| opendevreview | chandan kumar proposed openstack/cyborg-specs master: Add generic NVMe driver spec with secure cleanup https://review.opendev.org/c/openstack/cyborg-specs/+/985349 | 08:12 |
|---|---|---|
| opendevreview | chandan kumar proposed openstack/cyborg-specs master: Add generic NVMe driver spec with secure cleanup https://review.opendev.org/c/openstack/cyborg-specs/+/985349 | 08:23 |
| chandankumar | sean-k-mooney: jgilaber hello, https://review.opendev.org/c/openstack/cyborg-specs/+/985349: Add generic NVMe driver spec with secure cleanup is ready for review now, feel free to add it to your review queue. thank you! | 08:29 |
| opendevreview | chandan kumar proposed openstack/cyborg master: Add VM setup documentation for Cyborg development https://review.opendev.org/c/openstack/cyborg/+/983964 | 09:01 |
| opendevreview | chandan kumar proposed openstack/cyborg master: Add Cyborg NVMe driver development environment documentation https://review.opendev.org/c/openstack/cyborg/+/982711 | 09:02 |
| opendevreview | chandan kumar proposed openstack/cyborg master: Add user guide for creating instances with accelerators https://review.opendev.org/c/openstack/cyborg/+/983027 | 09:02 |
| opendevreview | chandan kumar proposed openstack/cyborg master: Add doc around cyborg tempest test and grenade testing https://review.opendev.org/c/openstack/cyborg/+/983030 | 09:02 |
| sean-k-mooney[m] | [chandankumar](https://matrix.to/#/@_oftc_chandankumar:matrix.org) i have read over the spec. ill review it again later but there are a number of things i think we should change | 09:44 |
| sean-k-mooney[m] | i do not want to reuse disable and enable to also track cleaning i also dont want to add deiver specific columns to the device table i.e. cleaning format | 09:46 |
| sean-k-mooney[m] | i also dont thing pooling and parsing a sanatisation log is a good approch and i have reservation about the file locks being implemented in cyborg | 09:47 |
| sean-k-mooney[m] | the core seams ok but there is a non trivial amount if duplication and some non relevent info in the spec that should be removed | 09:49 |
| sean-k-mooney[m] | ill do a proper review and leave comments inline in the next day or two | 09:49 |
| chandankumar | thank you sean-k-mooney[m] for the feedback, I will wait for your comments! | 10:01 |
| chandankumar | sean-k-mooney: jgilaber based on discussion during PTG around PCI driver, I have added https://bugs.launchpad.net/openstack-cyborg/+bug/2152545 to track all the duplicate code around lspci and its parsing code. | 12:48 |
| chandankumar | I will add this bug to the upcoming cyborg meeting for discussion on how to move forward. | 12:49 |
| sean-k-mooney | ack | 12:49 |
| jgilaber | ack, thanks for documenting it chandankumar | 12:49 |
| sean-k-mooney | i have been debating if i will start on the v2 driver this release or not | 12:49 |
| sean-k-mooney | i may owrk on this dedup | 12:49 |
| sean-k-mooney | but im currently tyring ot blance soem pros and cons | 12:50 |
| sean-k-mooney | one thing i want to look at is what external tools we use today | 12:50 |
| sean-k-mooney | what driver specific deps we have in general and if those are proeprly modeled in extras and bindep | 12:51 |
| sean-k-mooney | lspci i think is a resonabel depency to take | 12:51 |
| sean-k-mooney | as is nmcli | 12:51 |
| sean-k-mooney | but beyond the code duplciation i want ot make sure we are nto includign tools that are vendor specific if we dont need them | 12:52 |
| sean-k-mooney | so im really not sure how to scoep the tech debt reduction here | 12:52 |
| sean-k-mooney | i.e. shoudl we start with just the lspci cleanup | 12:52 |
| sean-k-mooney | or do a borader pass | 12:52 |
| sean-k-mooney | even if we do a broader pass we will want to break down the clean up into small commits | 12:53 |
| sean-k-mooney | so we can use this bug as a tracker in anycase | 12:53 |
| sean-k-mooney | yesterday i spend some time workign on my poc of a kernel modlule to emulsate sriov | 12:53 |
| sean-k-mooney | and i have a workign poc end to end that shoudl work in upstream ci | 12:54 |
| sean-k-mooney | so im also thinkign about this form a testign perspective | 12:54 |
| sean-k-mooney | with that we coudl have temepst test that dont need real hardware and fully test the pci driver | 12:54 |
| sean-k-mooney | i might be able to exend it to supprot other types fo devices liek gpus or nvme eventually too | 12:55 |
| chandankumar | ah, that would be great | 12:57 |
| chandankumar | But we still need iommu enabled on the testing node. | 12:57 |
| sean-k-mooney | the two things i was planning to focus on this cycle were srbac and the pci driver but we can dicuss priority i guess here or in the irc meetign | 12:57 |
| sean-k-mooney | chandankumar: no we dont | 12:57 |
| sean-k-mooney | chandankumar: i(gpt 5.5 at my direct) implemente a softiommu implenation that make it work without needing viommu | 12:58 |
| sean-k-mooney | my current approch shoudl work in a standard vm that we get form zuul | 12:58 |
| sean-k-mooney | but i need to test that end to end :) | 12:59 |
| sean-k-mooney | im condiering pocing that over then next week or so | 12:59 |
| chandankumar | great, that will do the job. | 12:59 |
| opendevreview | Takashi Kajinami proposed openstack/cyborg master: Remove ineffective static_root https://review.opendev.org/c/openstack/cyborg/+/987095 | 13:32 |
| opendevreview | Takashi Kajinami proposed openstack/cyborg master: sqlalchemy: Use built-in declarative https://review.opendev.org/c/openstack/cyborg/+/987089 | 13:39 |
| opendevreview | Takashi Kajinami proposed openstack/cyborg master: sqlalchemy: Use built-in declarative https://review.opendev.org/c/openstack/cyborg/+/987089 | 14:55 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!