| clarkb | it is surprisingly difficult to get bookworm's dkms packages to build with gcc 11 instead of 12 | 00:39 |
|---|---|---|
| clarkb | changing gcc/cc to gcc-11 didn't do it. Uninstalling gcc-12 removes the linux kernel headers. | 00:39 |
| clarkb | I know why they do it, the kernel was built with 12 so you should build things with 12 too. But I wanted to see if gcc 11 produces a different relocation result in the elf | 00:40 |
| *** travisholton0 is now known as travisholton | 00:41 | |
| clarkb | I think I'm going to give up on this for now. Maybe tomorrow we put together something we can file against debian and point to the nixos problem as being similar/the same | 00:42 |
| clarkb | I've learned a bit about things though which is nice I guess. | 00:42 |
| clarkb | extern struct key_type key_type_keyring __attribute__((weak)); | 01:07 |
| clarkb | my latest suspicion is that we're dealing with the __attribute__((weak)); definition here | 01:07 |
| clarkb | I suspect that the linker has done some relocation magic to resolve the weak key_type_keyring definition | 01:09 |
| clarkb | which I think is actually in the kernel | 01:09 |
| clarkb | ya the linux kernel seems to define it in security/keys/keyring.c | 01:10 |
| clarkb | there is another struct defined in a similar way but in the case of key_type_keying we make a pointer to point at its address in openafs: static struct key_type *__key_type_keyring = &key_type_keyring; so we've got to resolve things more concretely than the other weakly defined symbol? | 01:13 |
| clarkb | I half wonder if we can just refer to the address of key_type_keyring within the function there and have that move outside of the global stuff in the elf | 01:15 |
| clarkb | since we'll be handling that in CODE instead | 01:15 |
| clarkb | ? | 01:15 |
| clarkb | ok I said I was done but then I couldn't help myself. I was able to build a libafs.ko that insmodded by dropping that __attribute__((weak)); | 01:52 |
| clarkb | that said I don't know if the key_type_keyring is the one I want to be referring to after all the linking is done | 01:55 |
| clarkb | but it is defined extern so in theory it is locating the variable defined by the kernel? | 01:55 |
| clarkb | and now I'm actually going to stop as I'm slightly paranoid I could do something to make afs unhappy. The modified and loadable module has been rmmod'd and I never actually performed any actions against afs other than insmodding the module | 02:17 |
| clarkb | I think if an unhappy client could create problems we would have bigger issues, but I'd rather not find out right now and let it be for now | 02:17 |
| *** diablo_rojo is now known as Guest500 | 03:09 | |
| *** enick_952 is now known as diablo_rojo | 03:09 | |
| fungi | just a heads up, whatever crud i've come down with is still making me a bit spacey... i'll try to be around some today, but also trying to sleep this off | 13:38 |
| clarkb | I've joined the openafs channel on libera and did a summarized braindump of what I learned yesterday there | 15:01 |
| clarkb | someone in the openafs channel reports they will try to reproduce and if they manage to attempt to fix it | 15:50 |
| fungi | yay! | 16:06 |
| frickler | corvus: in https://zuul.opendev.org/t/openstack/project/opendev.org/openstack/kolla-ansible I'm seeing two tabs per branch with different content. not sure if that is somehow intentional or a bug? | 16:07 |
| frickler | oh, kolla even has six tabs per branch. might be related to those being defined in multiple files in .zuul.d/* ? | 16:08 |
| corvus | yeah those are project stanzas, the tab titles are misleading/incomplete | 16:16 |
| frickler | corvus: now that you mention this, I think I may have asked this already earlier and forgotten about it. would it be possible to include the file reference somewhere or does that get lost while parsing the config? | 16:25 |
| clarkb | fungi: did you want to weigh in on dropping opensuse functional jobs for bindep: https://review.opendev.org/c/opendev/bindep/+/909771 we have coverage in the unittests which should be sufficient going forward | 16:26 |
| fungi | yep, sorry, been a bit out of it yesterday/today | 16:27 |
| clarkb | yup understood. More just wanting to avoid approving it if you are interested in looking it over | 16:28 |
| fungi | i was hoping to find time to review them all | 16:28 |
| clarkb | ok I did approve the bindep fallback job removal for opensuse and setting of max-servers to 0 | 16:28 |
| clarkb | I'll hold off on additional approvals so that you can take a look at the changes | 16:29 |
| fungi | but if there are already enough reviews, don't hold it up waiting for me | 16:29 |
| opendevreview | Merged openstack/project-config master: Drop bindep fallback testing on opensuse https://review.opendev.org/c/openstack/project-config/+/909775 | 16:34 |
| corvus | frickler: possible, i think | 16:36 |
| opendevreview | Merged openstack/project-config master: Set opensuse-15 min-ready to 0 https://review.opendev.org/c/openstack/project-config/+/909774 | 16:37 |
| opendevreview | Merged opendev/bindep master: Drop bindep-opensuse-15 job https://review.opendev.org/c/opendev/bindep/+/909771 | 17:29 |
| opendevreview | Merged zuul/zuul-jobs master: Drop opensuse-15 jobs https://review.opendev.org/c/zuul/zuul-jobs/+/909772 | 17:41 |
