openstack-ironic | 11:27 | |
| *** mkrai_ has quit IRC | 11:37 | |
| openstackgerrit | Mark Goddard proposed openstack/tenks master: WIP: Support linux bridge in addition to OVS https://review.opendev.org/711682 | 11:43 |
| openstackgerrit | Mark Goddard proposed openstack/tenks master: WIP: Support linux bridge in addition to OVS https://review.opendev.org/711682 | 11:48 |
| *** ociuhandu has quit IRC | 11:54 | |
| *** ociuhandu has joined #openstack-ironic | 11:55 | |
| *** ociuhandu has quit IRC | 11:55 | |
| *** ociuhandu has joined #openstack-ironic | 11:55 | |
| dtantsur | if any stable core has a minute: https://review.opendev.org/#/c/712627/ | 12:05 |
| patchbot | patch 712627 - ironic (stable/train) - Change force_raw_images to use sha256 if md5 is se... - 1 patch set | 12:05 |
| dtantsur | rpittau|bbl: ^^ | 12:05 |
| openstackgerrit | Mark Goddard proposed openstack/tenks master: WIP: Support linux bridge in addition to OVS https://review.opendev.org/711682 | 12:18 |
| openstackgerrit | Mark Goddard proposed openstack/bifrost master: Fix Ubuntu keystone deployment https://review.opendev.org/712112 | 12:21 |
| *** rh-jelabarre has joined #openstack-ironic | 12:24 | |
| *** rcernin has quit IRC | 12:25 | |
| *** tkajinam has quit IRC | 12:27 | |
| *** rloo has joined #openstack-ironic | 12:30 | |
| *** rloo has quit IRC | 12:32 | |
| *** rloo has joined #openstack-ironic | 12:33 | |
| openstackgerrit | Mark Goddard proposed openstack/tenks master: Stop gzipping logs in get-logs.sh https://review.opendev.org/712668 | 12:33 |
| *** Lucas_Gray has joined #openstack-ironic | 12:43 | |
| *** sziviani_ has quit IRC | 12:47 | |
| *** jerry2 has quit IRC | 12:48 | |
| *** jerry2 has joined #openstack-ironic | 12:49 | |
| *** Lucas_Gray has quit IRC | 12:49 | |
| *** ociuhandu has quit IRC | 12:50 | |
| *** ociuhandu has joined #openstack-ironic | 12:51 | |
| *** Lucas_Gray has joined #openstack-ironic | 12:53 | |
| *** mmethot has joined #openstack-ironic | 12:56 | |
| *** ociuhandu has quit IRC | 12:57 | |
| *** ociuhandu has joined #openstack-ironic | 13:00 | |
| *** rpittau|bbl is now known as rpittau | 13:02 | |
| rpittau | lol dhcp job works but non-dhcp is broken | 13:06 |
| *** Algeroth has quit IRC | 13:10 | |
| *** Lucas_Gray has quit IRC | 13:10 | |
| *** ianychoi has quit IRC | 13:13 | |
| iurygregory | O.o | 13:13 |
| *** amoralej is now known as amoralej|lunch | 13:15 | |
| *** ktibi has joined #openstack-ironic | 13:15 | |
| *** ociuhandu has quit IRC | 13:15 | |
| ktibi | Hi guys, I try to deploy a HP gen10 with ilo driver. Inspection works but power control fail with error : "Failed to change power state to 'power off'. Error: _op() got an unexpected keyword argument 'allow_redirects' " | 13:15 |
| *** ociuhandu has joined #openstack-ironic | 13:16 | |
| ktibi | inspector is already in debug mode but nothing in log :/ any idea ? | 13:16 |
| *** ociuhandu has quit IRC | 13:21 | |
| *** k_mouza has quit IRC | 13:21 | |
| *** ociuhandu has joined #openstack-ironic | 13:22 | |
| *** derekh has quit IRC | 13:29 | |
| *** ociuhandu has quit IRC | 13:30 | |
| TheJulia | good morning | 13:30 |
| *** ijw has quit IRC | 13:31 | |
| etingof | TheJulia o/ | 13:31 |
| TheJulia | ktibi: Oo | 13:31 |
| etingof | ktibi, may be your sushy is not fresh enough? | 13:31 |
| dtantsur | morning TheJulia | 13:31 |
| TheJulia | ktibi: any chance you can provide us version s of ironic, sushy, python-proliantutils, and possibly a bit more of the logs? | 13:31 |
| TheJulia | etingof: you win for making a code reference that sounds like a food reference | 13:32 |
| *** ociuhandu has joined #openstack-ironic | 13:32 | |
| ktibi | TheJulia, etingof I use rocky last stable version. Proliantutils is in 2.7.1 | 13:32 |
| ktibi | sushy in 1.6.1 | 13:33 |
| ktibi | HP ILO is in 1.45 version | 13:33 |
| * TheJulia wishes ilo developers hanged out here through the week | 13:33 | |
| * TheJulia goes and gets the magical coffee from the dispencer of wake up juice | 13:34 | |
| etingof | this is where it might fail -- https://github.com/openstack/sushy/blob/stable/rocky/sushy/connector.py#L65 | 13:35 |
| etingof | because since stein it accepts **kwargs -- https://github.com/openstack/sushy/blob/stable/stein/sushy/connector.py#L64 | 13:36 |
| rpittau | ktibi: ou'll need a more recent version of proliantutils https://opendev.org/x/proliantutils/commit/0196cd316236ca48f41ace48c662f4735b574b7b | 13:36 |
| *** dtantsur is now known as dtantsur|brb | 13:36 | |
| rpittau | or an older version of sushy :) | 13:36 |
| ktibi | rpittau, I use sushy in 1.6, the version if affect ? | 13:37 |
| TheJulia | the issue is in proliantutils it looks like | 13:38 |
| * dtantsur|brb -> walk | 13:38 | |
| ktibi | can update proliantutils and sushy and keep ironic in rocky version ? | 13:38 |
| *** mkrai_ has joined #openstack-ironic | 13:39 | |
| etingof | is it the right version of proliantutils for rocky? | 13:39 |
| TheJulia | I'm not sure, looking because that call is brand new with in the last 3 weeks in proliant utils | 13:40 |
| rpittau | ktibi: you'll need at least version 2.8.0 of proliantutils | 13:41 |
| iurygregory | good morning TheJulia | 13:41 |
| etingof | on the other hand, the patch rpittau linked does not change allows_redirects, does it? | 13:41 |
| rpittau | etingof: no, that's for sushy 3.10 | 13:41 |
| TheJulia | so.. there is no cap on proliantutils | 13:41 |
| rpittau | the patch involved is this one https://opendev.org/x/proliantutils/commit/254318f608577523c1727d42ca05e1cf258b832b | 13:41 |
| ktibi | in rocky requierment => proliantutils>=2.6.0 | 13:41 |
| ktibi | no version fix on sushy | 13:42 |
| ktibi | so I think I can update sushy + proliantutils | 13:42 |
| TheJulia | in theory, you might see some issues with the more fine grained items we expect in the API, but the ilos offer what may be the most compete redfish API | 13:43 |
| etingof | but even master does not have **kwargs in the overridden `_op` -- https://opendev.org/x/proliantutils/src/branch/master/proliantutils/redfish/connector.py#L38 | 13:45 |
| TheJulia | so I guess the real question is... where is the _op that things are going sideways | 13:46 |
| *** amoralej|lunch is now known as amoralej | 13:46 | |
| etingof | right, there are two possibilities... | 13:46 |
| etingof | we will never know unless ktibi reveals the traceback ;) | 13:47 |
| ktibi | etingof, My inspector is in debug mode but nothing :/ | 13:47 |
| etingof | why inspector, not conductor? | 13:49 |
| TheJulia | I'd look in the conductor log because it is telling ironic "hey, power this off" if memory serves | 13:49 |
| ktibi | etingof, oh I'll inspcet conductor log | 13:50 |
