2023-12-01 10:13:35 <ncopa> i would like to rename dl-cdn.alpinelinux.org to cdn.alpinelinux.org and add a CNAME dl-cdn. where do I do that?
2023-12-01 10:13:46 <ncopa> we also need to update the https cert
2023-12-01 10:15:52 <ikke> ncopa: before we do that, we need to make sure fastly accepts it
2023-12-01 10:18:55 <ncopa> ok. how do we do that?
2023-12-01 10:19:11 <ikke> https://gitlab.alpinelinux.org/alpine/infra/linode-tf/-/blob/master/domains_mirrors.tf?ref_type=heads#L65
2023-12-01 10:19:31 <ikke> dl-cdn is already a cname, so we would just add another 
2023-12-01 10:19:52 <ncopa> aha
2023-12-01 10:19:54 <ncopa> good
2023-12-01 10:21:12 <ncopa> do you want me to create an MR?
2023-12-01 10:22:17 <ikke> Yes please 
2023-12-01 10:29:32 <ncopa> https://gitlab.alpinelinux.org/alpine/infra/linode-tf/-/merge_requests/76
2023-12-01 10:30:05 <ncopa> maybe I should have used a branch in the same repo instead of forking it?
2023-12-01 10:32:15 <ikke> Yeah
2023-12-01 10:37:26 <ncopa> should I create a new one?
2023-12-01 10:41:29 <ikke> Once merged, it will work 
2023-12-01 10:41:49 <ncopa> can i merge it?
2023-12-01 10:42:19 <ncopa> we may need to add it to fastly config too, under domains
2023-12-01 10:43:37 <ikke> And a cert 
2023-12-01 11:18:05 <ikke> ncopa: I ran the deploy job
2023-12-01 11:18:35 <ncopa> thanks!
2023-12-01 12:12:46 <mps> I skimmed through latest ncurses changes and I see some fixes for terminfo and small changes, no ABI change. Maybe we should upgrade it for 3.19 release
2023-12-01 12:55:05 <ncopa> mps: sure. lets go for it
2023-12-01 13:32:36 <mps> ok
2023-12-01 13:39:22 <mps> !56425
2023-12-01 14:30:48 <mps> ncopa: should we try to revert changes in linux-firmware which requires rdfind !55436
2023-12-01 14:32:09 <ncopa> https://gitlab.alpinelinux.org/alpine/aports/-/commit/6e480e3ec55e091d00d9f1d0ed8f8a52ffa1e9d9
2023-12-01 14:32:26 <mps> oh, it is moved
2023-12-01 14:33:12 <ncopa> you just need to rebase it
2023-12-01 14:34:03 <mps> wouldn't simple merge work
2023-12-01 14:34:37 <mps> ah, you mean rebase in gitlab
2023-12-01 14:45:27 <mps> hm, 'fast-forward merge is not possible. To merge this request, first rebase locally.' how to fix this
2023-12-01 14:46:13 <mps> I tried `git rebase master linux-firmware-20231111` but it fails
2023-12-01 14:52:24 <mps> guides on internet about this are cumbersome
2023-12-01 14:54:21 <mps> hmm, manually only
2023-12-01 15:04:38 <ikke> Does it fail because of merge conflicts? Because that's exactly why you need to fix it locally 
2023-12-01 15:07:37 <mps> ikke: idk, gitlab says 'fast-forward merge is not possible. To merge this request, first rebase locally.'
2023-12-01 15:07:51 <ikke> That means there are conflicts 
2023-12-01 15:08:03 <mps> I fixed it manually with 'trick'
2023-12-01 15:08:43 <mps> and now I understand little better this ... uhm ...
2023-12-01 15:09:10 <mps> now it could be merged
2023-12-01 15:09:23 <mps> !55436
2023-12-01 15:12:28 <mps> last 'thing' in my branches for 3.19 is librtlsdr
2023-12-03 09:41:04 <mps> ncopa: I read your release announce mail. I think we should merge !56293 and !54530
2023-12-03 10:10:46 <ncopa> I don’t get why you insist on Linux-headers. What benefit does that bring, given that all packages are already built? What is the downside of waiting til after 3.19?
2023-12-03 10:13:54 <mps> nothing should be rebuilt for release if headers are installed
2023-12-03 10:14:29 <ncopa> ?
2023-12-03 10:15:20 <mps> upgrading linux-headers doesn't require rebuild any package
2023-12-03 10:16:27 <ncopa> true. But what value does it provide to update it?
2023-12-03 10:17:13 <ncopa> it does not give users any new features. It does not simplify maintenance for the 3.19 branch
2023-12-03 10:17:42 <mps> if we upgrade it for 3.19 then when someone want to build locally newer packages which require upgraded UAPI it would be useful
2023-12-03 10:18:16 <mps> only this is reason I think it could upgraded
2023-12-03 10:18:48 <mps> for me it is not problem, I already upgraded on all my machine
2023-12-03 10:18:50 <ncopa> Those users can update the headers themselves from edge?
