| opendevreview | Julia Kreger proposed openstack/ironic master: WIP: Add HTTP versions of network boot interfaces https://review.opendev.org/c/openstack/ironic/+/900965 | 00:17 |
|---|---|---|
| opendevreview | Julia Kreger proposed openstack/ironic master: DNM: CI test for httpboot jobs https://review.opendev.org/c/openstack/ironic/+/901182 | 00:17 |
| opendevreview | Julia Kreger proposed openstack/ironic-tempest-plugin master: Add snmp variant of ramdisk iso boot test https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/901213 | 01:06 |
| TheJulia | the first http boot in CI https://a37003535f40ac4a096b-3fa6c7b03298e65764ba8baf4a8a5f26.ssl.cf2.rackcdn.com/901182/3/check/ironic-tempest-bfv/b6bba8f/controller/logs/ironic-bm-logs/node-0_console_2023-11-17-01%3A10%3A05_log.txt | 01:24 |
| opendevreview | Julia Kreger proposed openstack/ironic-tempest-plugin master: Add snmp variant of ramdisk iso boot test https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/901213 | 01:50 |
| opendevreview | Julia Kreger proposed openstack/sushy-tools master: Simplify UEFI logic and change the UefiHttp flow https://review.opendev.org/c/openstack/sushy-tools/+/901208 | 02:00 |
| TheJulia | https://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_3a9/901182/3/check/ironic-tempest-ipa-partition-uefi-pxe-grub2/3a92312/controller/logs/ironic-bm-logs/node-0_console_2023-11-17-01%3A13%3A00_log.txt <-- second, grub. *almost* there. just needs the extra grubby things deposited | 02:01 |
| timeu | Hi all, I noticed a difference in behavior when deploying Centos7 vs RHEL8 on a baremetal node using a whole disk image (with EFI partition): In the centos7 case the root partition is not automatically grown to the size of the disk (it stays at 4GB = size of the image) while in the rhel8 case the root partition is grown to the size of the disk. Is this is a known issue ? | 08:02 |
| timeu | both deployments are done with the same IPA image (ussuri). Ironic version is the one from OSP16.2 | 08:05 |
| dtantsur | timeu: it could be difference in cloud-init versions or configuration. That's what grows your partitions. It's also possible that in C7 the root partition is not the last on the disk. | 08:39 |
| timeu | @dtantsur: thanks for the info. I was wondering what is responsible for growing the partition. As far as I can tell the partition schema is the same for both OS (EFI, BIOS boot, root, 64 MB linux fs at the end). | 08:51 |
| timeu | so ironic just writes the image to the disk and the cloud-init of the image will take care of growing the partition | 08:51 |
| dtantsur | correct | 09:03 |
| dtantsur | TheJulia: really nice! (re HTTP boot). Is it redfish or DHCP? | 09:10 |
| iurygregory | morning Ironic | 11:25 |
| TheJulia | dtantsur: both the logs are dhcp based. Redfish is being a little problematic in that I added a sanity check into the driver code to generate an exception and fail validation if the hardware doesn't support it which seemed reasonable to me. And then I had a bad conditional that was not quite working causing the template to not work. I can't run any of the unit tests locally for sushy-tools since I'm on debian and it | 13:44 |
| TheJulia | requires the libvirt module to be compiled :\ | 13:44 |
| TheJulia | and it is not like there really are many unit tests in sushy-tools | 13:52 |
| -opendevstatus- NOTICE: Gerrit will be unavailable for a short time starting at 15:30 UTC as it is upgraded to the 3.8 release. https://lists.opendev.org/archives/list/service-announce@lists.opendev.org/thread/XT26HFG2FOZL3UHZVLXCCANDZ3TJZM7Q/ | 14:06 | |
| timeu | @dtantsur: FYI: using the latest centos7 cloud image seems to have fixed the issue | 14:22 |
| dtantsur | good! | 14:36 |
| TheJulia | sigh.... | 15:33 |