| etingof | on the other hand, there is no 'allow_redirects' in ironic/sushy codebase presently, it's only in proliantutils | 13:51 |
| *** ijw has joined #openstack-ironic | 13:51 | |
| etingof | that hints this comes from proliantutils calling older sushy and passing `allow_redirects` to ir | 13:52 |
| ktibi | ha :) | 13:52 |
| TheJulia | but the version doesn't match, so.... unkown | 13:52 |
| TheJulia | known | 13:52 |
| * TheJulia focuses on caffination | 13:53 | |
| etingof | if there is proliantutils version that does not pass `allow_redirects`, it should work with your sushy | 13:53 |
| ktibi | login failed >< | 13:53 |
| etingof | or you could try to bump up sushy to at least the version that accepts **kwargs in _op() | 13:53 |
| rpittau | 2.7.1 should not have allow_redirects | 13:53 |
| *** jdandrea has joined #openstack-ironic | 13:54 | |
| ktibi | Got invalid response with message: 'Login failed.' | 13:54 |
| ktibi | http://paste.openstack.org/show/790598/ | 13:54 |
| rpioso | Good morning, ironic | 13:54 |
| rpioso | etingof: Regarding https://bugzilla.redhat.com/show_bug.cgi?id=1804127#c11, which version of sushy was used? | 13:55 |
| openstack | bugzilla.redhat.com bug 1804127 in python-sushy "Importing results in "Error: Cannot mix str and non-str arguments"" [Medium,On_dev] - Assigned to ietingof | 13:55 |
| etingof | ktibi, this URL does not smell redfishy -- https://10.126.126.201:443/ribcl | 13:56 |
| *** ijw has quit IRC | 13:56 | |
| rpioso | etingof: mraineri asked me, "Does the "OData-Version" get supplied by Ironic when making an HTTP request to the service?" | 13:56 |
| etingof | ktibi, may be you have something wrong in node config? | 13:57 |
| ktibi | etingof, I'll paste the node show | 13:57 |
| * etingof has no experience with ilo config | 13:57 | |
| ktibi | http://paste.openstack.org/show/790600/ | 13:58 |
| ktibi | I masked mac and serial | 13:59 |
| etingof | rpioso, o/ that might be train sushy | 13:59 |
| rpioso | etingof: He went on to say some implementations have keyed off something very subtle in the spec with regards to how links are presented, including the one used in that bugz. And, "If "OData-Version" was NOT provided, they'd return with links in their own format. But if it was provided, then they'd be Redfish conformant." | 13:59 |
| rpioso | etingof: Yeah, I saw the bugz mentioned train, but you never know, right? | 14:00 |
| etingof | nooo, with software you never know what's happening under the hood | 14:00 |
| etingof | rpioso, that's interesting, off the top of my head, I am not certain we cap that odata version | 14:01 |
| rpioso | etingof: That header is not present in 1.x.x versions of sushy. | 14:01 |
| rpioso | s/present/sent/ | 14:01 |
| rpioso | etingof: mraineri says there's only one version. | 14:02 |
| rpioso | 4.0 | 14:02 |
| openstackgerrit | Riccardo Pittau proposed openstack/ironic-inspector master: Adapt ironic inspector jobs to ironic dib-based jobs https://review.opendev.org/712516 | 14:03 |
| *** derekh has joined #openstack-ironic | 14:03 | |
| etingof | rpioso, is it that we must always send odata-version=4.0? | 14:03 |
| etingof | ...and all our redfishy problems magically disappear | 14:03 |
| * rpioso taps elbows with mraineri | 14:03 | |
| etingof | please, confirm | 14:03 |
| TheJulia | ktibi: etingof so looks like the _op is from deep inside sushy | 14:04 |
| rpioso | etingof: mraineri will be joining us shortly. | 14:05 |
| ktibi | TheJulia, I add a new right on ILO interface for my user, and know I think it's good (missing login privileges) | 14:06 |
| ktibi | the server is always power on but no error in conductor | 14:06 |
| ktibi | the answer is good : http://paste.openstack.org/show/790601/ | 14:07 |
| openstackgerrit | khansa proposed openstack/sushy-cli master: Add system BIOS management command https://review.opendev.org/711541 | 14:07 |
| TheJulia | i bet when it was trying to touch it with redfish as a fallback to ribcl, it was getting a redirect and things were going sideways then | 14:07 |
| * TheJulia really wishes we had a complete stack trace | 14:08 | |
| etingof | ktibi, that request/response is iLO response, not redfish request | 14:09 |
| etingof | TheJulia, for the whole thing to burp up no redirect response is requires, it seems | 14:10 |
| ktibi | etingof, I can try to upgrade proliantutils to 2.9.0 | 14:10 |
| etingof | any redfish request made through proliantutils should hit that problem on its way to sushy | 14:10 |
| openstackgerrit | Riccardo Pittau proposed openstack/bifrost master: [DNM] test ci https://review.opendev.org/712692 | 14:10 |
| *** mraineri has joined #openstack-ironic | 14:10 | |
| *** ociuhandu has quit IRC | 14:11 | |
| ktibi | etingof, tips for debug redfish request ? or extract a log ? | 14:13 |
| TheJulia | etingof: I think it was failing, and falling back | 14:13 |
| TheJulia | and then getting redirected | 14:14 |
| TheJulia | and then boom | 14:14 |
| * TheJulia has grown tired of ipmitool debugging output | 14:14 | |
| *** sziviani_ has joined #openstack-ironic | 14:14 | |
| TheJulia | creepy is: ordering two new disks my supermicro server using amazon... and now getting adds in one of my mailboxes for the exact same thing I ordered, but from newegg. | 14:15 |
| mraineri | etingof: I hear you have some questions about the OData-Version header? | 14:17 |
| mraineri | So, up until version 1.6.0 of the specification, OData-Version was required to be sent | 14:18 |
| ktibi | etingof, TheJulia I upgrade proliantutils in 2.9.0, sushi upgrade to 3.1.0 :/ | 14:18 |
| mraineri | HPE had a little hook in place to key off the presence of the OData-Version header; if it was provided by the client, then do everything 100% Redfish, but if not, then revert to their own pre-Redfish API | 14:19 |
| openstackgerrit | khansa proposed openstack/sushy-cli master: Add system BIOS management command https://review.opendev.org/711541 | 14:19 |
| mraineri | At least until version 1.6.0 of the spec removed the client requirement for sending the header, the argument was "if the client doesn't send it, then they're operating outside the Redfish spec" | 14:19 |
| rpioso | mraineri: etingof asked, "is it that we must always send odata-version=4.0? ...and all our redfishy problems magically disappear" | 14:20 |
| mraineri | For the sake of safety, yes, it should always be sent | 14:20 |
| mraineri | That'll let you work with pre-1.6.0 implementations | 14:20 |
| mraineri | For pre-1.6.0 implementations, it is required for the client to send OData-Version | 14:21 |
| mraineri | On all requests | 14:21 |
| *** ijw has joined #openstack-ironic | 14:22 | |
| openstackgerrit | Julia Kreger proposed openstack/ironic master: Move ipmi logging to a separate option https://review.opendev.org/712695 | 14:24 |