2023-12-03 10:19:08 <mps> right
2023-12-03 10:19:43 <ncopa> also, often apps does #ifndef … #define MiSSING …
2023-12-03 10:19:56 <mps> but will give 'not so nice image' about alpine, kernel is 6.6 and headers are older
2023-12-03 10:19:59 <ncopa> so they get the missing UAPI
2023-12-03 10:20:26 <mps> my reason is only 'marketing'
2023-12-03 10:20:38 <ncopa> then I don’t think it is worth it
2023-12-03 10:21:03 <ncopa> updating headers is a risk of complicating future maintenance 
2023-12-03 10:21:07 <mps> ok, however you like
2023-12-03 10:23:04 <mps> btw, kernel 6.6.4 is released few hours ago
2023-12-03 10:23:32 <mps> hope we will not have 6.6.6 version :)
2023-12-03 15:40:14 <mps> do builders works
2023-12-03 15:41:33 <mps> algitbot: retry master
2023-12-03 15:42:27 <mps> looks like some 'hangs' on uploading pkgs
2023-12-03 15:43:53 <mps> aha, now they work
2023-12-03 16:17:04 <mps> algitbot: retry master
2023-12-04 14:59:47 <celie> I know everyone is busy with 3.19, so i hesitate to make such a request, but would it be possible to get !56298 merged in time for 3.19, so we don't ship 3.19 with a version of Perl that upstream considers "broken"
2023-12-04 15:10:09 <mps> celie: does something block it? also I think it *must* be merged to 3.19
2023-12-04 15:10:27 <ncopa> I'm looking at shipping rpi disk images. did we get any consensus where to put generic cloud images? https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/52993
2023-12-04 15:10:55 <ncopa> i merged !56298
2023-12-04 15:11:05 <mps> nice
2023-12-04 15:11:21 <ncopa> ikke: where will we store the cloud images? how are they generated?
2023-12-04 15:11:55 <celie> Thanks
2023-12-04 15:12:11 <ncopa> i wonder if we could generate the rpi images together with the cloud images, as they will likely be geneated similar way
2023-12-04 15:12:19 <mps> when is last day for 3.19
2023-12-04 15:12:30 <ncopa> depends on the ^^^
2023-12-04 15:12:56 <ncopa> im ready for release, but woudl be nice to also include the rpi imager support
2023-12-04 15:13:08 <ncopa> but we need to figure out how to do those releases
2023-12-04 15:14:39 <mps> I have no idea about this and how can help
2023-12-04 15:14:43 <mps> if at all
2023-12-04 15:14:59 <ncopa> i think there was an issue about where to pu thte clud images. but i cannot find it
2023-12-04 15:15:23 <mps> ah, yell at ikke :)
2023-12-04 15:15:26 <ikke> ncopa: they are currently generated by tomalok. I'm working in a process to pull them from a location and push them to dl-master 
2023-12-04 15:16:09 <ncopa> do we have an open issue/MR/email-thread/something about what the plan there is?
2023-12-04 15:16:23 <ncopa> because the rpi images will be similar
2023-12-04 15:16:46 <ikke> There should be an issue in the infra project 
2023-12-04 15:17:23 <ncopa> i think i found it
2023-12-04 15:31:18 <ikke> ncopa: eventually the goal is to automate this in a pipeline 
2023-12-04 15:31:27 <ncopa> understood
2023-12-04 16:23:33 <ikke> ncopa: What was your idea for generating the rpi images?
2023-12-04 16:25:09 <ncopa> im currently at create a .img.xz from scripts/mkimg.*
2023-12-04 16:25:40 <ncopa> its a simple fat image, no partition table (afaics)
2023-12-04 16:26:08 <ncopa> then will ship the json with alpine-mksite
2023-12-04 16:36:43 <ikke> So it would not be too difficult to generate such an image in a pipeline
2023-12-04 16:51:14 <ncopa> hum. it was harder than anticipated
2023-12-04 16:51:43 <ncopa> to create alpine-rpi-*.tar.gz + alpine-rpi-*.img.xz
2023-12-04 20:10:21 <ncopa6> alpine-rpi.img.xz is around 60MB
2023-12-04 20:12:26 <ikke> ncopa: I may know what happened with the secfixes-tracker
2023-12-04 20:12:47 <ikke> The cron job runs different tasks in the background at the same time
2023-12-04 20:12:56 <ncopa> oh
2023-12-04 20:12:56 <ikke> And they run into locked databases
2023-12-04 20:13:04 <ncopa> so a race condition
2023-12-04 20:13:07 <ikke> Due to commits being deferred
2023-12-04 20:13:10 <ikke> yes
2023-12-04 20:13:15 <ikke> well, not race conditions
2023-12-04 20:13:17 <ikke> just plain errors
2023-12-04 20:13:36 <ikke> update-states was run seperately as last
2023-12-04 20:13:46 <ncopa> so sqlite does not queue the transactions
2023-12-04 20:14:03 <ncopa> maybe we should switch to postgresql or something
2023-12-04 20:14:07 <ikke> sqlite3.OperationalError: database is locked
2023-12-04 20:14:18 <vkrishn> thoughts for rpi images, /alpine/v3.19/rpi.releases
2023-12-04 20:14:41 <ikke> ncopa: since each job is running a lot faster now, I guess it makes sense to just run them sequentiallyh
2023-12-04 20:15:00 <ncopa> ikke: cant we make it block until its unlocked?