| opendevreview | Julia Kreger proposed openstack/sushy-tools master: Simplify UEFI logic and change the UefiHttp flow https://review.opendev.org/c/openstack/sushy-tools/+/901208 | 15:43 |
| -opendevstatus- NOTICE: The Gerrit upgrade is complete, however we have Zuul offline in parallel for a schema migration, so any events occurring during this time will be lost (requiring a recheck or similar to trigger jobs once it returns to service); we'll update again once this is complete. | 16:34 | |
| -opendevstatus- NOTICE: Zuul is fully back in service now, but any events occurring prior to 17:05 UTC may need a recheck to trigger jobs. | 17:13 | |
| JayF | TheJulia: I am hunting thru email history for the emails about the ARM CI project proposed for a university group last cycle | 17:28 |
| JayF | TheJulia: trying to find it to re-propose it for the New Mexico State thing posted to the list this week | 17:28 |
| dtantsur | JFYI I think https://github.com/dtantsur/ironic-operator has reached the state where an interested person could start playing with it. Beware: rough edges everywhere, nearly untested. | 17:28 |
| dtantsur | Have a nice weekend o/ | 17:28 |
| JayF | dtantsur: is that for a single process/sqlite | 17:28 |
| JayF | or a full throttle ironic? | 17:28 |
| JayF | have a nice one o/ see you in a week (I'm gone all next week) | 17:28 |
| dtantsur | JayF: single process, supports sqlite and mariadb | 17:28 |
| JayF | ack | 17:29 |
| dtantsur | ah, nice, enjoy! | 17:29 |
| iurygregory | So, I was looking at the idea of caching https://opendev.org/openstack/ironic-python-agent/src/branch/master/ironic_python_agent/hardware.py#L237 to avoid spending a lot of time when the hardware has a lot of devices, after looking at our code base for a while my idea is to include the information we get in _get_multipath_parent_device when it's called in list_all_block_devices https://opendev.org/openstack/ironic-python-agent/src/branch/ | 17:29 |
| iurygregory | master/ironic_python_agent/hardware.py#L566 into the BlockDevice, does this make sense or I'm in the wrong path? if anyone has ideas please let me know =) | 17:29 |
| dtantsur | actually, it has code to support 3 ironic instances (1 per control plane node), but that probably does not work correctly, especially with sqlite | 17:29 |
| dtantsur | o/ | 17:29 |
| dtantsur | s/sqlite/dnsmasq/ in the previous message (of course, it does not work with sqlite, but that's expected) | 17:30 |
| JayF | I've mentioned this a few places, going to say it here: it's my last day online working until Nov 27 (a week from Monday). If you need something from me speak soon :) | 17:39 |
| JayF | iurygregory: I wonder if that would cause the fault tolerance to be less-tolerant of faults | 17:40 |
| JayF | iurygregory: e.g. if a failure in one path occurs between "when the cache was pulled" and "when we use it", will it be problematic? | 17:40 |
| iurygregory | JayF, from the previous discussion we had (on Oct) I don't think this would be a problem | 17:42 |
| JayF | ack | 17:42 |
| JayF | I don't know much about multipath in general | 17:42 |
| JayF | that was just the thought I had :) | 17:42 |
| iurygregory | I had the link handy if you want to check the irc conversation we had | 17:42 |
| TheJulia | JayF: ack, so fwiw, the week before christmas, I'll be in NM ~30 minutes from UNM | 17:43 |
| * JayF wonders if NM means something different than New Mexico | 17:43 | |
| JayF | oooooh | 17:43 |
| JayF | I just grasped the relevance | 17:43 |
| JayF | sorry, had to pick the context up off the ground lol | 17:43 |
| * TheJulia is always happy to go go have a cup of coffee in new mexico with people | 17:43 | |
| TheJulia | iurygregory: trying to grok/think through it | 17:45 |