| TheJulia | bfournie: ^^ | 14:25 |
| bfournie | TheJulia: nice \o/ | 14:26 |
| bfournie | TheJulia: will make people happy | 14:26 |
| * TheJulia has grown tired of the errors | 14:26 | |
| TheJulia | iurygregory: are you aware of master branch current state grenade just falling down while executing? | 14:27 |
| *** ijw has quit IRC | 14:27 | |
| ktibi | so, I tried with last version of proliantutils 2.9.4 and sushy 3.1, nothing better :/ | 14:28 |
| ktibi | all is good for ironic, but server is always power on | 14:29 |
| iurygregory | TheJulia, will look now, I was in a meeting about COVID19 instructions for people in CZ =) | 14:32 |
| rpittau | I can't believe we hit another issue with setuptools + pip :/ | 14:32 |
| iurygregory | rpittau, it's life =( | 14:33 |
| etingof | CZ is almost locked down by now, apparently | 14:33 |
| iurygregory | yeah | 14:36 |
| * iurygregory will only go to the office to grab the monitor to take home hehehe | 14:37 | |
| ktibi | TheJulia, I see on doc, I can use redfish driver directly instead of ilo, maybe works better ? | 14:37 |
| rpittau | I'm surprised and worried that here we're still not in lockdown | 14:38 |
| TheJulia | ktibi: I really suspect you need to downgrade, tbh as rpittau suggested originally | 14:38 |
| TheJulia | you _should_ be able to use the redfish driver directly though. | 14:38 |
| ktibi | to 1.6.0 for proliantutils ? | 14:38 |
| TheJulia | I think 1.6.0 was broken, the latest 1.6.x is likely the best place to be | 14:39 |
| TheJulia | rpittau: I'm wishing we basically were, but instead the government banned the EU *facepalm* | 14:40 |
| TheJulia | iurygregory: monitor ++ | 14:40 |
| rpittau | they have elections here on sunday, it;s going to be a bloodbath | 14:40 |
| ktibi | I can't find 1.6.0, 0.1.1 ==> 2.0.0 | 14:41 |
| *** ricolin_ has quit IRC | 14:41 | |
| TheJulia | ugh | 14:41 |
| *** ricolin_ has joined #openstack-ironic | 14:41 | |
| ktibi | maybe talk about sushy ? | 14:41 |
| TheJulia | if anyone has stock in the US, just... don't look at the markets right now | 14:41 |
| * TheJulia wonders if the ilo folks pulled older versions | 14:42 | |
| *** ociuhandu has joined #openstack-ironic | 14:42 | |
| rpittau | EU is the same :( | 14:42 |
| rpittau | yeah proliantutils version starts from 2.0.0 | 14:42 |
| *** gyee has joined #openstack-ironic | 14:43 | |
| * TheJulia sighs | 14:44 | |
| *** ociuhandu has quit IRC | 14:47 | |
| *** TxGirlGeek has joined #openstack-ironic | 14:48 | |
| *** ricolin_ has quit IRC | 14:48 | |
| rpittau | ktibi: sorry, I'm going a bit blind here, but can you try with proliantutils 2.6.1 ? | 14:48 |
| * rpioso hopes etingof is free :-) | 14:49 | |
| rpioso | etingof: Did you see mraineri's messages above? | 14:49 |
| etingof | rpioso, yes | 14:50 |
| openstackgerrit | khansa proposed openstack/sushy-cli master: Add system BIOS management command https://review.opendev.org/711541 | 14:50 |
| rpioso | etingof: +1. Thx :-) | 14:50 |
| ktibi | I try to debug with the doc of pypi, first command in error "from proliantutils.ilo import client' | 14:51 |
| etingof | rpioso, let me dig into sushy to see how it behaves in odata respect | 14:52 |
| etingof | mraineri, thanks for the clarification! | 14:52 |
| mraineri | No problem | 14:52 |
| ktibi | http://paste.openstack.org/show/790607/ | 14:52 |
| *** diurnalist has joined #openstack-ironic | 14:53 | |
| *** ijw has joined #openstack-ironic | 14:53 | |
| ktibi | ok need python 3 >< | 14:54 |
| rpittau | ktibi: which version of sushy is that? doesn't look compatible with python 2 | 14:54 |
| ktibi | 3.1 | 14:54 |
| rpittau | yeah, that's py3 only | 14:54 |
| ktibi | so whats last version of sushy compatible with python 2 ? | 14:55 |
| rpittau | any version up to train | 14:55 |
| rpittau | ktibi: 2.0.2 should work | 14:56 |
| *** ociuhandu has joined #openstack-ironic | 14:57 | |
| ktibi | rpittau, yes import works ;) | 14:57 |
| rpittau | eheh ok :) | 14:57 |
| ktibi | I'll try manual power off with python command | 14:57 |
| *** ijw has quit IRC | 14:58 | |
| *** uzumaki has joined #openstack-ironic | 14:59 | |
| openstackgerrit | Mark Goddard proposed openstack/tenks master: Force Python 2 for virtualenvs on CentOS 7 https://review.opendev.org/712703 | 14:59 |
| *** Lucas_Gray has joined #openstack-ironic | 15:00 | |
| *** ociuhandu has quit IRC | 15:02 | |
| ktibi | get_host_power_state() works :/ | 15:03 |
| iurygregory | always nova tests failling in master .-. | 15:05 |
| iurygregory | I'm checking differences in configuration for one patch that has green and red for the job to see if something went wrong | 15:06 |
| openstackgerrit | Merged openstack/ironic stable/train: Change force_raw_images to use sha256 if md5 is selected https://review.opendev.org/712627 | 15:06 |
| rpittau | fantastic, the bug with setuptools is in 3.6.8 but not 3.6.9 :D | 15:06 |
| iurygregory | rpittau, good! | 15:09 |
| rpittau | iurygregory: not good, bionic has 3.6.9, but centos8 has 3.6.8 | 15:09 |
| iurygregory | rpittau, oh okay, but we can upgrade no? | 15:09 |
| rpittau | I'm not sure there's an official package, I found a workaround though | 15:10 |
| * rpittau tea time | 15:12 | |
| *** mkrai_ has quit IRC | 15:13 | |
| *** mkrai has joined #openstack-ironic | 15:13 | |
| *** Lucas_Gray has quit IRC | 15:14 | |
| *** ijw has joined #openstack-ironic | 15:16 | |
| *** Lucas_Gray has joined #openstack-ironic | 15:16 | |
| *** Goneri has joined #openstack-ironic | 15:18 | |
| *** Lucas_Gray has quit IRC | 15:30 | |
| *** Lucas_Gray has joined #openstack-ironic | 15:32 | |
| rpioso | etingof: Sounds good. Stay safe! | 15:36 |
| etingof | rpittau, mraineri seems like we have it in place since Train -- https://github.com/openstack/sushy/blob/stable/train/sushy/connector.py#L83 | 15:36 |
| etingof | rpioso, ^ | 15:37 |
| etingof | that BZ is supposedly about Train, so odata header should be sent... I will ask there to be sure | 15:39 |
| rpioso | etingof: It's in all releases >= 2.0.0. It is absent from all *-em extended maintenance tags. | 15:39 |
| rpioso | etingof: Right! One never knows. | 15:40 |
| * rpioso suspects there's a good deal of fishiness in the ocean | 15:41 | |
| * TheJulia wonders how long before we create a redfish api validation tool... maybe call it acidification ? | 15:41 | |
| rpittau | What about "redhook" ? | 15:42 |
| * TheJulia may be calling to the current state of the ocean and http://acid3.acidtests.org/ | 15:42 | |
| etingof | do not we have a story for that? | 15:44 |
| etingof | by arne_wiebalck may be? | 15:44 |
| arne_wiebalck | there is | 15:45 |
| *** TxGirlGeek has quit IRC | 15:45 | |