2023-12-04 20:15:29 <ncopa> i suppose we could have gotten into the same issue before, but it would only be a single entry that got lost
2023-12-04 20:15:39 <ncopa> so we wouldnt notice
2023-12-04 20:19:20 <ikke> I think the most simple solution is to just run them sequentially and avoid the problem completely
2023-12-04 20:19:26 <ncopa> i like that
2023-12-04 20:20:05 <ikke> I've switched to the new NVD API now
2023-12-04 20:20:10 <ncopa> but we may run into the problem if we connet it to an mqtt hook
2023-12-04 20:20:12 <ncopa> nice!
2023-12-04 20:20:13 <ncopa> thanks!
2023-12-04 20:20:31 <ikke> We don't need to run all jobs then
2023-12-04 20:20:37 <ikke> just the import-secfixes job
2023-12-04 20:20:44 <ikke> and afterwards update-states
2023-12-04 20:22:25 <ikke> Just as a heads-up, I didn't implement pagination yet. The default page size is 2000 entries, which is plenty for getting the changes for the last 48h
2023-12-04 20:22:54 <ikke> The advantage of the new API is that we do not need to separately import yearly feeds anymore
2023-12-04 20:23:14 <ikke> updates and new CVEs all come in through the same feed
2023-12-04 20:23:35 <ncopa> https://bobbyhadz.com/blog/operational-error-database-is-locked#try-to-increase-the-timeout
2023-12-04 20:23:50 <ncopa> i suppose we can increase the timeout to a few minutes
2023-12-04 20:24:19 <ncopa> looks like the default is 5 seconds
2023-12-04 20:24:40 <ikke> I'm now running it with all jobs sequentially, and that works fine
2023-12-04 20:25:05 <ncopa> i was thinking for importing secfixes updates from git push
2023-12-04 20:25:14 <ncopa> which may happen at any time
2023-12-04 20:25:35 <ncopa> those will probably not need to hold the db for more than 5 sec though
2023-12-04 20:26:05 <ncopa> but if the long feed import is running, it may need more than 5 seconds
2023-12-04 20:27:01 <ncopa> very nice to have the cve feed fixes in place
2023-12-04 20:27:06 <ncopa> thank you alot!
2023-12-04 20:27:10 <ikke> np :)
2023-12-04 20:33:04 <ncopa> wow, the alpine-rpi.img actually works
2023-12-04 20:34:27 <ikke> nice
2023-12-04 20:35:10 <ikke> ncopa: What my idea was for the cloud images is to have a drop-off point on dev.a.o where tomalok can put the images, and a process that will fetch them from there and push it to dl-master
2023-12-04 20:35:24 <ikke> (This is what we came up in our meeting with Tomalok)
2023-12-04 20:35:38 <ncopa> yeah, i saw the summary in the ticket
2023-12-04 20:35:53 <ikke> Ok, good, wasn't sure what was in there
2023-12-04 20:36:05 <ncopa> the rpi img is simpler than a cloud image
2023-12-04 20:36:43 <ncopa> its only fat image without partition table, with the files
2023-12-04 20:36:43 <ikke> Are you going to handle that in the normal release process?
2023-12-04 20:36:47 <ncopa> yeah
2023-12-04 20:36:49 <ikke> ok
2023-12-04 20:37:07 <ncopa> the meta data for the rpi-imager needs to be handled other place
2023-12-04 20:37:16 <ncopa> and im going for simple and stupid there as well
2023-12-04 20:37:50 <ncopa> the metadata needs decompressed size, but the release process has no convienient way to provide that info
2023-12-04 20:38:14 <ncopa> so i plan to just download it and decompress it when generating meta data :)
2023-12-04 20:44:12 <ikke> Simple enough
2023-12-04 20:57:42 <ikke> I'm always very puzzled how these custom rewriters are supposed to work
2023-12-04 20:57:48 <ikke> It seems to just ignore it
2023-12-04 21:02:12 <ikke> Hmm, they are processed as part of the nvd import
2023-12-05 01:49:08 <tomalok> ncopa ikke - fwiw, i worked a bit on some revisions to alpinelinux.org/cloud templates to support download alongside launch (for the clouds that can do that) -- i think i got it sorted except i need to add my public key (used to sign images) and update the text at the top of the page
2023-12-05 01:50:51 <tomalok> apart from that all i need is (a) the go-ahead for building 3.19.0 images and (b) some dev.a.o directory to copy (symlink?) the image directory tree for rsync out to dl-master (etc)
2023-12-06 14:47:53 <mps> ncopa: should we mention in release notes that we now have everything in repositories for apple silicon
2023-12-06 15:29:50 <ncopa> can we add the 3.19 branch to pkgs.alpinelinux.org now? or do we need to wait for the final release?
2023-12-06 15:34:25 <ikke> We can already start it, but it takes a while 
2023-12-06 15:43:24 <clandmeter> should probably make it auto pickup
2023-12-06 17:57:17 <ikke> ncopa: running the update job now for 3.19
2023-12-07 11:15:03 <AljazS> Hello - I was told to join in here and ping *someone* in order to get my Alpine mirror added to the official mirror list.
