| opendevreview | Jacob Anders proposed openstack/sushy stable/zed: Improve resiliency of eTag handling https://review.opendev.org/c/openstack/sushy/+/862846 | 01:17 |
|---|---|---|
| opendevreview | Jacob Anders proposed openstack/sushy stable/yoga: Improve resiliency of eTag handling https://review.opendev.org/c/openstack/sushy/+/862847 | 01:18 |
| opendevreview | Jacob Anders proposed openstack/sushy stable/xena: Improve resiliency of eTag handling https://review.opendev.org/c/openstack/sushy/+/862848 | 01:23 |
| opendevreview | Jacob Anders proposed openstack/sushy stable/wallaby: Improve resiliency of eTag handling https://review.opendev.org/c/openstack/sushy/+/862849 | 01:24 |
| opendevreview | Harald Jensås proposed openstack/ironic master: [WIP] Use join to get node_uuid for ports https://review.opendev.org/c/openstack/ironic/+/862852 | 02:09 |
| opendevreview | Harald Jensås proposed openstack/ironic master: [WIP] Use join to get node_uuid for ports https://review.opendev.org/c/openstack/ironic/+/862852 | 02:18 |
| opendevreview | Harald Jensås proposed openstack/ironic master: [WIP] Use join to get node_uuid for ports https://review.opendev.org/c/openstack/ironic/+/862852 | 03:17 |
| *** v1k0d3n_ is now known as v1k0d3n | 08:07 | |
| *** cstone5 is now known as cstone | 08:07 | |
| *** buhman_ is now known as buhman | 08:07 | |
| *** kubajj_ is now known as kubajj | 08:07 | |
| *** NobodyCam_ is now known as NobodyCam | 08:07 | |
| *** erbarr_ is now known as erbarr | 08:07 | |
| *** TheJulia_ is now known as TheJulia | 08:07 | |
| mnasiadka_ | Good morning | 09:47 |
| mnasiadka_ | Is there a way to change deploy interface for nodes in active status somehow? (upgraded OpenStack to Xena and iscsi deploy interface is gone) | 09:47 |
| mnasiadka_ | ok, solved with setting maintenance, changing, and going out of maintenance | 10:18 |
| iurygregory | good morning | 12:06 |
| opendevreview | Aija Jauntēva proposed openstack/ironic master: Create 'redfish' driver Redfish Interop Profile https://review.opendev.org/c/openstack/ironic/+/754061 | 12:38 |
| opendevreview | Aija Jauntēva proposed openstack/ironic master: Create 'redfish' driver Redfish Interop Profile https://review.opendev.org/c/openstack/ironic/+/754061 | 12:38 |
| opendevreview | Harald Jensås proposed openstack/ironic master: [WIP] Use join to get node_uuid for ports https://review.opendev.org/c/openstack/ironic/+/862852 | 12:45 |
| hjensas | good morning | 12:57 |
| hjensas | TheJulia: ^ very, WIP but wdyt? | 12:57 |
| TheJulia | good morning! | 12:59 |
| TheJulia | mnasiadka_: That is the trick for *most* fields if your an admin and they are state restricted :) | 13:00 |
| TheJulia | hjensas: what does the db response object look like, out of curiosity? | 13:04 |
| hjensas | It query.one() returns a tuple([models.Port, models.Node]), query.all returns [tuple([models.Port, models.Node]), tuple([models.Port, models.Node]), ...] | 13:06 |
| TheJulia | Does any of this hits sa.execute? | 13:19 |
| hjensas | ironic/db/sqlalchemy/api.py:342 does session.execute | 13:25 |
| TheJulia | okay, yeah | 13:26 |
| TheJulia | I meant more about an orm query logic path being followed | 13:26 |
| TheJulia | You know... we could actually change the object definition... | 13:27 |
| TheJulia | for an inherent node_uuid join | 13:28 |
| TheJulia | that way we're not doing odd response manipulation | 13:28 |
| hjensas | you mean add a second foreign key in the models? | 13:28 |
| TheJulia | no | 13:28 |
| TheJulia | uhhh | 13:28 |
| TheJulia | one moment let me get an example | 13:29 |
| * TheJulia just stumbled upon mapped expressions in sqlalchemy and has crazy bad ideas | 13:29 | |
| TheJulia | err, not object | 13:30 |
| TheJulia | the model | 13:30 |
| TheJulia | https://github.com/openstack/ironic/blob/master/ironic/db/sqlalchemy/models.py#L232 | 13:31 |
| TheJulia | https://github.com/openstack/ironic/blob/master/ironic/db/sqlalchemy/models.py#L298 | 13:32 |
| TheJulia | That way we can have sqlalchemy do the population itself, and we don't have to have any logic on it | 13:33 |
| TheJulia | and if we do selectinload it won't have the chance of being an exploded thing requiring orm to do additional things (besides, auto-dedupe is going away on orm) | 13:34 |
