| opendevreview | Verification of a change to openstack/ironic master failed: Introduce conductor touch while offline https://review.opendev.org/c/openstack/ironic/+/897375 | 00:58 |
|---|---|---|
| opendevreview | Verification of a change to openstack/ironic stable/yoga failed: redfish_address - wrap_ipv6 address https://review.opendev.org/c/openstack/ironic/+/896209 | 01:06 |
| opendevreview | Verification of a change to openstack/ironic stable/xena failed: redfish_address - wrap_ipv6 address https://review.opendev.org/c/openstack/ironic/+/896210 | 01:31 |
| opendevreview | Merged openstack/ironic master: Introduce conductor touch while offline https://review.opendev.org/c/openstack/ironic/+/897375 | 02:27 |
| opendevreview | Merged openstack/ironic stable/yoga: redfish_address - wrap_ipv6 address https://review.opendev.org/c/openstack/ironic/+/896209 | 03:13 |
| opendevreview | Merged openstack/ironic stable/wallaby: redfish_address - wrap_ipv6 address https://review.opendev.org/c/openstack/ironic/+/896376 | 03:13 |
| opendevreview | Verification of a change to openstack/ironic stable/xena failed: redfish_address - wrap_ipv6 address https://review.opendev.org/c/openstack/ironic/+/896210 | 03:19 |
| opendevreview | Boushra Sondos Bettir proposed openstack/ironic-python-agent master: Replace shlex module with helper function https://review.opendev.org/c/openstack/ironic-python-agent/+/898058 | 05:20 |
| rpittau | good morning ironic! o/ | 07:14 |
| opendevreview | Adam Rozman proposed openstack/ironic master: implement basic-auth support for user image download process https://review.opendev.org/c/openstack/ironic/+/890411 | 07:41 |
| opendevreview | Adam Rozman proposed openstack/ironic master: implement basic-auth support for user image download process https://review.opendev.org/c/openstack/ironic/+/890411 | 07:43 |
| rpittau | dtantsur: good timing! :D | 08:03 |
| dtantsur | mmm? | 08:03 |
| rpittau | this > https://review.opendev.org/c/openstack/ironic-python-agent/+/898058 | 08:03 |
| dtantsur | ahhh, I was 1 minute earlier, lol | 08:04 |
| rpittau | yeah :D | 08:05 |
| dtantsur | Need to also put the Adam's image auth patches on the queue. I suspect he'll have an issue with image_download_source=local and its lack of authenticaiton support, just like we hit it with the new non-bootable image API | 08:05 |
| * dtantsur is pissed because weechat somehow broken nick autocompletion on tab | 08:06 | |
| opendevreview | Verification of a change to openstack/ironic stable/xena failed: redfish_address - wrap_ipv6 address https://review.opendev.org/c/openstack/ironic/+/896210 | 08:31 |
| opendevreview | Mahnoor Asghar proposed openstack/ironic master: Add inspection hooks https://review.opendev.org/c/openstack/ironic/+/892661 | 08:57 |
| opendevreview | Merged openstack/ironic-python-agent stable/2023.2: Retry on checksum failures https://review.opendev.org/c/openstack/ironic-python-agent/+/897961 | 09:29 |
| opendevreview | Merged openstack/ironic-python-agent stable/2023.1: Retry on checksum failures https://review.opendev.org/c/openstack/ironic-python-agent/+/897950 | 09:29 |
| opendevreview | Merged openstack/ironic stable/xena: redfish_address - wrap_ipv6 address https://review.opendev.org/c/openstack/ironic/+/896210 | 11:52 |
| TheJulia | good morning | 13:38 |
| opendevreview | Julia Kreger proposed openstack/ironic-tempest-plugin master: WIP: Add test for dhcp-less vmedia based deployment https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/898006 | 13:48 |
| opendevreview | Julia Kreger proposed openstack/ironic master: WIP/DNM: Advanced vmedia deployment test ops https://review.opendev.org/c/openstack/ironic/+/898010 | 13:49 |
| dtantsur | TheJulia: good morning, hopefully better than we have here in OCP :D | 14:08 |
