21:00:01 #startmeeting EPEL (2022-03-09) 21:00:01 Meeting started Wed Mar 9 21:00:01 2022 UTC. 21:00:01 This meeting is logged and archived in a public location. 21:00:01 The chair is carlwgeorge. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions. 21:00:01 Useful Commands: #action #agreed #halp #info #idea #link #topic. 21:00:01 The meeting name has been set to 'epel_(2022-03-09)' 21:00:07 #meetingname epel 21:00:07 The meeting name has been set to 'epel' 21:00:11 #chair carlwgeorge 21:00:11 Current chairs: carlwgeorge 21:00:53 hey guys 21:00:54 .hi 21:00:55 pgreco: pgreco 'Pablo Sebastian Greco' 21:01:14 #chair pgreco 21:01:14 Current chairs: carlwgeorge pgreco 21:01:44 #topic aloha 21:02:31 .hi 21:02:32 salimma: salimma 'Michel Alexandre Salim' 21:02:44 #chair salimma 21:02:44 Current chairs: carlwgeorge pgreco salimma 21:02:51 hi salimma 21:03:05 .hi 21:03:06 dherrera: dherrera 'None' 21:03:28 morning 21:04:00 #chair nirik 21:04:00 Current chairs: carlwgeorge nirik pgreco salimma 21:04:05 carlwgeorge, I have something for open floor if there's enough time 21:04:46 .hi 21:04:48 rcallicotte: rcallicotte 'Robby Callicotte' 21:04:51 Davide and smooge send their regrets 21:05:14 pgreco: sure thing 21:05:18 (this is the one meeting a month that clashes with CentOS Board meeting for Davide, and smooge is shedding blood somewhere) 21:05:26 :p 21:06:42 guess that's long enough for hellos 21:06:47 #topic EPEL Issues 21:06:50 :) 21:07:05 #link https://pagure.io/epel/issues?tags=meeting&status=Open 21:07:30 we've only got one tagged, and we agreed to push it back to march 23rd 21:08:01 yep 21:08:05 we also had a new issue today that by our docs we should wait 1 week to allow voting 21:08:08 #link https://pagure.io/epel/issue/164 21:08:28 now tagged as meeting for next week 21:08:29 let's tag that as meeting now for next time 21:08:32 ah 21:08:45 but please do add your individual vote in there if you're a member of the epel packagers sig 21:08:52 its Adam! 21:09:12 * carlwgeorge waves at maxamillion 21:09:48 any other issues from the tracker we need to talk about before moving on? 21:10:33 smooge wanted to bring up the packages he branched for fedpkg, but that can go to EPEL8 later 21:10:55 * carlwgeorge nods 21:11:36 #topic Old Business 21:12:44 * nirik always feels seen with this one. 21:12:56 hehe 21:13:20 .hello2 21:13:23 maxamillion: maxamillion 'Adam Miller' 21:13:40 is anybody else having trouble logging into pagure? 21:13:45 Strange state: failure 21:13:49 * rcallicotte waves hello 21:14:19 .hello 21:14:19 themayor: (hello ) -- Alias for "hellomynameis $1". 21:14:30 pgreco: working fine for me, just logged in again in a private tab to check 21:14:30 pgreco: I had a similar issue recently where the redirect failed, but I'm still logged in on most devices so didn't check further 21:15:05 if we've got nothing for old business we can move on to... 21:15:09 yeah, I'm never logged, I opened it in a private window and after the password just goes to that mode.. 21:15:16 #topic EPEL-7 21:15:28 are you entering your username (not password) and do you have a otp enrolled? 21:15:36 carlwgeorge: you emailed maxamillion@redhat.com, which is ironically an existing thing because I accidentally put that on a bunch of package changelogs that shipped in OpenShift Enterprise 1.0 forever ago 21:16:12 i just assumed you had that alias and it came up in the gmail search so i went with it :D 21:16:42 nirik, full email, no otp 21:16:55 #info epel7 will be retired on 2024-06-30 in line with the end of the rhel7 maintenance 2 phase 21:17:13 carlwgeorge: +1 21:17:16 good times :) 21:17:52 nirik: username works, email fails, both on pagure.io and src.fedoraproject.org.... 21:17:54 ah, EPEL7 21:17:55 one sec 21:18:26 pgreco: yes, email worked on the old account system by accident. It's not on the new one. 21:18:26 I know we'll do the security review on March 23, but quick recap, 3 Chromium CVEs closed, but we gained one for python-twisted 21:18:46 ack 21:19:01 (we are fixing it, but it's proving to be a long painfull road. ) 21:19:49 sounds like progress 21:20:55 anything else for epel7? 