| arne_wiebalck | https://storyboard.openstack.org/#!/story/2007345 | 15:46 |
| etingof | the Golden Fish award has been proposed to compliant implementations | 15:47 |
| *** TxGirlGeek has joined #openstack-ironic | 15:47 | |
| etingof | the fishy challenge: how long does it take to read power state from iDRAC over Redfish? | 15:50 |
| etingof | spoiler link -- http://paste.openstack.org/show/790615/ | 15:50 |
| mraineri | TheJulia: There are actually a few different validation tools out there at the moment | 15:52 |
| mraineri | This tool verifies all payloads read from the service via GET conform to the schema definitions: https://github.com/DMTF/Redfish-Service-Validator | 15:52 |
| mraineri | I know that one gets heavily beaten up by different vendors | 15:53 |
| mraineri | This tool does more protocol types of testing (like HTTP header verification): https://github.com/DMTF/Redfish-Service-Conformance-Check | 15:53 |
| mraineri | That one hasn't been as successful though and we have some internal discussions going on now for what to do with it | 15:53 |
| mraineri | This one is a set of tests that does more functional testing with a service based on expected client workflows: https://github.com/DMTF/Redfish-Usecase-Checkers | 15:54 |
| mraineri | What's in there now is stable and useful, but it definitely needs to grow more (feel free to open issues with test ideas!) | 15:55 |
| TheJulia | dtantsur|brb: to answer the checksum question, it seems we have no positive path tests, looking at adding one. | 15:55 |
| TheJulia | Interesting | 15:55 |
| TheJulia | thanks mraineri | 15:55 |
| mraineri | No problem | 15:55 |
| mraineri | Definitely would like more eyes on the tools and feedback | 15:55 |
| *** dtantsur|brb is now known as dtantsur | 15:56 | |
| mraineri | Maybe a pull request or two for new features ;) | 15:56 |
| TheJulia | sadly time is a major issue for me | 15:56 |
| mraineri | Eh, but the accountants treat engineering time as free! | 15:56 |
| dtantsur | heh | 15:57 |
| *** mkrai has quit IRC | 15:58 | |
| etingof | in the linked story the idea is to have an ironic-specific Redfish compliance checker, not a general purpose one | 16:00 |
| etingof | the goal is to quickly determine that given BMC would work with ironic (considering its quite limited use of Redfish atm) | 16:01 |
| TheJulia | ++ | 16:01 |
| mraineri | There are things available for that too | 16:03 |
| mraineri | Redfish has the concept of a "Redfish Profile" documented here: https://www.dmtf.org/sites/default/files/standards/documents/DSP0272_1.3.0.pdf | 16:03 |
| mraineri | Essentially it's a JSON document that describes baseline requirements for an implementation (like implementing the ComputerSystem resource with a certain set of required properties) | 16:04 |
| *** khansa has quit IRC | 16:04 | |
| mraineri | You could make one specific for what Ironic requires out of an implementation, and use it with a tool that we have to check the service against the profile | 16:04 |
| mraineri | https://github.com/DMTF/Redfish-Interop-Validator | 16:04 |
| dtantsur | OH, this is a great idea | 16:06 |
| etingof | mraineri, thanks for the heads up concerning these tools! | 16:06 |
| dtantsur | we absolutely should have our profile | 16:06 |
| etingof | mraineri, I've read the README but what's not clear is this: will it be able to execute Actions? | 16:07 |
| etingof | or, more generally, modify the resources that our profile requires | 16:07 |
| mraineri | It doesn't execute actions, but it should verify they exist | 16:07 |
| etingof | I am afraid that's insufficient for us... | 16:08 |
| mraineri | The same goes for writable properties; we have an outstanding item for a safe way to test it | 16:08 |
| dtantsur | etingof: insufficient, but a huge step forward | 16:08 |
| etingof | because life experience is that the vendors tend to announce things, but they might not always work as expected | 16:08 |
| mraineri | I think the more "Action" based testing would be useful for the Usecase tools | 16:09 |
| etingof | having profile is a good thing, but it won't answer the simple question: if I use ironic in my cloud, should I buy this machine? | 16:09 |
| mraineri | Since that's designed more for functional testing | 16:09 |
| dtantsur | etingof: it will, however, tell you which machines NOT to buy | 16:10 |
| mraineri | It could be part of a purchasing spec; one of the check boxes could be "Support the Ironic Redfish profile" | 16:10 |
| etingof | that's why I think that ironic profile is better than nothing, but, again, the devil is in details | 16:11 |
| dtantsur | unrelated: has anyone tested our CI with cirros 0.5.1? rpittau? | 16:11 |
| rpittau | dtantsur: mmm no, not really | 16:11 |
| dtantsur | okay, I'll through a patch | 16:11 |
| *** uzumaki has quit IRC | 16:11 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: DNM test with cirros 0.5.1 https://review.opendev.org/712728 | 16:12 |
| dtantsur | WAIT, it has Grub2?? | 16:13 |
| dtantsur | I'll test boot_option:local if this ^^ passes | 16:14 |
| *** ociuhandu has joined #openstack-ironic | 16:15 | |
| openstackgerrit | Riccardo Pittau proposed openstack/bifrost master: Fix for PEP517 issue with Python 3.6.8 https://review.opendev.org/712692 | 16:17 |
| rpittau | so bifrost CI with centos8 is broken, this ^ should fix the issue | 16:17 |
| rpittau | and too much stuff, wrong rebase | 16:18 |
| *** ijw has quit IRC | 16:19 | |
| *** ociuhandu has quit IRC | 16:19 | |
| openstackgerrit | Riccardo Pittau proposed openstack/bifrost master: Fix for PEP517 issue with Python 3.6.8 https://review.opendev.org/712692 | 16:19 |
| rpittau | ok should be fine now :) | 16:20 |
| rpittau | or well we could also not upgrade pip at first, but I'm afraid 3.6.8 in centos8 comes with pip higher than 19.0 | 16:22 |
| *** ijw has joined #openstack-ironic | 16:24 | |
| openstackgerrit | Mark Goddard proposed openstack/networking-generic-switch master: Create and delete networks in parallel https://review.opendev.org/692404 | 16:24 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: WIP: Decompose core deploy step https://review.opendev.org/698309 | 16:28 |
| *** ijw has quit IRC | 16:29 | |
| openstackgerrit | Merged openstack/ironic master: Extend power sync timeout for Ericsson SDI https://review.opendev.org/712592 | 16:30 |
| *** Blinkiz has quit IRC | 16:30 | |
| *** Blinkiz has joined #openstack-ironic | 16:31 | |
| dtantsur | ugh, this in-band deploy steps work is already worse than the driver composition reform when it comes to eating my brain :) | 16:32 |
| rpittau | braaaains | 16:32 |
| iurygregory | brain images for rpittau | 16:35 |
| dtantsur | whole disk or partition? | 16:36 |