| dtantsur | actually, PSA: https://issues.redhat.com/browse/RHEL-12870 | 14:08 |
| dtantsur | people may start seeing this in Stream 8 | 14:08 |
| TheJulia | wheeeeee | 14:27 |
| TheJulia | hmm, our API has improved mysteriously since the early days of me sending json patches to it | 14:32 |
| opendevreview | Julia Kreger proposed openstack/ironic-tempest-plugin master: WIP: Add test for dhcp-less vmedia based deployment https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/898006 | 14:34 |
| TheJulia | iurygregory: I have a question for you on https://review.opendev.org/c/openstack/ironic/+/897989, maybe I'm not understanding why we have to special case to continue. | 14:43 |
| TheJulia | well, not to continue, as it were | 14:43 |
| -opendevstatus- NOTICE: The lists.openstack.org site will be offline over the next few hours for migration to a new server | 15:31 | |
| opendevreview | Julia Kreger proposed openstack/ironic-tempest-plugin master: WIP: Add test for dhcp-less vmedia based deployment https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/898006 | 15:42 |
| TheJulia | \o/ | 15:42 |
| rpittau | good night! o/ | 15:50 |
| JayF | \o | 15:56 |
| dtantsur | So folks, when is the right time to migrate off eventlet? :D | 16:30 |
| JayF | I have actually thought about this and come to the conclusion the only way that actually happens is if someone wrote an eventlet-compatible library mapping to one of the real-async libraries | 16:32 |
| dtantsur | I don't think monkey-patching makes it possible | 16:32 |
| JayF | because mixing greenthreads and actual threads is likely a really bad idea and eventlet is baked in deep, deep, deep to the codebase | 16:32 |
| dtantsur | (and if we keep the monkey patching, we keep all the downsides) | 16:32 |
| dtantsur | mmm, mixing greenthreads and actual threads is a great idea, eventlet just cannot do it | 16:32 |
| dtantsur | (both Go and Rust's Tokyo can) | 16:33 |
| JayF | my experience is around eventlet/greenlet | 16:33 |
| JayF | so that tracks :D | 16:33 |
| JayF | extremely unpopular opinion: we'd be in a MASSIVELY better place if OpenStack was written around twisted instead :D | 16:34 |
| JayF | lol | 16:34 |
| dtantsur | I will side with you on this one | 16:34 |
| dtantsur | https://github.com/eventlet/eventlet/issues/673 (the temoto's suggestion) make me think that we could potentially isolate an asyncio executor in a green thread | 16:36 |
| JayF | honestly I only feel that way because I know it was migrated to some of the native async stuff on the backend, and I'm friends with Glyph so I know someone would care about our bugs ;) | 16:36 |
| dtantsur | and then grow the asyncio part and reduce the eventlet one | 16:36 |
| dtantsur | I've done a lot of Twisted in both of my previous jobs | 16:37 |
| clarkb | JayF: are you aware of the history there with twisted and eventlet? | 16:37 |
| JayF | No, actually. I've written basically zero anything in twisted despite Glyph teaching me quite a bit of the advanced python stuff I know now | 16:38 |
| JayF | b/c he was at a desk about 20 ft from me at Rackspace SF :D | 16:38 |
| clarkb | tl;dr nova was twisted, swift was eventlet. Before I got involved apparently the decisions was made to standardize on eventlet. Then a few yaers later swift realized eventlet was not the greatest and that led to the go experiment | 16:38 |
| JayF | I traded devops knowledge for python knowledge | 16:38 |
| JayF | oh :( | 16:38 |
| dtantsur | btw how did the go experiment go? | 16:38 |
| JayF | dtantsur: I believe rackspace ran the go fork downstream in production for ... as long as I was aware? | 16:39 |
| JayF | dtantsur: but was a situation where I think they stopped investing in it once they got the API coverage they cared about implemented | 16:39 |
| dtantsur | I see | 16:39 |