| TheJulia | so selectinload is likely the safest bet overall | 13:34 |
| TheJulia | I guess it boils down to 2 is better than one and possibly having to do additional dedupe/object manipulation in my mind | 13:36 |
| TheJulia | Overall, not bad! | 13:36 |
| TheJulia | I'm super impressed you got a wip that quickly :) | 13:37 |
| TheJulia | This will mean though, if we go down the path of editing the model with a join, the object model will get a new field | 13:38 |
| TheJulia | err | 13:38 |
| TheJulia | that could compound | 13:38 |
| TheJulia | hmmmm | 13:38 |
| hjensas | k, I'll give that mapped relationship a go. It would be very good if we don't need to do the object manipulation. | 13:38 |
| TheJulia | we may need more lower level logic though, since a normal joined response would be best for a single port | 13:39 |
| TheJulia | 1 node, 1 port | 13:39 |
| TheJulia | maybe normal joined response would be okay | 13:39 |
| * TheJulia suspects we shouldn't need to so anything like NodeBase and Node that I did | 13:40 | |
| TheJulia | (so you can get nodes, without the joins) | 13:40 |
| TheJulia | JayF: we might need to think about benchmarking ports too..... | 13:42 |
| TheJulia | which I believe would make folks like rloo super happy | 13:42 |
| rloo | :D | 13:43 |
| TheJulia | and... speed up networking-baremetal | 13:44 |
| JayF | gotta go fast | 13:57 |
| JayF | If we do that; we should also try to run some py39 vs py311 benchmarks | 13:58 |
| JayF | apparently just going to py311 has stunning perf uplifts | 13:58 |
| TheJulia | ohhh ahhhh | 14:05 |
| TheJulia | did they change the recompile pattern or something? | 14:05 |
| JayF | I do not know the source of the improvements; only that I've seen it confirmed multiple places :) | 14:07 |
| TheJulia | Interesting! | 14:09 |
| JayF | https://docs.python.org/3.11/whatsnew/3.11.html#whatsnew311-faster-cpython | 14:09 |
| JayF | honestly, given some of the things I've seen benchmarked | 14:10 |
| JayF | I'm really interesting to see if the type of data transformatinos and processing we do will be one of the more improved cases | 14:10 |
| TheJulia | yeah, they basically are freezing the imported code | 14:17 |
| *** larsks_ii is now known as larsks | 16:02 | |
| opendevreview | Harald Jensås proposed openstack/ironic master: [WIP] Use ORM relationship for node_uuid for ports https://review.opendev.org/c/openstack/ironic/+/862933 | 16:13 |
| hjensas | TheJulia: ^ using association_proxy seems to work. But lot's of "parent instance ... is not bound to a Session" errors in unit tests. | 16:16 |
| TheJulia | ewww | 16:18 |
| TheJulia | I'd check the relationship mapping docs | 16:21 |
| opendevreview | Merged openstack/bifrost master: Move bifrost tests to Ubuntu Jammy https://review.opendev.org/c/openstack/bifrost/+/848828 | 16:34 |
| opendevreview | Merged openstack/bifrost master: Upgrade from zed https://review.opendev.org/c/openstack/bifrost/+/859075 | 16:39 |
| TheJulia | woot, I actually have a single inspector functional test working \o/ | 16:48 |
| TheJulia | and I'm finally to the pain that is the state machine in inspector | 17:01 |
| * TheJulia thinks not today | 17:01 | |
| hjensas | ... association_proxy does weird stuff on create, back to the docs ;-E | 17:14 |
| TheJulia | :( | 17:14 |
| opendevreview | Jay Faulkner proposed openstack/ironic-specs master: Add Ironic work items for 2023.1 https://review.opendev.org/c/openstack/ironic-specs/+/862538 | 21:06 |
| opendevreview | Merged openstack/sushy stable/zed: Improve resiliency of eTag handling https://review.opendev.org/c/openstack/sushy/+/862846 | 21:14 |
| opendevreview | Merged openstack/sushy stable/yoga: Improve resiliency of eTag handling https://review.opendev.org/c/openstack/sushy/+/862847 | 22:20 |
| JayF | sushy gates are butter | 22:20 |
| TheJulia | Nice | 22:29 |
| opendevreview | Merged openstack/sushy stable/xena: Improve resiliency of eTag handling https://review.opendev.org/c/openstack/sushy/+/862848 | 22:31 |
| TheJulia | \o/ 5 passing integration tests | 22:55 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!