| TheJulia | new type | 16:36 |
| TheJulia | but maybe like a ramdisk | 16:36 |
| iurygregory | dtantsur, container images | 16:36 |
| dtantsur | oh, containers with brains? | 16:36 |
| dtantsur | yummy | 16:36 |
| * rpittau suffering and despair | 16:36 | |
| dtantsur | wait, it's not Friday yet, is it? | 16:36 |
| iurygregory | dtantsur, no =( | 16:36 |
| iurygregory | wait | 16:37 |
| dtantsur | sort of Friday for me, I have to travel to pick up my passport tomorrow.. | 16:37 |
| iurygregory | in Japan is Friday already | 16:37 |
| arne_wiebalck | TheJulia: did you see the s/w RAID, UEFI, efibootmgr discussion we had earlier? | 16:37 |
| TheJulia | arne_wiebalck: nope | 16:37 |
| TheJulia | arne_wiebalck: will it make me want to have whiskey? | 16:37 |
| arne_wiebalck | ha! | 16:37 |
| dtantsur | (yes) | 16:37 |
| arne_wiebalck | not strong enough | 16:37 |
| iurygregory | https://review.opendev.org/#/c/696156/7/ironic_python_agent/extensions/image.py | 16:37 |
| patchbot | patch 696156 - ironic-python-agent - Software RAID: Add UEFI support - 7 patch sets | 16:37 |
| * rpittau grabs a random bottle | 16:37 | |
| * iurygregory grabs a beer | 16:38 | |
| TheJulia | so a summary is? | 16:38 |
| TheJulia | or do I just need to scroll up? | 16:38 |
| dtantsur | I have a 46% Highland Park, will it work? | 16:38 |
| arne_wiebalck | please have a look at the last two comments, from iurygregory and myself | 16:38 |
| iurygregory | dtantsur, it will | 16:39 |
| dtantsur | I'm prepared! | 16:39 |
| arne_wiebalck | everyone has a drink? let's go then | 16:39 |
| TheJulia | uhh, not prepared yet | 16:39 |
| TheJulia | was looking at the review | 16:39 |
| iurygregory | you should only look after having the drink D: | 16:40 |
| arne_wiebalck | while drinking, through the bottom of the glass | 16:40 |
| *** tesseract has quit IRC | 16:40 | |
| TheJulia | arne_wiebalck: for a short term your last comment makes sense | 16:40 |
| arne_wiebalck | the logic gets really twisted | 16:41 |
| TheJulia | Although, I'm still wondering HOW we get to the point where we're not running grub-install | 16:41 |
| TheJulia | or, grub2-install to be more precise | 16:41 |
| dtantsur | side question: is it possible to test secure boot in the CI? | 16:41 |
| TheJulia | And I suspect part of the question how EFI firmware handles it | 16:41 |
| TheJulia | dtantsur: in theory with ubuntu it should be possible | 16:42 |
| TheJulia | dtantsur: although, I think yolanda may tell me of issues there with network booting | 16:42 |
| dtantsur | TheJulia: well, the new cirros seems to have grub2, soooo | 16:42 |
| TheJulia | CentOS... I think may now have an actually signed shim | 16:42 |
| arne_wiebalck | TheJulia: since the boot mode is UEFI and efibootmgr may be in the image | 16:42 |
| TheJulia | dtantsur: does it have a signed shim though? | 16:42 |
| dtantsur | mmm, a good question | 16:42 |
| * dtantsur suspects no | 16:43 | |
| arne_wiebalck | TheJulia: and then we take a turn and never look back | 16:43 |
| dtantsur | anyway, we don't have to test netboot | 16:43 |
| TheJulia | They key really is the signed shim which is why I put the grub network booting job in, because it goes shim->grub2 | 16:43 |
| yolanda | so i was able to secure boot, by grabbing shim and grubx64 from centos rpms, then placing on the pxe directories | 16:43 |
| TheJulia | arne_wiebalck: but efibootmgr should be in the ramdisk...hmm | 16:44 |
| TheJulia | yolanda: that is awesome | 16:44 |
| arne_wiebalck | TheJulia: yes, hence we never run grub2 for s/w RAID | 16:44 |
| * TheJulia feels moderately lost at the moment | 16:45 | |
| * TheJulia should have gotten a drink | 16:45 | |
| dtantsur | I'm afraid I still cannot get why we need to run grub2 for software RAID | 16:45 |
| dtantsur | TheJulia++ I think that's what lets us down | 16:45 |
| TheJulia | I'm thinking the same as dtantsur, I'm just.. not groking it | 16:45 |
| TheJulia | perhaps we need to talk through this? | 16:45 |
| iurygregory | basically we don't run _install_grub2 anymore (in most scenarios)https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/agent_base.py#L959-L961 | 16:46 |
| iurygregory | for software raid this seems to be a problem =) | 16:46 |
| dtantsur | yeah, I cannot quite understand why? | 16:46 |
| dtantsur | why it is a problem | 16:46 |
| iurygregory | dtantsur, in this I need to agree with you =) | 16:46 |
| dtantsur | I thought I understood it today... but I did not. | 16:47 |
| dtantsur | We relocate the EFI partition from inside the image to outside RAID | 16:48 |
| dtantsur | This is what I hope I understood | 16:48 |
| iurygregory | since they have a reason to need that ... we would need to add a validation on the if "and not hardware.is_md_device(device)" <- if this ensures that SW Raid is in use ofc | 16:48 |
| dtantsur | I'm still not sure why we cannot just point EFI to the new partition? | 16:48 |
| dtantsur | iurygregory: we can, short term. Long term we need to have software RAID working with secure boot. | 16:49 |
| arne_wiebalck | I guess the BIOS would need to assemble the RAID to find it otherwise, no? | 16:49 |
| openstackgerrit | Merged openstack/tenks master: CI: Switch to kayobe-overcloud-centos8 https://review.opendev.org/712136 | 16:49 |
| iurygregory | dtantsur, agree | 16:49 |
| dtantsur | arne_wiebalck: okay, so you relocate the partition. Does it necessary require calling grub? | 16:49 |
| dtantsur | note that at least one person has indicated to me that he needed our efimgr patch to make UEFI work at all (no secure boot). I forgot why. | 16:50 |
| TheJulia | Well, so EFI firmware may be able to read the existing partitio | 16:50 |
| TheJulia | in the whole disk image | 16:50 |
| dtantsur | actually, with RAID-1 (?) we don't need to assemble RAID to read a partition | 16:50 |
| TheJulia | so for bios mode, it made sense to point the pointer to the bootloader in the mbr | 16:50 |
| TheJulia | dtantsur: bingo | 16:50 |
| dtantsur | and I think we have a requirement of using RAID-1 for root, no? | 16:50 |
| arne_wiebalck | yes | 16:51 |
| TheJulia | yes | 16:51 |
| * dtantsur is getting himself even more confused | 16:51 | |
| TheJulia | dtantsur: take a sip | 16:51 |
| dtantsur | TheJulia: I'm planning on some exercising soon(ish), so skipping drinks for now :) | 16:51 |
| dtantsur | you'll have to deal with sober me :) | 16:51 |
| TheJulia | lol | 16:51 |
| arne_wiebalck | I still doubt the bios would find it since it is a partition inside a partition | 16:52 |
| TheJulia | I really can't go off drinking, I have another meeting in a few hours | 16:52 |
| TheJulia | I think | 16:52 |