2023-12-07 11:15:24 <AljazS> I was unable to find the minimum requirements for official mirrors - so I'd also like to know what those are?
2023-12-07 11:16:49 <AljazS> I am currently hosting the /3.19/ repo and I'd like to keep it so if possible.
2023-12-07 11:17:34 <AljazS> I've been hosting an archlinux repo for a few years now too (if this info helps in any way).
2023-12-07 11:18:31 <AljazS> And - the mirror is hosted on a server in a faculty's server room in Slovenia. On my own hardware. With a symetric 1Gbps link (in the enxt 6 months I should be getting a 10Gbps link too, not 100% confirmed yet tho)
2023-12-07 11:20:28 <mps> AljazS: somewhat old but ... https://wiki.alpinelinux.org/wiki/How_to_setup_a_Alpine_Linux_mirror
2023-12-07 11:21:12 <AljazS> mps: Yeah I already have it set up.
2023-12-07 11:21:23 <AljazS> I'd like to have it added to the mirror list if that's possible
2023-12-07 11:22:12 <mps> wait some time here, someone will answer I think
2023-12-07 11:22:27 <AljazS> +1
2023-12-07 11:22:32 <ikke> AljazS: we are a bit behind responding to mirror requests due to the 3.19 release
2023-12-07 11:22:59 <AljazS> Aight.
2023-12-07 11:23:48 <ikke> AljazS: I assume you already sent an e-mail to the mirrors address?
2023-12-07 11:23:53 <mps> AljazS: OT, I speak slovenian
2023-12-07 11:24:13 <AljazS> ikke: No I have not. I was not aware that's an option.
2023-12-07 11:24:18 <AljazS> mps: weila, lepa :)
2023-12-07 11:24:53 <ikke> AljazS: ah, that would be good to do then
2023-12-07 11:24:54 <mps> my lastname ends with 'ć'
2023-12-07 11:25:19 <AljazS> mps: So a Balcan brother? :P
2023-12-07 11:25:29 <AljazS> ikke: Will do.
2023-12-07 11:25:36 <mps> AljazS: yes
2023-12-07 11:26:29 <AljazS> Aight. Well thank you for the info. I'll send the email to the mirrors addr. Have a good one guys o/
2023-12-07 21:32:26 <ncopa> I would love to have cdn.alpinelinux.org/edge/main
2023-12-07 21:33:37 <ncopa> and cdn.alpinelinux.org/v3.19/main
2023-12-07 21:36:57 <ncopa> need to configure fastly now that the cname is in place. and the certs
2023-12-07 21:37:05 <ikke> yes
2023-12-07 21:37:50 <ncopa> i wonder if we can use something.alpinelinux.org/alpine as backend
2023-12-07 21:38:25 <ikke> We use t1.alpinelinux.org
2023-12-07 21:38:59 <ncopa> can we use t1.alpinelinux.org/alpine? the subdir
2023-12-07 21:39:27 <ikke> For what?
2023-12-07 21:39:28 <ncopa> and can we tell fastly that it is an alias, so we reuse the dl-cdn cache?
2023-12-07 21:39:32 <ncopa> as backend
2023-12-07 21:39:49 <ikke> oh rather than t1.alpinelinux.org
2023-12-07 21:40:00 <ncopa> so we end up with cdn.alpinelinux.org/edge/main instead of cdn.alpinelinux.org/alpine/edge/main
2023-12-07 21:40:17 <ncopa> i want get rid of the /alpine/ path element in between
2023-12-07 21:40:22 <ikke> yeah, understood
2023-12-07 21:40:33 <ikke> Not sure, I'm not too familiar with fastly
2023-12-07 21:40:53 <ncopa> or we could move it for all mirrors
2023-12-07 21:41:00 <ncopa> and add symlink alpine -> .
2023-12-07 21:41:18 <ikke> Wouldn't that cause a lot of bandwidth
2023-12-07 21:41:19 <ncopa> would be even better, but im afraid of all mirrors resyncing everything at the same time
2023-12-07 21:41:24 <ncopa> exactly
2023-12-07 21:41:24 <ikke> as rsync would not see it as a move
2023-12-07 21:41:44 <ncopa> if we can move it i think we should, but i dont think we can
2023-12-07 21:42:11 <ncopa> so i thought that we point cdn.a.o to t1.a.o/alpine as backend, and get rid of the element in between
2023-12-07 21:42:16 <ncopa> i believe it should be possible
2023-12-07 21:43:34 <ncopa> i suppose other way to get rid of the /alpine/ element would be to add symlinks at the top level dir it would be to 
2023-12-07 21:44:06 <ncopa> s/ it would be to //
2023-12-07 21:46:28 <ncopa> but its annoying to add a symlink for every new release
2023-12-07 21:46:46 <ncopa> eg v3.19 -> alpine/v3.19
2023-12-07 21:58:25 <ikke> Perhaps we need something like https://developer.fastly.com/solutions/examples/rewrite-url-path/
2023-12-08 00:56:00 <iXNyNe> ncopa ikke i might have an interesting idea re: URLs and redirects. have you heard of https://scarf.sh/ ? LSIO uses scarf to have a vanity URL on our docker images and some metrics on usage (docker registries are a bit inconsistent with that info). anyway they have the ability to do file URLs that works for package management
2023-12-08 00:59:53 <iXNyNe> btw, congrats on 3.19 release. lsio is already prepping our base image 😄
2023-12-10 18:54:19 <mps> https://lwn.net/Articles/954285/   Ext4 data corruption in stable kernels
2023-12-10 18:55:02 <mps> looks like it related to 6.1.64 and 6.1.65 kernels
2023-12-10 18:55:37 <minimal> has been mentioned in both alpine-devel and alpine-linux in the past day
2023-12-10 18:55:57 <mps> ah ok, I'm not there
2023-12-11 11:31:57 <ncopa> I filed this: https://gitlab.alpinelinux.org/alpine/infra/mirrors/-/issues/730 not sure mirrors is the right project
2023-12-11 11:33:17 <ikke> https://gitlab.alpinelinux.org/alpine/infra/infra/
2023-12-11 11:33:41 <ncopa> moved. thanks!