| TheJulia | iurygregory: so, if you mean by that, so we return the md device *instead* of the base devices, sure. | 17:46 |
| iurygregory | maybe, ok I've found something strange when looking at the logs and our code .-. | 18:21 |
| iurygregory | "WARNING root [-] We have identified a multipath device sdei, this is being ignored in favor of (35000c500d706f623) and its related child devices" | 18:21 |
| iurygregory | (35000c500d706f623) shouldn't be dm-46? .-. https://paste.opendev.org/show/bxyDYugFagTGnWNgSYoU/ | 18:22 |
| iurygregory | my idea is that we could have some mapping when we execute the multipath -ll <device> and we could have {'sdei': 'dm-46', 'sdbc': 'dm-46', 'sdho': 'dm-46', 'sdku': 'dm-46', ....}, and we wouldn't need to check 3 devices (this would improve the time) not even sure if would make sense ... | 18:26 |
| iurygregory | if we don't have the information for the device we normally execute _get_multipath_parent_device (and when doing we save this information for and keep it) so when we go to verify sdbc we won't execute _get_multipath_parent_device | 18:27 |
| TheJulia | so I guess the thing is we should be ignoring the underlying devices completely | 19:18 |
| TheJulia | if there is a dm-device | 19:18 |
| TheJulia | I guess this might just be easier to mentally model with code | 19:23 |
| TheJulia | 15T device 8| | 19:23 |
| iurygregory | yeah, in case we find one we could ignore the others | 19:31 |
| * iurygregory trying to get a parser for the output that would work | 19:32 | |
| iurygregory | TheJulia, we probably need to update our current code, to provide the right device (not the numbers right?) | 19:33 |
| iurygregory | wondering if output differs based on the fiber channel etc | 19:33 |
| iurygregory | since I know the current code we have worked for a Customer when we had an escalation | 19:34 |
| opendevreview | Merged openstack/bifrost master: ironic: Perform online data migrations with localhost DB https://review.opendev.org/c/openstack/bifrost/+/895948 | 20:57 |
| opendevreview | Julia Kreger proposed openstack/sushy-tools master: Simplify UEFI logic and change the UefiHttp flow https://review.opendev.org/c/openstack/sushy-tools/+/901208 | 21:51 |
| TheJulia | iurygregory: what do you mean device numbers? | 21:51 |
| TheJulia | iurygregory: sorry, had my 1-on-1 and got distracted | 21:52 |
| iurygregory | TheJulia, currently our code thinks the output is always in this format https://opendev.org/openstack/ironic-python-agent/src/branch/master/ironic_python_agent/hardware.py#L271-L280 | 21:53 |
| iurygregory | in the bug I'm working we are returning "(35000c500d706f623)" instead of "dm-46" | 21:53 |
| iurygregory | https://paste.opendev.org/show/bxyDYugFagTGnWNgSYoU/ | 21:54 |
| TheJulia | oh! | 21:54 |
| TheJulia | yeah | 21:54 |
| TheJulia | it would be | 21:54 |
| iurygregory | so we can't always think the second element of the split is the right thing | 21:54 |
| TheJulia | ewwwwww | 21:54 |
| iurygregory | =( | 21:54 |
| TheJulia | ugh | 21:54 |
| iurygregory | my idea so far is to just check the split and grab the one that startswith "dm" | 21:55 |
| iurygregory | not sure if is the right approach... | 21:55 |
| iurygregory | but it works on my test here using the output from the bug XD | 21:56 |
| iurygregory | I will finish things and push the patches on monday | 22:00 |
| iurygregory | going to turn off the laptop o/ have a great weekend | 22:01 |
| TheJulia | iurygregory: that is likely a good approach | 22:01 |
| iurygregory | ack o/ | 22:02 |
| opendevreview | Julia Kreger proposed openstack/ironic master: WIP: Add HTTP versions of network boot interfaces https://review.opendev.org/c/openstack/ironic/+/900965 | 22:03 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!