| TheJulia | arne_wiebalck: its not looking at partitions as much as the mbr pointing to the address to continue to load from | 16:52 |
| TheJulia | which is why bios booting needs to have the bootloader in the first... 100GB or something like that? | 16:52 |
| dtantsur | mmm, maybe with UEFI we actually need the partition of the right type | 16:53 |
| arne_wiebalck | I thought it would look for efi partitions, no? | 16:53 |
| dtantsur | now I think I see why we need to relocate the EFI partition | 16:53 |
| arne_wiebalck | the bios tries to find partitions of type efi | 16:53 |
| dtantsur | I'm not sure why we need to call grub2.. | 16:53 |
| arne_wiebalck | I think we use grub2 as a convenience wrapper | 16:54 |
| TheJulia | but as long as we do, we won't get to having secure boot :( | 16:54 |
| TheJulia | which is why half of us are looking at beverages going "why?!?" | 16:54 |
| arne_wiebalck | right | 16:54 |
| dtantsur | right, but dd from one partition to the other shouldn't have such an impact | 16:54 |
| TheJulia | I'm not even super sure we need to do that | 16:55 |
| dtantsur | I'm trying to understand why it's more complex than just dd'ing | 16:55 |
| TheJulia | guess it depends on the uefi firmware | 16:55 |
| dtantsur | TheJulia: I think arne_wiebalck is right, and GPT will need to list the EFI partition | 16:55 |
| dtantsur | not nested inside another partition | 16:55 |
| TheJulia | yeah, likely | 16:55 |
| *** ijw has joined #openstack-ironic | 16:56 | |
| *** lucasagomes has quit IRC | 16:56 | |
| arne_wiebalck | secure boot does not work with partition images either FWIU | 16:56 |
| iurygregory | yeah | 16:56 |
| dtantsur | correct | 16:56 |
| TheJulia | only when network booted with particular configuration | 16:56 |
| dtantsur | but this is about whole disk partitions as well? | 16:56 |
| TheJulia | mainly whole disk, partition images we're stuck with grub2 | 16:57 |
| dtantsur | what is wrong with the following: 1) relocate EFI partition with dd, 2) call efibootmgr to point at it? | 16:57 |
| dtantsur | no grub2 needed, no? | 16:57 |
| * dtantsur must be missing some critical bit | 16:57 | |
| arne_wiebalck | well, the code does not call grub2 for partition images | 16:57 |
| TheJulia | dtantsur: that sshoul d work I think | 16:58 |
| dtantsur | arne_wiebalck: if it doesn't, it's probably a bug | 16:58 |
| TheJulia | conductor does it | 16:58 |
| iurygregory | at least it seems to not call (from my point of view .-.) | 16:58 |
| *** ijw has quit IRC | 16:58 | |
| arne_wiebalck | dtantsur: there's an if to bypass grub in uefi mode | 16:58 |
| * TheJulia opens the uefi spec | 16:59 | |
| *** ijw has joined #openstack-ironic | 16:59 | |
| iurygregory | partition image would be call when root_uuid and not whole_disk_image | 16:59 |
| dtantsur | arne_wiebalck: it also depends on having the EFI partition properly populated | 16:59 |
| dtantsur | I'm not sure it happens for partition images | 16:59 |
| dtantsur | so it would fall back to grub2 again | 17:00 |
| arne_wiebalck | dtantsur: properly populated? | 17:00 |
| iurygregory | https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/extensions/image.py#L519-L530 if it's using uefi for partition images would go on the if no? .-. | 17:00 |
| dtantsur | check various conditions inside _manage_uefi | 17:00 |
| iurygregory | so probably it will fail during checking the partition for uefi | 17:00 |
| dtantsur | it may return False, in which case we still call brub | 17:00 |
| dtantsur | * grub | 17:01 |
| TheJulia | confirmed, has to be in the main GPT table | 17:01 |
| TheJulia | uefi has no nested concept | 17:01 |
| dtantsur | particularly, https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/extensions/image.py#L186-L208 | 17:01 |
| *** priteau has quit IRC | 17:01 | |
| dtantsur | TheJulia: so many features - and still some are missing! | 17:02 |
| TheJulia | gah | 17:02 |
| iurygregory | now my mind is back to the normal state | 17:02 |
| arne_wiebalck | dtantsur: maybe manage_uefi fails for s/w RAID as well ... problem solved :-P | 17:02 |
| dtantsur | iurygregory: good for you, I don't even remember how it feels :D | 17:02 |
| iurygregory | arne_wiebalck, we can make it fail if you wish =P | 17:03 |
| dtantsur | we shouldn't | 17:03 |
| TheJulia | iurygregory: what is this normal you speak of?!? | 17:03 |
| TheJulia | tell us the story of "normal" | 17:03 |
| iurygregory | TheJulia, that I was not going crazy yet | 17:03 |
| dtantsur | that's what I'm trying to say: falling back to grub2 at least rules out secure boot, but probably also valid UEFI cases | 17:03 |
| iurygregory | I need to save space for friday work | 17:03 |
| dtantsur | (at least according to incomplete field reports) | 17:03 |
| * iurygregory because we have state of emergency in CZ XD | 17:04 | |
| arne_wiebalck | dtantsur: yes | 17:04 |
| dtantsur | but I still don't see a reason why we need grub2 for proper whole disk images *after we relocate the partition* | 17:04 |
| dtantsur | maybe we're trying to mix up two processes here? | 17:05 |
| dtantsur | relocating the partition seems unconditionally needed | 17:05 |
| TheJulia | I think we are and I think we agree | 17:05 |
| dtantsur | THEN we can check if we need grub2 or can do with efibootmgr only | 17:05 |
| TheJulia | I think | 17:05 |
| TheJulia | yes | 17:05 |
| arne_wiebalck | isn't grub doing the relocation atm, no? | 17:05 |
| TheJulia | no | 17:05 |
| arne_wiebalck | no? | 17:06 |
| TheJulia | grub installs fresh copies from the OS | 17:06 |
| TheJulia | those copies, are also re-assembled | 17:06 |
| TheJulia | because grub is some fancy dynamic bynary | 17:06 |
| TheJulia | binary | 17:06 |
| TheJulia | with built in modules... n'stuff | 17:06 |
| TheJulia | it has "stuff" | 17:06 |
| dtantsur | a small OS to run an OS | 17:06 |
| arne_wiebalck | ok | 17:06 |
| * TheJulia has lost it | 17:06 | |
| arne_wiebalck | lol | 17:06 |
| * TheJulia goes and makes something for lunch | 17:07 | |
| dtantsur | arne_wiebalck: in your patch, _prepare_boot_partitions_for_softraid doesn't touch grub2. I think it should be called first in install_bootloader. | 17:08 |
| dtantsur | then figure out if grub2 is actually needed or not. | 17:08 |
| dtantsur | wdyt? am I still missing something? | 17:08 |
| * TheJulia goes and tries to figure out what to make for lunch during a pandemic | 17:09 | |
| dtantsur | heh | 17:09 |
| dtantsur | it seems that the locals prefer pasta | 17:09 |
| rpittau | pasta is always good for lunch | 17:09 |
| dtantsur | at least according to empty shelves with it | 17:09 |
| iurygregory | dtantsur, _prepare_boot_partitions_for_softraid it's call on line 475 in _install_grub2 O.o | 17:09 |
| dtantsur | iurygregory: yeah, I think it should be moved | 17:09 |