2023-12-11 12:14:03 <ncopa> FYI, dmvpn works on alpine 3.19 
2023-12-11 12:37:56 <clandmeter> do you mean replace dl-cdn.a.o?
2023-12-11 13:51:42 <ncopa> yes
2023-12-11 13:51:55 <ncopa> but we need keep dl-cdn.a.o for backwards compat
2023-12-11 14:04:33 <clandmeter> ncopa: ok, i think its a typo on the ticket
2023-12-11 21:15:44 <clandmeter> ikke: regarding the audio on meet.a.o, i think its related to cpu load.
2023-12-11 21:16:21 <clandmeter> i notice my cpu fan is going up when my audio starts to break down
2023-12-11 21:16:31 <ikke> hmm, ok
2023-12-11 21:16:47 <clandmeter> cause i think in the beginning you can understand ok right?
2023-12-11 21:17:05 <ikke> I believe so
2023-12-11 21:17:29 <clandmeter> so a macbook pro from 2017 is not fast enough :|
2023-12-11 21:18:07 <clandmeter> im going to order a new macbook pro, so that will probably solve the issue. 
2023-12-11 21:18:19 <clandmeter> but its not the reason for the purchase :D
2023-12-11 21:18:36 <ikke> haha
2023-12-11 21:18:44 <ikke> I don't have such a beestly machine
2023-12-11 21:19:25 <clandmeter> if that new machine wont work, then ill rm -rf meet.a.o :)
2023-12-11 21:19:37 <ikke> :D]
2023-12-11 21:19:55 <ikke> clandmeter: Should we try to upgrade it?
2023-12-11 21:20:16 <clandmeter> sure, but i dont have time now. maybe later this week
2023-12-11 21:20:35 <ikke> yeah, there's no rush
2023-12-11 21:20:54 <ikke> But it's a bit out of date
2023-12-11 21:21:01 <clandmeter> right
2023-12-11 21:21:12 <clandmeter> how are video meetings on matrix?
2023-12-11 21:21:48 <clandmeter> or does it use jiri
2023-12-11 21:21:52 <clandmeter> jitsi
2023-12-11 21:22:14 <ikke> I have no clue
2023-12-11 21:22:23 <clandmeter> i think having jitsi could be a better alternative
2023-12-11 21:22:34 <clandmeter> atleast "an" alternative
2023-12-11 21:43:35 <mps> clandmeter: I have macbook m1pro but also it could sometimes run fans
2023-12-12 07:26:50 <Saijin_Naib[m]> <clandmeter> "how are video meetings on matrix..." <- They are indeed embedded Jitsi
2023-12-12 14:42:55 <Kladky> Matrix is moving to their own native VOIP solution for group calls, the biggest implementation being Element Call: call.element.io
2023-12-12 19:09:44 <Saijin_Naib[m]> Sweet! Is that part of the upcoming 2.0 protocol and ElementX work?
2023-12-13 11:31:10 <ikke> large spamwave of users, luckily all with the same mail domain
2023-12-13 20:23:46 <mps> we shouldn't upgrade linux-edge for 3.18 anymore I think?
2023-12-13 20:39:49 <ikke> It is EOL
2023-12-13 20:42:06 <mps> nice, less work
2023-12-14 08:46:07 <clandmeter> ikke: does gitlab no longer have preview button when submitting an issue?
2023-12-14 08:46:13 <clandmeter> or maybe it never did
2023-12-14 08:48:20 <clandmeter> oh it does, but its in the editor itself
2023-12-14 08:48:44 <clandmeter> im so used that such button is below the textfield
2023-12-15 06:44:31 <clandmeter> ikke: have you used !reset in compose already?
2023-12-15 07:08:29 <ikke> clandmeter: no
2023-12-15 07:09:12 <clandmeter> i just tried but it did not completely work on the .18 version we have
2023-12-15 07:09:20 <clandmeter> so i copied a binary from upstream
2023-12-15 07:09:27 <clandmeter> on deu1-dev1
2023-12-15 07:18:57 <ikke> What does it do?
