| vanou | good morning ironic | 02:20 |
|---|---|---|
| vanou | TheJulia JayF: Thanks review for https://review.opendev.org/c/openstack/ironic/+/867808 https://review.opendev.org/c/openstack/ironic/+/865075 . The reason I created 867808 is, if I adopt try-fallback approach on 865075, IPMI try takes 60seconds then fallback to Redfish. So this makes iRMC driver very slow because IPMI has been used many places in code. TheJulia's advice on | 02:31 |
| vanou | 865075, which patch 867808 should introduce "don't retry automatically", is helpful. | 02:31 |
| kubajj | Good morning vanou and Ironic! | 05:43 |
| vanou | good morning kubajj | 07:34 |
| opendevreview | Vincent proposed openstack/bifrost master: Feature: Enable user_data's file generation https://review.opendev.org/c/openstack/bifrost/+/870017 | 08:09 |
| iurygregory | good morning Ironic | 11:37 |
| opendevreview | Baptiste Jonglez proposed openstack/networking-generic-switch master: Document ngs_trunk_ports and ngs_physical_networks https://review.opendev.org/c/openstack/networking-generic-switch/+/868313 | 12:24 |
| opendevreview | Baptiste Jonglez proposed openstack/networking-generic-switch master: Add support for Cisco Nexus devices (NX-OS) https://review.opendev.org/c/openstack/networking-generic-switch/+/868481 | 12:26 |
| kubajj | dtantsur: I am clueless in how to integrate the NodeInventoryController into the _subcontroller_map. Any suggestions? | 12:33 |
| dtantsur | kubajj: are there any issues with doing it the way other controllers are integrated? | 12:38 |
| kubajj | dtantsur: still the node_ident error | 12:39 |
| dtantsur | kubajj: did you copy the part TheJulia suggested? could you show me the current diff? | 12:40 |
| kubajj | dtantsur: I did. Can't right now. I needed to go to the embassy to vote in elections, but will be back in a bit. | 12:43 |
| dtantsur | sure, important things first | 12:44 |
| opendevreview | Baptiste Jonglez proposed openstack/networking-generic-switch master: config: Ignore unknown options starting with ngs_ https://review.opendev.org/c/openstack/networking-generic-switch/+/868300 | 12:50 |
| kubajj | dtantsur: https://paste.opendev.org/show/bJZRImUiBJsxJed4iCZ4/ | 13:18 |
| * dtantsur applies locally | 13:18 | |
| opendevreview | Baptiste Jonglez proposed openstack/networking-generic-switch master: Add ngs_ssh_disabled_algorithms setting https://review.opendev.org/c/openstack/networking-generic-switch/+/868316 | 13:22 |
| dtantsur | kubajj: ah! you need to remove the node_ident argument from get() and use self.node_ident now | 13:25 |
| kubajj | dtantsur: Ok, that worked :D Thanks | 13:29 |
| opendevreview | Jakub Jelinek proposed openstack/ironic master: API for node inventory https://review.opendev.org/c/openstack/ironic/+/866876 | 13:30 |
| *** tosky_ is now known as tosky | 14:47 | |
| TheJulia | vanou: one thing I was thinking of last night was, you also don't need to always try ipmi. you could use power status checking as a "does ipmi work" tracker that could set/record, and then the other methods wouldn't need to always attempt ipmi. The reason why I was thinking about this that way, was because each time the conductor does something user-interactive, it is still going to be a long delay because ipmitool is | 14:50 |
| TheJulia | going to take some time unless it gets an immediate reply from the bmc saying "not permitted", eventually I could see operators firewalling off ipmi entirely which would then mean the same base problem would exist | 14:50 |
| vanou | TheJulia: Thanks for comment. You mean let 'power status checking' in verify step (get_power_state method of power interface?) checks whether IPMI operation works. Then that check stores if IPMI works in e.g. driver_internal_info. Then another iRMC driver methods determine which protocol(IPMI/Redfish) to use based on that stored information. Is my understandig correct? | 15:34 |
| TheJulia | vanou: no, I mean https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/irmc/power.py#L242 instead of verify steps | 16:06 |
| TheJulia | that is not a verify step, that method gets called periodically to record/track the power status of the BMC | 16:07 |
| TheJulia | but otherwise I think we're on the same page | 16:07 |
| vanou | Yeah get_power_state method is good. It's called periodically and also called in verifying. If we put logic which set flag of driver_internal_info into get_power_state method, is that flag can be firmware version as earlier patch? Or that flag should be more simple one (boolean)? | 16:21 |
| TheJulia | vanou: I would just do it as a boolean | 16:49 |
| TheJulia | vanou: that way we don't really even need to be precisely aware of the version, we just try to gracefully handle it | 16:50 |
| vanou | TheJulia: Thanks. I'll add boolean flag on that method and change another related method to check that flag to determine IPMI or Redfish. | 16:58 |
| kubajj | TheJulia: could you please re-review https://review.opendev.org/c/openstack/ironic/+/866876 when you have a minute so I can work on follow-ups or fixing what is missing tomorrow? | 18:45 |
| TheJulia | kubajj: I’m out at the moment but I’ll try to when I get back | 18:46 |
| kubajj | TheJulia: thanks | 18:47 |
| kubajj | TheJulia: the inspect utils import change is one of the follow-ups I'll work on next. Thanks for the review. | 20:51 |
| TheJulia | The race for api microversion 1.81 intensivies | 20:52 |
| TheJulia | intensifies | 20:52 |
| TheJulia | JayF: also, ./ironic/tests/unit/api/controllers/v1/test_node.py:7963:80: E501 line too long (80 > 79 characters) | 20:52 |
| JayF | ugh | 20:52 |
| JayF | is that the only feedback, or is there more? | 20:52 |
| JayF | if that's it, I'll do a quick fix | 20:53 |
| JayF | TheJulia: I'll express an explicit desire for kubajj to win the race if it gets close; I think I probably have more time to rebase it out than they do :) | 20:53 |
| kubajj | JayF: I'm happy to tease as well. No worries. | 20:54 |
| kubajj | *rebase. Not tease 😃 | 20:54 |
| TheJulia | JayF: I think there is something going on with the rbac tests | 21:01 |
| TheJulia | I've pulled it down locally and launched the tests | 21:01 |
| JayF | TheJulia: heh, I was thinking for project_scoped.yaml it returned a 500 for success b/c it tried to do something but didn't | 21:05 |
| JayF | TheJulia: and saw other examples of tests returning 500s | 21:06 |
| TheJulia | i still think you should get a 200 back on that | 21:06 |
| JayF | TheJulia: honestly, if it's supposed to be 200, I bet I have a bug where if you have no shards you err | 21:06 |
| TheJulia | which is odd | 21:06 |
| TheJulia | oh, I see it | 21:06 |
| TheJulia | will leave a comment on the change | 21:06 |
| JayF | good stuff, ty | 21:07 |
| TheJulia | comment sent | 21:07 |
| TheJulia | that should result in a 200 getting returned | 21:08 |
| JayF | aha, so real bug | 21:08 |
| JayF | thank you | 21:08 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!