| iurygregory | dtantsur, oh ok gotcha | 17:10 |
| TheJulia | I need to make sauce though, and I'm not in that mood... yet | 17:10 |
| TheJulia | but 3 hours until my next meeting... so Hmmm | 17:10 |
| dtantsur | TheJulia: we're lazy, so we usually buy ready-to-use sauces | 17:10 |
| iurygregory | when you said "doesn't touch" I thought it's not being called in the _install_grub2 | 17:10 |
| rpittau | dtantsur: heresy! | 17:10 |
| dtantsur | right? | 17:10 |
| TheJulia | rpittau: ++ | 17:11 |
| dtantsur | in my defence, they're pretty good and only contain vegetables | 17:11 |
| dtantsur | no fancy output of chemistry industry | 17:11 |
| iurygregory | I don't think you will win an argument about sauces with someone from Italy =) | 17:11 |
| rpittau | :D | 17:11 |
| iurygregory | we went from drinks to images and now food | 17:12 |
| dtantsur | iurygregory: don't underestimate me, I can start telling what people call pizza here! | 17:12 |
| iurygregory | it's time to prepare dinne ... | 17:12 |
| dtantsur | going to have a devastating psychological effect | 17:12 |
| rpittau | TheJulia: you can make a very good and fast sauce with fresh tomatoes, garlic, olive oil, chili pepper, onions | 17:13 |
| iurygregory | dtantsur, hehehehe I know this feeling in Brazil pizza is a bit different | 17:13 |
| TheJulia | dtantsur: better or worse than brno? | 17:13 |
| dtantsur | TheJulia: I've seen... truly creative things here | 17:13 |
| TheJulia | "truly creative" | 17:13 |
| dtantsur | exactly | 17:13 |
| iurygregory | pizza with pineapple? | 17:14 |
| dtantsur | fancier | 17:14 |
| TheJulia | We should devote a release of ironic to "truly creative" pizza | 17:14 |
| iurygregory | oh cheese and guava... | 17:14 |
| dtantsur | iurygregory: how would you like a pizza with potato chips? | 17:14 |
| rpittau | approved! | 17:14 |
| TheJulia | ham and pineapple == yum | 17:14 |
| iurygregory | dtantsur, there is one in Brazil | 17:14 |
| iurygregory | cheese and potato chips on top and some corn | 17:14 |
| rpittau | one of my favourites is salmon and orange slices | 17:14 |
| dtantsur | wow | 17:15 |
| rpittau | no tomato sauce | 17:15 |
| iurygregory | =O | 17:15 |
| dtantsur | sounds evil, would totally try | 17:15 |
| iurygregory | now we can say heresy to rpittau | 17:15 |
| rpittau | I'm a progressive Italian :D | 17:15 |
| iurygregory | =D | 17:15 |
| openstackgerrit | Julia Kreger proposed openstack/ironic master: Make image_checksum optional if other checksum is present https://review.opendev.org/711816 | 17:16 |
| clarkb | a local pizza place does a white cheddar, potato, and bacon pizza | 17:16 |
| dtantsur | yummy | 17:16 |
| clarkb | I would never think to add potato but it works | 17:16 |
| rpittau | clarkb: that's not bad | 17:16 |
| iurygregory | *-* | 17:16 |
| iurygregory | wow | 17:16 |
| rpittau | potato on pizza is good | 17:16 |
| dtantsur | and then there is a local variant of pizza called Flammkuchen | 17:16 |
| rpittau | you need good potatoes | 17:16 |
| * iurygregory hopes they can do delivery in Brno =P | 17:16 | |
| dtantsur | sour cream instead of tomato | 17:17 |
| dtantsur | https://en.wikipedia.org/wiki/Flammekueche | 17:17 |
| rpittau | that looks interesting | 17:18 |
| arne_wiebalck | I am not sure I will have the bandwidth to test/experiment with a grub2-less s/w RAID in the coming weeks | 17:18 |
| dtantsur | arne_wiebalck: we can probably merge your patch since it doesn't make anything worse | 17:19 |
| dtantsur | (with a new condition on software RAID) | 17:19 |
| dtantsur | but I'd prefer us not dropping it on the floor afterwards | 17:19 |
| arne_wiebalck | there is a risk | 17:19 |
| iurygregory | we can only merge after removing import six =) | 17:19 |
| arne_wiebalck | the current code with grub2 adds UEFI on RAID | 17:20 |
| dtantsur | well, some updates are needed anyway | 17:20 |
| rpittau | I think dd will work | 17:20 |
| rpittau | I was reading some stuff while following the discussion | 17:21 |
| arne_wiebalck | I think we agree that grub2 is probably not required, but I think we would need to test this | 17:21 |
| arne_wiebalck | before we add such a patch | 17:21 |
| arne_wiebalck | the current patch has been tested on real hardware | 17:22 |
| rpittau | grub is not required for uefi, so if the partition is migrated for the softraid, we shouldn't need that either, but I agree testing is required | 17:22 |
| iurygregory | yeah ++ for testing | 17:23 |
| dtantsur | arne_wiebalck: okay, let's finish and merge it with understanding that more work is required | 17:23 |
| dtantsur | unless anybody disagrees.. | 17:23 |
| arne_wiebalck | I agree that this all not ideal. | 17:24 |
| rpittau | let's clean it up and merge, maybe add a note to the etherpad for future work on that ? | 17:24 |
| iurygregory | if it wants to use grub2 (and we merge) it will only need to update the conditions etc... | 17:24 |
| arne_wiebalck | and six :-D | 17:24 |
| iurygregory | please <3 | 17:25 |
| iurygregory | remove six! | 17:25 |
| rpittau | clean it up = remove import six | 17:25 |
| arne_wiebalck | I didn't add it ... or did I? | 17:25 |
| arne_wiebalck | anyway | 17:25 |
| rpittau | arne_wiebalck: it was there since the beginning, ti was in november | 17:25 |
| rpittau | if you look at my comment at end of january it mentions it | 17:25 |
| arne_wiebalck | I am happy to have the additional work for the grub2 less UEFI RAID on me | 17:26 |
| arne_wiebalck | rpittau: yes, sorry I missed that | 17:26 |
| arne_wiebalck | so, you can ping me on this all the time :-D | 17:26 |
| * rpittau leaves until there's still some light outside | 17:27 | |
| rpittau | bye! good night! o/ | 17:27 |
| *** rpittau is now known as rpittau|afk | 17:27 | |
| arne_wiebalck | I'll update the patch tomorrow, thanks iurygregory rpittau dtantsur TheJulia for your time and thoughts! | 17:27 |
| dtantsur | thank you arne_wiebalck! | 17:28 |
| iurygregory | arne_wiebalck, thank you o/ | 17:28 |
| TheJulia | thanks! | 17:30 |
| dtantsur | yay, no travel to CZ as well.. | 17:30 |
| TheJulia | dtantsur: huh? | 17:31 |
| dtantsur | TheJulia: looking at travel bans expanding and border closing.. | 17:31 |
| TheJulia | eek | 17:31 |
| dtantsur | I guess it's justified. Sad to see these times though. | 17:31 |
| TheJulia | yeah | 17:32 |
| * dtantsur hopes iurygregory and etingof didn't plan out-of-country travels in the near future | 17:32 | |
| iurygregory | yeah | 17:32 |
| TheJulia | to flatten the curve it will be at least the next two to three weeks | 17:32 |
| dtantsur | yep. a month in practice | 17:33 |
| iurygregory | well I was planning to go to Japan this year, but I think I wont and that in the end of the year things will be more calm and I will spend xmas and new years in Brazil XD | 17:33 |