2023-12-15 07:20:57 <ikke> https://docs.docker.com/compose/compose-file/13-merge/#reset-value
2023-12-15 07:23:50 <clandmeter> ie reset ports:
2023-12-15 07:24:04 <clandmeter> for override compose files
2023-12-15 07:24:41 <clandmeter> i have a copy of jitsi running on that host
2023-12-15 07:24:50 <clandmeter> not sure its any better compared to meet.a.o
2023-12-15 10:53:39 <ikke> clandmeter: ncopa do we want to move the arm  builders back to che?
2023-12-15 12:03:19 <ncopa> I don't mind as long as we can trust that it is up and running
2023-12-15 12:05:50 <ikke> There has been some downtime lately, but all recovered within a couple of hours
2023-12-15 16:40:05 <ikke> rebooting deu7-dev1 (wg bridge) for updates
2023-12-15 16:48:55 <ikke> deu7-dev1 upgraded to 3.19
2023-12-15 20:28:46 <mps> anyone with rust/cargo experience could help with !57363
2023-12-15 20:29:38 <mps> it need to pull some crates from the net and I don't know how to do this with aports
2023-12-15 20:36:54 <mps> it builds fine on my local machine but not on CI
2023-12-16 01:30:51 <celie> I think the `cargo fetch` options (--locked instead of --offline) used in newapkbuild (https://gitlab.alpinelinux.org/alpine/abuild/-/blob/master/newapkbuild.in#L35) should work
2023-12-16 07:43:34 <mps> celie: thanks. but I tried this from beginning and got error 'error: the lock file /home/mps/aports/testing/speakersafetyd/src/speakersafetyd-0.1.7/Cargo.lock needs to be updated but --locked was passed to prevent this'
2023-12-16 07:44:19 <mps> and 'If you want to try to generate the lock file without accessing the network, remove the --locked flag and use --offline instead.' because this I used --offline
2023-12-16 08:05:52 <celie> I think that's because the version numbers in Cargo.lock and Cargo.toml don't match
2023-12-16 08:09:33 <celie> I just tried it, running `cargo update -p speakersafetyd` updates the version number in Cargo.lock, and that allows `cargo fetch --locked` to complete
2023-12-16 08:11:22 <celie> So, adding `cargo update -p speakersafetyd` before `cargo fetch --target="$CTARGET" --locked` in prepare() should work
2023-12-16 08:12:51 <celie> Usually, for other aports, we make a patch instead of running `cargo update` in APKBUILD, but in this case, the only thing it updated was the version number, and if upstream forgets to update it again for the next version, you'll have to update the patch too if you used that
2023-12-16 08:17:42 <mps> celie: aha, but it works fine on my local machine with --offline. Though I don't understand 'fine points' of cargo
2023-12-16 08:23:05 <celie> I don't really understand cargo either, it's just that i've seen the error message before, and that's how i solved it then
2023-12-16 08:23:52 <celie> As for it building fine on your local machine, did you run `cargo build` or `cargo fetch` without --locked or --frozen before trying --offline?
2023-12-16 08:24:29 <mps> now I got 'error: failed to run custom build command for `alsa-sys v0.3.1`'
2023-12-16 08:24:32 <celie> If you did, i think that may have updated the crates.io index or already cached the crates, so cargo didn't need to use the network anymore and --offline works
2023-12-16 08:25:17 <mps> I run on local machine 'abuild -r'
2023-12-16 08:25:55 <celie> Ok, not really sure why then
2023-12-16 08:27:05 <mps> but your note about index makes sense because earlier I run make and maybe this is cached somewhere
2023-12-16 08:27:54 <mps> (all this rust ecosystem is unfriendly to distros)
2023-12-16 08:28:10 <celie> Yea, i feel that way too sometimes
2023-12-16 08:28:33 <celie> Anyway, alsa-sys probably needs alsa-lib-dev for its "custom build command"
2023-12-16 08:29:18 <mps> yes, also I think so, even tried it earlier
2023-12-16 08:29:53 <mps> and with it 'abuild -r' passed
2023-12-16 08:31:46 <mps> also 'abuild rootbld'
2023-12-16 08:32:04 <mps> looks like we got it. thank you again
2023-12-16 08:32:22 <celie> You're welcome
2023-12-16 08:35:59 <mps> and it passed in CI \o/
2023-12-16 08:37:06 <mps> now alpine have speakers on apple silicon machines
2023-12-16 13:39:48 <pj> it's not unfriendly, you just have to RTFM
2023-12-16 16:17:38 <clandmeter> ikke: i think we could wait until we move one server to the other place
2023-12-19 09:54:54 <ncopa> i have been thinking that we should probably modernize the alpinelinux.org website
2023-12-19 09:55:08 <ncopa> been thinking of switching to hugo
2023-12-19 12:26:24 <clandmeter> elaborate please 
2023-12-19 12:26:29 <clandmeter> whats wrong with the current one?