| clarkb | ya the code is out there somewhere but it never ended up in swift proper aiui | 16:39 |
| clarkb | funny though go has the same underlying issue though | 16:39 |
| clarkb | it just scales up better on top of the greentrhead design so you notice less | 16:40 |
| JayF | https://github.com/troubling/hummingbird | 16:40 |
| dtantsur | clarkb, IIUC it uses an M:N thread model | 16:40 |
| dtantsur | which can result in much better scaling with CPU numbers | 16:40 |
| dtantsur | but also, they don't deal with non-sense like "eventlet overriding things the way it no longer works" like in https://issues.redhat.com/browse/RHEL-12870 | 16:41 |
| clarkb | dtantsur: yup and you don't get strict control of it which is the problem aiui. Specifically disk/fs io is often blockign (there are no non blocking options) which means you basically need to park an entire thread for it | 16:41 |
| clarkb | both go and python make that painful | 16:41 |
| clarkb | go just happens to not hit the limits nearly as quickly so its probably fine | 16:41 |
| dtantsur | okay, you convinced me, Rust it is :D | 16:41 |
| JayF | I am not 100% sure because I mostly worked on the outskirts of the code at this job | 16:41 |
| JayF | but I worked at a place that did a lot of go that interfaced with the DB | 16:41 |
| JayF | and IIRC we did a completely separate set of goroutines off a queue for the actual DB write bits | 16:41 |
| JayF | like ensured no matter what I/O never happened in the primary set of greenthreads | 16:42 |
| dtantsur | Maybe not using OS threads was a mistake to begin with.. | 16:43 |
| JayF | I honestly don't know, all these problems are hard and it's easy to pretend the alternatives don't have their own set of problems but they do; we just don't know what they are | 16:45 |
| dtantsur | True. I'm just upset because I spent the whole day chasing this eventlet regression | 16:46 |
| dtantsur | (I wonder how many people will see it, potentially everyone still using Stream 8 / RHEL 8) | 16:46 |
| JayF | I mean, that's unfortunate; but it is part of the burden taken on when a vendor wants to support python for longer. | 16:47 |
| JayF | I can't imagine that was a pleasant one to track either | 16:48 |
| * JayF wishes "apt-get install python3.8" would be a valid fix for those folks | 16:48 | |
| JayF | I've been running Gentoo desktops/homelabs and not enough actual enterprise linux servers for too long ;) | 16:48 |
| dtantsur | :) | 16:49 |
| TheJulia | *sigh* "Network 530dd336-f70a-4339-ad68-2aaf704d186b could not be found." | 16:50 |
| opendevreview | Julia Kreger proposed openstack/ironic-tempest-plugin master: WIP: Add test for dhcp-less vmedia based deployment https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/898006 | 17:29 |
| TheJulia | woot, my dhcp-less ramdisk booted | 18:55 |
| TheJulia | muahahahahaha | 18:55 |
| opendevreview | Julia Kreger proposed openstack/ironic master: WIP/DNM: Advanced vmedia deployment test ops https://review.opendev.org/c/openstack/ironic/+/898010 | 19:09 |
| opendevreview | Boushra Sondos Bettir proposed openstack/ironic-python-agent master: Replace shlex module with helper function https://review.opendev.org/c/openstack/ironic-python-agent/+/898058 | 19:52 |
| opendevreview | Julia Kreger proposed openstack/ironic master: WIP/DNM: Advanced vmedia deployment test ops https://review.opendev.org/c/openstack/ironic/+/898010 | 20:20 |
| opendevreview | Boushra Sondos Bettir proposed openstack/ironic-python-agent master: Replace shlex module with helper function https://review.opendev.org/c/openstack/ironic-python-agent/+/898058 | 20:37 |
| opendevreview | Julia Kreger proposed openstack/ironic master: WIP/DNM: Advanced vmedia deployment test ops https://review.opendev.org/c/openstack/ironic/+/898010 | 22:59 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!