21:22:12 #topic EPEL-8 21:22:38 #info epel8-next will be retired on 2024-05-31 in line with the c8s eol (end of rhel8 full support phase) 21:22:41 sort of related, I just discovered COPR doesn't support epel-next 21:23:21 not with a checkbox, but i don't think it's necessary 21:23:48 coprs should target epel, and on the off chance they need to build with epel-next they can add it as a extra repo 21:24:26 I want to bring something up 21:24:39 shoot 21:24:58 i've been working on checking the packages that don't install (from this list https://tdawson.fedorapeople.org/epel/willit/epel8/status-wont-install.html) 21:25:33 and found drbd-pacemaker doesn't work because it is missing the pacemaker dependency 21:26:18 investigating, I saw that pacemaker exists on EPEL7 because it's built against the HighAvailability repo 21:26:38 but that repo is not available neither in EPEL8 or EPEL9 by policy 21:27:52 my question, how should this get resolved? 21:28:01 i can think of a few possibilities 21:28:19 1) change the dep to a recommends instead of a requires 21:29:04 "coprs should target epel, and on..." <- ah, my use case is for packit 21:29:05 2) someone build pacemaker in epel directly (a bit odd because it is readily available in the highavailability or resilientstorage repos) 21:29:20 we have some packages that doesn't support epel8 yet at the moment (due to updated deps) 21:29:39 3) retire drbd-pacemaker 21:29:46 so... configuring extra repos in packit is probably tricky 21:29:48 we used to add/use the HA repo for epel6... 21:30:03 the real question for the committee is are epel packages allowed to require ha or rs packages 21:30:31 hum. 21:30:47 salimma: copr allows adding extra repos, so if packit doesn't expose that then it sounds like a packit feature request or bug 21:31:07 carlwgeorge: ack 21:31:12 re: HA or RS.. hmm 21:31:50 I suppose we could add them, but I really would like to avoid adding a bunch of channels... 21:31:58 if we were to add ha and/or rs to the list of repos epel builds against, then that would make some epel packages in violation of policy, and lead to several unshipped package requests 21:32:19 i don't remember off hand which ones they are, but i believe i've seen them before 21:32:39 what's the use case of packages in EPEL that are the same as in HA? newer versions? 21:33:08 subpackages that aren't shipped in ha/rs 21:33:26 I've seen this in infra repo too 21:33:27 by policy now it's allowed to overlap, because ha/rs aren't considered the "base" 21:34:37 fedora has a policy that all packages deps must be satisfied within the official repos https://docs.fedoraproject.org/en-US/packaging-guidelines/#_package_dependencies 21:35:15 the strictest translation of that to epel would be that epel packages can't depend on ha/rs packages, but that's sure to upset people 21:35:24 ah, so if we add this we'll have to rename them to $foo-epel I guess 21:35:49 and make sure the versions match, etc. ... what's going to be annnoying is that now people have to flip /3/ repos to enabled 21:36:05 i think this is an interesting and complex issue, but i don't think we'll come to a resolution on it during the time of this meeting 21:36:15 we perhaps should push DNF devs to have a way to override repo configs sooner rather than later 21:36:24 yeah, we should move on for now 21:36:41 dherrera: thanks for bringing it up, can you also start a thread on #epel-devel to summarize the situation and gather more feedback? 21:37:00 We really need to avoid adding all rhel channels... that way lies madness. ;) 21:37:01 salimma: such a dnf change would probably be unlikely to be backported to el8 21:37:30 👍️ 21:37:49 i'll reply on the thread when it exists, but i'm leaning towards a policy of recommends, even if it should be requires otherwise, i.e. if fedora requires, if epel recommends 21:38:06 any other epel8 stuff before we move on? 21:38:30 if we have an exception, it should be very clear that it is related to additional rhel repos 21:38:40 and not that it can depend on anything external 21:38:43 carlwgeorge: agreed, just that we have time to get it in so it will work for el10 and above 21:39:03 pgreco: agreed 21:39:33 I'm not opposed to just added HA... it's pretty common I suspect... 