| dtantsur | we still hope to get some hiking in Switzerland in May. The chances don't look awesome though.. | 17:33 |
| dtantsur | Ending up with no vacation is *exactly* what my psychological health needs here and now :-/ </rant> | 17:34 |
| dtantsur | on the "positive" note I'll wish you good night | 17:34 |
| *** dtantsur is now known as dtantsur|afk | 17:34 | |
| iurygregory | =/ | 17:35 |
| openstackgerrit | Julia Kreger proposed openstack/ironic-python-agent master: Agent token support https://review.opendev.org/692614 | 17:35 |
| * etingof is stuck in his village surrounded mostly by cows, hens and redfish | 17:42 | |
| *** xXraphXx has joined #openstack-ironic | 17:45 | |
| *** kozhukalov has joined #openstack-ironic | 17:46 | |
| * TheJulia imagines etingof have created redfish all around him | 17:55 | |
| *** derekh has quit IRC | 18:01 | |
| iurygregory | updates on the stable/stein branch for grande ( tempest venv is recreated by grenade which removed the already installed ironic-tempest-plugin and so does tempest did not loaded plugins things. ) | 18:01 |
| TheJulia | wheeee | 18:04 |
| *** Lucas_Gray has quit IRC | 18:05 | |
| arne_wiebalck | bye everyone o/ | 18:05 |
| etingof | breeding redfish is difficult in a landlocked country | 18:06 |
| etingof | it's easier with ironic bears though | 18:07 |
| TheJulia | lol | 18:14 |
| TheJulia | what about red submarine fish? | 18:14 |
| etingof | that's musical! | 18:17 |
| etingof | so far idrac is totally non-cooperative in the bios tackling business | 18:18 |
| etingof | let's try to bump up the firmware | 18:18 |
| rpioso | etingof: Be careful about bumping the firmware. | 18:26 |
| rpioso | etingof: Which h/w are you using? And what's its current iDRAC firmware version? | 18:26 |
| TheJulia | etingof: set bios and boot mode at once or?? | 18:31 |
| ktibi | Re, so I switch to ipmi instead of ILO and works. But know, IPA return a empty list in disks value. | 18:47 |
| ktibi | now* | 18:48 |
| ktibi | any tips for debug why my disks list is empty ? | 18:48 |
| openstackgerrit | Harald Jensås proposed openstack/ironic master: Additional IP addresses to IPv6 stateful ports https://review.opendev.org/700002 | 18:50 |
| *** erbarr has joined #openstack-ironic | 18:59 | |
| *** gmann is now known as gmann_lunch | 19:00 | |
| *** dsneddon has joined #openstack-ironic | 19:02 | |
| *** kozhukalov has quit IRC | 19:13 | |
| *** diurnalist has quit IRC | 19:14 | |
| *** amoralej is now known as amoralej|off | 19:24 | |
| etingof | rpioso, it's 3.21.26.22 | 19:29 |
| rpioso | etingof: Hardware model? | 19:29 |
| etingof | R640 | 19:29 |
| rpioso | etingof: The most recent version we've tested is 3.30.30.30. | 19:34 |
| etingof | right, this is historical! from early 2019 | 19:35 |
| etingof | what should I be careful about when kicking idrac firmware? | 19:35 |
| rpioso | etingof: Right, historical and solid. | 19:36 |
| rpioso | etingof: 4.00.00.00 and 4.10.10.10 have the Continuous issue. We may not have tried to downgrade from that version. We have encountered downgrade issues with a recent release of 13G firmware. A workaround to facilitate downgrading is available, and we believe it should work with 14G, too, although we haven't tried it. | 19:39 |
| * TheJulia needs to stop writing books of text | 19:45 | |
| TheJulia | ktibi: your IPA ramdisk likely lacks drivers to see the disks. Sounds like your ramdisk needs a driver for the raid controller? | 19:46 |
| TheJulia | ktibi: alternatively, a raid set may not already be created. | 19:46 |
| *** trident has quit IRC | 19:51 | |
| *** trident has joined #openstack-ironic | 19:59 | |
| etingof | rpioso, should I go for 3.30.30.30? is it possible to choose the desired version on fw upgrade? | 19:59 |
| etingof | TheJulia, let's write code that will write books of text? | 20:00 |
| rpioso | etingof: The version depends on what you're trying to do. | 20:00 |
| rpioso | etingof: 3.30.30.30 does not have the Continuous issue. | 20:00 |
| etingof | my goal is to see if bios management works | 20:01 |
| etingof | e.g. read/change/reset | 20:01 |
| *** gmann_lunch is now known as gmann | 20:01 | |
| rpioso | etingof: 3.30.30.30 should be sufficient. | 20:01 |
| etingof | awesome! | 20:01 |
| rpioso | etingof: And, yes, you can choose the version from support.dell.com. | 20:01 |
| etingof | good! | 20:02 |
| rpioso | Gotta click on the right links. It's a UI :-) | 20:02 |
| rpioso | etingof: Enter the service tag number and rock on. | 20:02 |
| * etingof has fought his Mac to downgrade to OS version that still works | 20:02 | |
| * rpioso advises etingof to purchase a Dell Precision :-) | 20:03 | |
| * etingof has owned some Dell Latitude in late nineties | 20:09 | |
| *** Goneri has quit IRC | 20:18 | |
| *** rh-jelabarre has quit IRC | 20:22 | |
| *** rh-jelabarre has joined #openstack-ironic | 20:29 | |
| *** TxGirlGeek has quit IRC | 20:32 | |
| *** diurnalist has joined #openstack-ironic | 20:33 | |
| *** alexmcleod has quit IRC | 20:33 | |
| *** ociuhandu has joined #openstack-ironic | 20:42 | |
| *** ktibi has quit IRC | 20:45 | |
| *** ociuhandu has quit IRC | 21:02 | |
| *** ociuhandu has joined #openstack-ironic | 21:03 | |
| *** ociuhandu has quit IRC | 21:08 | |
| openstackgerrit | Julia Kreger proposed openstack/ironic master: Hash the rescue_password https://review.opendev.org/691413 | 21:25 |
| *** rh-jelabarre has quit IRC | 21:26 | |
| *** rh-jelabarre has joined #openstack-ironic | 21:27 | |
| openstackgerrit | Julia Kreger proposed openstack/ironic-python-agent master: Rescan after restarting the md device https://review.opendev.org/706632 | 21:28 |
| *** alexmcleod has joined #openstack-ironic | 21:39 | |
| *** ociuhandu has joined #openstack-ironic | 21:41 | |
| *** amoralej|off is now known as amoralej | 22:00 | |
| *** ociuhandu has quit IRC | 22:03 | |
| *** ociuhandu has joined #openstack-ironic | 22:04 | |
| *** rcernin has joined #openstack-ironic | 22:04 | |
| *** rcernin has quit IRC | 22:04 | |
| *** rcernin has joined #openstack-ironic | 22:05 | |
| *** ociuhandu has quit IRC | 22:09 | |
| *** diurnalist has quit IRC | 22:28 | |
| *** sziviani_ has quit IRC | 22:28 | |
| *** diurnalist has joined #openstack-ironic | 22:36 | |
| *** diurnalist has quit IRC | 22:43 | |
| *** diurnalist has joined #openstack-ironic | 22:44 | |
| *** tkajinam has joined #openstack-ironic | 22:56 | |
| *** Goneri has joined #openstack-ironic | 23:08 | |
| *** amoralej is now known as amoralej|off | 23:29 | |
| *** threestrands has joined #openstack-ironic | 23:50 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!