| clarkb | https://review.opendev.org/c/openstack/openstack-zuul-jobs/+/909770 is that last change remaining before I think we can start forcibly removing opensuse 15 | 17:43 |
| clarkb | its the change above then https://review.opendev.org/c/opendev/base-jobs/+/909773 then the nodepool cleanup then the mirror cleanup | 17:43 |
| clarkb | woot the ozj change merged. fungi has reviewed 909773. Any objections to me approving it now? | 17:57 |
| clarkb | I think this will potentially create problems for keystone, devstack, and tempest | 17:57 |
| fungi | i just didn't want to approve it before the deps were merged | 17:57 |
| fungi | approved now | 17:58 |
| clarkb | makes sense. Its also the one that will have an impact if any do | 17:58 |
| fungi | we can push up patches to remove the lingering bits in those projects if necessary | 17:58 |
| clarkb | ya I've got a few up. But I'm not going to bother pushing changes for every stable branch | 18:01 |
| clarkb | particular in the case of opensuse these jobs haven't really ever worked and I don't feel like we should be responsible for figuring out how to clean that up | 18:02 |
| clarkb | that does bring up an idea for me: maybe the openstack release process should have a "prune your zuul configs before branching occurs so that we don't carry old unused broken stuff forward another branch" | 18:02 |
| opendevreview | Merged opendev/base-jobs master: Drop the opensuse-15 nodeset https://review.opendev.org/c/opendev/base-jobs/+/909773 | 18:06 |
| clarkb | alright the next step is to remove the nodes and images. I'm going to make sure that all the booted nodes are gone | 18:09 |
| clarkb | corvus: it should be safe to remove the image definitions and the node defs from nodepool all at once right? re https://review.opendev.org/c/openstack/project-config/+/909776 | 18:11 |
| clarkb | There are no opensuse nodes according to nodepool list | 18:12 |
| clarkb | so I think 909776 is ready to go assuming we're happy with removing the image and node pools all at once | 18:12 |
| clarkb | er not pools but opensuse from pools | 18:13 |
| frickler | 30 new config errors due to missing opensuse nodeset. but that was kind of to be expected, I'm not complaining, just noticing | 18:14 |
| clarkb | the bifrost errors were not ones I had previously anticipated. The others are | 18:15 |
| frickler | iirc there was an intermediate step to be made for image removals in order to allow nodepool to remove them, but I'm not sure about the details | 18:17 |
| clarkb | oh right we need to remove them from the pool defs so that we delete the images from the clouds | 18:17 |
| clarkb | then we can remove the image from the builders | 18:17 |
| clarkb | I'll split the change | 18:17 |
| frickler | or maybe I'm mixing that up with removing a provider https://zuul-ci.org/docs/nodepool/latest/operation.html#removing-a-provider | 18:18 |
| * clarkb reads | 18:18 | |
| clarkb | I can do it that way to make sure we don't introduce unhappyness cleaning stuff up | 18:19 |
| opendevreview | Clark Boylan proposed openstack/project-config master: Drop opensuse image uploads from nodepool https://review.opendev.org/c/openstack/project-config/+/909776 | 18:28 |
| opendevreview | Clark Boylan proposed openstack/project-config master: Remove opensuse-15 image builds from nodepool https://review.opendev.org/c/openstack/project-config/+/909921 | 18:28 |
| clarkb | frickler: ^ I think that should do it | 18:28 |
| corvus | lgtm i approved the 1st | 19:09 |
| clarkb | thanks! | 19:09 |
| opendevreview | Merged openstack/project-config master: Drop opensuse image uploads from nodepool https://review.opendev.org/c/openstack/project-config/+/909776 | 19:21 |
| clarkb | I think the opensuse image uploads are gone now. I'm going to take a berak for lunch and a bike right though if we want ot hold off on approving the second change to cleanup nodepool config | 19:39 |
| clarkb | either way is fine with me | 19:39 |
| frickler | clarkb: when you return maybe you can also check stable branch cleanups at https://review.opendev.org/c/openstack/openstack-zuul-jobs/+/909606 (2 patches) | 19:44 |
| frickler | nodepool image-list doesn't show any opensuse images, so I'll approve the 2nd patch | 19:46 |
| opendevreview | Merged openstack/project-config master: Remove opensuse-15 image builds from nodepool https://review.opendev.org/c/openstack/project-config/+/909921 | 19:54 |
| clarkb | frickler: done | 23:09 |
| clarkb | infra-root https://review.opendev.org/c/opendev/system-config/+/909779 is the last step on our end for opensuse cleanup | 23:11 |
| clarkb | the final opensuse mirroring script will get cleaned up when we remove centos 7 since that script syncs some obs stuff for centos 7 | 23:11 |
| clarkb | but first I need to finish getting changes up to remove buster | 23:11 |
| clarkb | (since I said I would do buster before centos 7 in a couple of places) | 23:11 |
| clarkb | there are a handful of dib .d dirs and hash files for opensuse images on nb01 and nb02. I'll clean those up now | 23:13 |
| opendevreview | Clark Boylan proposed opendev/system-config master: Upgrade gitea to 1.21.6 https://review.opendev.org/c/opendev/system-config/+/909941 | 23:28 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!