21:39:37 adding 21:40:16 what's the rule on rhel to have access to HA? 21:41:00 i'm not even sure if it has an additional cost, or if it's just a repo you can enable at will 21:41:48 not sure. I think it varied in the past... 21:41:48 alright lets move on so we have time for the open floor 21:41:54 #topic EPEL-9 21:42:13 if can be enabled at will, I don't see an issue adding it to the repos for epel builds 21:44:04 this past week i added a could of useful python libs (outcome, async-generator, trio, h11, socksio, anyio) to epel9, they're in testing if anyone is interested in giving karma 21:45:16 * Eighth_Doctor waves 21:45:33 #chair Eighth_Doctor 21:45:33 Current chairs: Eighth_Doctor carlwgeorge nirik pgreco salimma 21:46:22 also mildly interesting, i did a twitter poll about epel9 desktops, and was surprised to see xfce was more desired than mate or cinnamon (in my small sample size) https://twitter.com/carlwgeorge/status/1498394802366332932 21:46:38 I finally put out the python-flit update, and there's also an update (tmuxp) that includes python-poetry-core 21:46:43 really? 21:46:53 so hopefully we have most of the Python build systems out there soon 21:47:14 that'd be nice 21:47:15 takeaway, if anyone wants to help bring xfce to epel9, reach out to the maintainers, as it would make quite a few people happy 21:47:21 what about poetry itself? 21:47:36 Conan Kudo: if you have something that needs it, I can take a look at it 21:47:53 I only got to poetry-core via libtmux which needs it 21:48:09 nonamedotc was going to do it... I think he even has a copr... not sure his timeline tho 21:48:35 Michel Alexandre Salim 🎩: I do have things that need it 21:48:39 hence the ask 21:49:01 some of the stuff I maintain needs it 21:49:19 probably reach out to nonamedotc directly first then 21:49:41 anything else for epel9 before we open the floor and let pgreco bring up his new thing? 21:51:05 #topic General Issues / Open Floor 21:51:09 I'm picking up python-bitcoinlib in fedora because it is ftbfs, and tracked down all its failures to something related to openssl 3.0 21:51:35 salimma: do you know if the "meta" guys have seen any crashes in the openssl 3 backport? 21:52:58 pgreco: love the quote marks haha 21:53:08 hehehe 21:53:19 went for the double meaning ;) 21:53:21 hmm, I have not heard of issues related to the backport but let me check 21:53:43 I have a simple reproducer with python in cs9/fedora 36 21:53:44 IIRC we backported it so we can use it in systemd 21:54:01 same version of python doesn't crash in f35 21:54:57 can you file a bug? 21:55:18 https://paste.centos.org/view/30b42dd1 21:55:40 if it's fixed in fedora I can rebase and issue an epel update too 21:55:55 I haven't had time to file it yet, but running that script with python linked against openssl3.0 turns into a core dump 21:56:01 it's not fixed in fedora 21:56:20 I'll file an issue this week 21:57:17 I've also found some other inconsistencies in it that I need to track down 21:57:23 it's a shame openssl 3 stuff didn't have time to bake in rawhide before going into c9s, but i know that the timelines (including upstream) just didn't work out that way 21:57:29 cs9's python is not linked against openssl3, so it's more about loading that library dynamically I guess 21:57:30 but wondering if you had any more insight 21:57:50 true 21:58:14 I'll check and let you know if I hear 21:58:20 thanks 21:58:29 I'll add you to the bug when I file it 21:58:41 salimma: python3-libs-3.9.9-4.el9 links against libssl.so.3 21:59:09 wait what? python in c9s isn't linked to openssl3? 21:59:12 how is that possible? 21:59:18 Eighth_Doctor: oh duh 21:59:32 sorry, brain fade, I'm thinking of epel8 which has openssl3 backported 22:00:26 well we're out of time this week, anything else before we wrap up? 22:02:34 I guess not 22:03:12 thanks everyone for coming, see yall next week 22:03:18 #endmeeting