2023-12-19 12:29:29 <ncopa> i'd like avoid maintain the full webpage generator, and re-use something someone else maintains
2023-12-19 13:52:16 <clandmeter> it sounds like it will be lots of work to convert things currently made for it? like the cloud integration also needs to be redone? sorry i dont know all the details, just something that came to mind. 
2023-12-19 14:02:00 <ncopa> yeah. that is holding me back from go ahaed and do it: lots of work
2023-12-19 14:22:38 <mps> I started long ago to convert one site from custom one to hugo but still didn't finished
2023-12-19 14:23:09 <mps> looked like it will be easy but reality is different
2023-12-19 18:15:12 <ptrc> ikke: https://gitlab.alpinelinux.org/-/snippets/265#note_363948
2023-12-19 20:08:13 <ikke> ptrc: thanks, removed
2023-12-20 00:11:15 <pj> is the host with my container dead?
2023-12-20 00:11:39 <pj> I've been having issues on and off trying to connect, when it works on just some days
2023-12-20 11:51:23 <ikke> pj: it's because of corerad crashing, trying to figure out why
2023-12-20 13:48:10 <ikke> 🤔
2023-12-20 13:48:53 <pj> ikke: thanks, I also did test a bit on my side, google tells me I'm not ipv6, ipv6 test tells me I'm, weird stuff
2023-12-20 16:54:12 <ncopa> i wonder if we should ask for more volunteers to help remove spam users
2023-12-20 16:54:34 <ikke> problem is that it would require administrator permissions
2023-12-20 16:57:58 <ncopa> https://gitlab.alpinelinux.org/Celeste/aports/-/jobs/1220164 its running on my board
2023-12-20 17:00:44 <ncopa> hi mick_ibm!
2023-12-20 17:01:04 <ncopa> I wonder if you could help us with some ppc64le related issues
2023-12-20 17:01:32 <ncopa> mick_ibm: https://sourceware.org/bugzilla/show_bug.cgi?id=30824
2023-12-20 17:02:39 <ncopa> mick_ibm: https://gitlab.alpinelinux.org/alpine/aports/-/jobs/1218465
2023-12-20 17:03:29 <ncopa> and this: https://gitlab.alpinelinux.org/alpine/aports/-/issues/15478
2023-12-20 17:09:10 <mick_ibm> oh boy, ok. ill send this to our kernel team because im going to be offline until early janurary traveling. so first link seems to be with gnu binutils (ld), third link is in bpf_legacy.c for 'Elf64_Relr'... but the second link, is the error with freerdp is is that just how the error is being reported?
2023-12-20 17:12:53 <ncopa> yes, the freerdp error is the bug in binutils
2023-12-20 17:13:08 <ncopa> i bumped into it again a couple of days ago
2023-12-20 17:13:45 <ncopa> mick_ibm: I wonder if you or someone else from IBM could register to gitlab.alpinelinux.org so we can assign issues there?
2023-12-20 17:13:56 <ncopa> so we have someone we can assign ppc64le related issues
2023-12-20 17:21:11 <mick_ibm> yeaa i will register there 
2023-12-20 17:22:28 <ncopa> Thanks!
2023-12-20 18:33:11 <ikke> ncopa: someone recommended me https://www.stopforumspam.com/ the other day
2023-12-21 09:42:46 <ncopa> https://security.alpinelinux.org/vuln/CVE-2023-40019
2023-12-21 09:42:58 <ncopa> yesterday edge and 3.19-stable was listed
2023-12-21 09:43:22 <ncopa> i fixed those, and expected to find them listed as fixed. instead they disappeared?
2023-12-21 09:43:33 <ncopa> 5745c542925397f583bdec3d859e87691f4d52ac
2023-12-21 10:39:50 <ncopa> it took 214 mins to buidl qemu on the starfive board: https://gitlab.alpinelinux.org/alpine/aports/-/jobs/1220330
2023-12-21 10:40:39 <ncopa> 15 mins on ppc64le (which I think is the slowest runner)
2023-12-21 10:40:45 <ncopa> I'd say 214 mins is useless
2023-12-21 11:17:17 <ptrc> perhaps we should just crowdfund one of those: https://www.crowdsupply.com/milk-v/milk-v-pioneer :p
2023-12-21 11:21:45 <ncopa> yeah
2023-12-21 11:27:44 <mps> ncopa: thank for fixing qemu MR and sorry for mess I made
2023-12-21 11:27:53 <mps> thanks*
2023-12-21 11:30:39 <ncopa> np
2023-12-21 11:31:05 <ncopa> i wonder if it would be enough with one milk-v pioneer
2023-12-21 11:31:13 <ncopa> we'd need for CI as well
2023-12-21 11:32:02 <mps> well, CI could be qemu-user on some of machines. slow yes but usable
2023-12-21 11:32:36 <ncopa> we currently don't run tests for the qemu-user riscv
2023-12-21 11:32:50 <ncopa> we'd like to run the tests
2023-12-21 11:33:15 <mps> ofc, but we have 'live' with this what we have
2023-12-21 11:33:24 <ncopa> i suppose the question is: will be be able to do 3.20-stable riscv64 with one of those machines?
2023-12-21 11:33:54 <mps> I guess milk-v could do this
2023-12-21 11:34:20 <mps> but ofc, didn't checked because we don't have it
2023-12-21 11:34:48 <ncopa> will it be capable of building edge+4 stable bracnhes (+ one extra branch when there is a new release) + CI?
2023-12-21 11:34:50 <mps> and I wonder why your qemu build takes so much
2023-12-21 11:35:01 <ncopa> on starfive2?
2023-12-21 11:35:05 <ncopa> because it is slow
2023-12-21 11:35:05 <mps> do you have NVME
2023-12-21 11:35:09 <ncopa> yes
2023-12-21 11:35:37 <mps> in my test build kernel took about hour few months ago
2023-12-21 11:35:51 <ncopa> yeah, that sounds plausible
2023-12-21 11:35:54 <mps> one hour*
2023-12-21 11:36:32 <ncopa> i had less than that when building in qemu-user on my i9 desktop
2023-12-21 11:36:32 <mps> though not sure if big packages could be built at all in visionfive v2
2023-12-21 11:36:54 <ncopa> i'd say no
2023-12-21 11:36:58 <ncopa> its too slow
2023-12-21 11:37:47 <mps> I hope with kernel 6.7 it would be better
2023-12-21 11:38:48 <mps> also qemu 8.2.0 have some improvements for riscv
2023-12-21 11:39:27 <mps> but don't have free time to test (want to finish with M1)
2023-12-21 11:40:41 <mps> last time number of people asking to help with alpine on private 'channels' increased
2023-12-21 11:45:15 <ncopa> https://forum.sophgo.com/t/risc-v-public-test-platform-released-7-zip-test/263 shows that it will be on pair with 6 cores of Ryzen 3950X
2023-12-21 11:45:23 <ncopa> it will still be slow for our use case
2023-12-21 11:49:43 <mps> for speed on riscv it is good/important to use latest kernels and toolchains because new instructions and features are constantly added
2023-12-22 08:55:12 <mps> we are using postfix? http://www.postfix.org/smtp-smuggling.html
2023-12-22 08:55:49 <ncopa> I believe we are
2023-12-22 08:57:01 <mps> for now fix is 'smtpd_data_restrictions = reject_unauth_pipelining'
2023-12-22 08:57:18 <mps> till the updated versions are released
2023-12-22 09:04:35 <ikke> Yes, will fix it today 
2023-12-22 09:56:05 <ncopa> seems like there are issues with lxc 5.0.3 in alpine 3.19 due to the change to cgroup2
2023-12-23 16:12:44 <ikke> ..
2023-12-23 18:41:43 <clandmeter> something looks unstable
2023-12-23 18:41:48 <ikke> ahuh
2023-12-23 18:42:19 <ikke> my ssh sessions are still connected
2023-12-23 21:58:03 <mps> also my ssh works
2023-12-23 21:59:16 <mps> and I don't like much openwrt, I use it when don't have choice
2023-12-24 08:27:16 <ikke> mps: that's different
2023-12-24 12:14:13 <mps> we don't have libexecinfo pkg? I thought we had it
2023-12-24 12:18:29 <ikke> 50795a14dee639ce2dcc836e2b2baca9bad4a1b1
2023-12-24 12:33:58 <mps> ikke: thanks
2023-12-24 12:34:07 <mps> what we can use instead
2023-12-24 12:34:21 <ikke> For what?
2023-12-24 12:34:59 <mps> some pkgs wants it
2023-12-24 12:36:03 <mps> it is used for backtrace in some pkgs, especially built in debug mode
2023-12-24 12:36:25 <mps> yes, it is glibc-ism, I know
2023-12-24 12:37:27 <ikke> It does not work properly with musl. All the packages that used it in aports just remove it
2023-12-24 12:40:44 <mps> ikke: right, just read issue report
2023-12-24 12:41:11 <mps> will tell upstream author of dosbox-staging to remove it
2023-12-24 12:41:39 <mps> thanks for help
2023-12-24 13:56:54 <ikke> pj: fixed an issue, your container should have an ipv6 address again
2023-12-24 17:02:07 <pj> ikke: thanks a lot, you're the best
2023-12-29 18:26:00 <ncopa> I have powered off the Ncopa-riscv64-runner. I don’t think it is usable for ci 
2023-12-29 18:37:29 <ikke> Alright 
2023-12-29 18:56:47 <ikke> I paused it
2023-12-31 15:46:42 <clandmeter> ikke: maybe you can add our other rv64 board as a runner, or maybe it is already equipped? 
2023-12-31 15:46:56 <clandmeter> or i can try it when i find the time
2023-12-31 15:49:52 <ikke> What other board? The vf2?
2023-12-31 15:53:11 <clandmeter> no
2023-12-31 15:53:14 <clandmeter> the other one
2023-12-31 15:53:19 <clandmeter> its on our infra already
2023-12-31 15:53:28 <clandmeter> faster board
2023-12-31 15:54:40 <clandmeter> unmatched one
2023-12-31 16:02:58 <clandmeter> are there any numbers how fast it can compile a kernel?
2023-12-31 16:22:21 <ikke> Oh, that one, forgot about it 
2023-12-31 16:22:29 <ikke> Is it still connected?