21:00:41 <tdawson> #startmeeting EPEL (2022-12-21)
21:00:41 <zodbot> Meeting started Wed Dec 21 21:00:41 2022 UTC.
21:00:41 <zodbot> This meeting is logged and archived in a public location.
21:00:41 <zodbot> The chair is tdawson. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions.
21:00:41 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
21:00:41 <zodbot> The meeting name has been set to 'epel_(2022-12-21)'
21:00:43 <tdawson> #meetingname epel
21:00:43 <zodbot> The meeting name has been set to 'epel'
21:00:44 <tdawson> #chair nirik tdawson pgreco carlwgeorge salimma dcavalca dherrera gotmax[m] smooge
21:00:45 <zodbot> Current chairs: carlwgeorge dcavalca dherrera gotmax[m] nirik pgreco salimma smooge tdawson
21:00:46 <tdawson> #topic aloha
21:00:48 <salimma> .hi
21:00:49 <zodbot> salimma: salimma 'Michel Alexandre Salim' <michel@michel-slm.name>
21:00:50 * smooge is here to hand over the nuclear football back to President Dawson.
21:00:54 <dherrera> .hi
21:00:56 <zodbot> dherrera: dherrera 'Diego Herrera' <dherrera@redhat.com>
21:01:01 <pgreco> .hi
21:01:02 <zodbot> pgreco: pgreco 'Pablo Sebastian Greco' <pablo@fliagreco.com.ar>
21:01:12 <carlwgeorge> .hi
21:01:13 <zodbot> carlwgeorge: carlwgeorge 'Carl George' <carl@redhat.com>
21:01:21 <tdawson> Hi salimma, smoog, pgreco and carlwgeorge
21:02:03 * tdawson takes the nuclear football and looks anxiously at it.
21:03:01 <pgreco> tdawson: I was at CERN today and your name came up, not gonna say in what context :D
21:03:02 * salimma thinks it's a trap
21:03:21 <carlwgeorge> itsatrap.gif
21:03:28 <salimma> thanks for spending your late evening with us pgreco
21:03:50 <carlwgeorge> same to you salimma, i hear you are in a similar timezone
21:04:45 <tdawson> That's right ... this is really late for Europe.
21:05:11 <tdawson> #topic End Of Life (EOL)
21:05:11 <smooge> Well it was sundown hours ago
21:05:12 <tdawson> RHEL 7 will go EOL on 2024-06-30
21:05:14 <tdawson> CentOS Stream 8 goes EOL in 2024-05-31
21:05:14 <pgreco> only if you adapt to the timezone...
21:05:15 <tdawson> CentOS Stream 9 goes EOL in 2027-05-31
21:05:22 <salimma> oh no, I came back on Monday
21:05:24 <salimma> so I'm just jet lagged
21:05:26 <tdawson> That's it folks ... thanks for comming ... talk to you next week ...
21:05:45 <tdawson> Oh wait ... I think there is a couple of things on the agenda ...
21:05:57 * smooge sits down again
21:06:18 <tdawson> #topic EPEL Issues https://pagure.io/epel/issues
21:06:19 <tdawson> https://pagure.io/epel/issues?tags=meeting&status=Open
21:06:37 <carlwgeorge> i think the oldest one (#212) is settled and can be untagged
21:07:07 <tdawson> Thanks ... I was just about to ask about that one.
21:07:35 <carlwgeorge> not sure if we normally leave those open until the underlying thing is resolved or just close after voting
21:07:52 <carlwgeorge> #link https://pagure.io/epel/issue/212
21:08:35 <rsc> .hello robert
21:08:36 <zodbot> rsc: robert 'Robert Scheck' <redhat@linuxnetz.de>
21:08:45 <tdawson> Hello rsc
21:09:18 <tdawson> I *think* we close it at this point.
21:09:35 <pgreco> I think so, the decision is made, now it's up to the maintainer
21:09:39 <smooge> .epel 212
21:09:40 <zodbot> smooge: Issue #212: 'novnc' major version/breaking update in EPEL7 - epel - Pagure.io - https://pagure.io/epel/issue/212
21:09:56 <tdawson> They've properly annoucned it, we've done the vote.  It might take some time before it get's built and released.
21:10:01 <smooge> close it
21:11:24 <tdawson> Now for the second easiest ticket ...
21:11:32 <tdawson> .epel 215
21:11:33 <zodbot> tdawson: Issue #215: remove singularity obsoletes from apptainer - epel - Pagure.io - https://pagure.io/epel/issue/215
21:12:05 <carlwgeorge> seems clear cut in this one that having the obsoletes is allowed by policy, as the package was a rename
21:12:13 <tdawson> Yep
21:13:01 <tdawson> Any negative votes about putting the obsoletes in?
21:13:48 <carlwgeorge> the obsoletes is already there, so it's a no-op unless we want to recommend moving the obsoletes to the apptainer-suid subpackage
21:14:24 <tdawson> I agree ... and it looks like they are planning on moving the obsoletes to the apptainer-suid package anyway.
21:14:47 <tdawson> I am going to mark this as passed unless anyone objects.
21:15:09 <dherrera> I think it's ok
21:15:10 <carlwgeorge> this would be option 3 from my list of all possible outcomes
21:15:23 <tdawson> :)  Yep
21:16:42 <tdawson> And now for the nuclear football
21:16:51 <tdawson> .epel 214
21:16:52 <zodbot> tdawson: Issue #214: remove singularity provides from apptainer - epel - Pagure.io - https://pagure.io/epel/issue/214
21:17:03 <carlwgeorge> i think this may be easier than we expected for the immediate question of removing the provides, as we have 3 +1s in the issue from committee members
21:17:10 <salimma> yeah
21:17:31 <salimma> I think even the maintainer now thinks adding the Provides to the -suid subpackage is better
21:17:45 <pgreco> https://src.fedoraproject.org/rpms/apptainer/pull-request/2
21:17:58 <salimma> as I noted I'd like to see a SC vote on that because of the compatibility issues surfaced so far - when it's ready
21:18:13 <carlwgeorge> that pr confused me, as it's basically a duplicate of my original pr with the changelog entry changed to give him credit for the changes
21:18:48 <pgreco> carlwgeorge: yeap, that was my take
21:19:06 <carlwgeorge> for the future question of re-adding the provides (whether to apptainer or apptainer-suid), i'm strongly opposed to doing something different in fedora and epel
21:19:23 <carlwgeorge> reminder that the apptainer maintainer agreed to remove the provides outright in fedora, but was asking for epel to diverge
21:19:37 <salimma> yeah, that part really concerned me in the original proposal
21:19:50 <salimma> that the maintainer doesn't care following guidelines for Fedora but wants to do their own thing for EPEL
21:20:12 <salimma> if anything, in EPEL the need to maintain compatibility is stronger, not weaker
21:20:25 <carlwgeorge> agreed
21:20:38 <dherrera> there is no justification to diverge from fedora
21:21:18 <pgreco> agreed
21:21:26 <carlwgeorge> for the future question of adding the provides back, i think it should be the same decision for fedora and epel, and we can punt that to fesco
21:21:33 <pgreco> I always try to keep all my branches the same, with minimal conditionals...
21:22:21 <dherrera> If you diverge from fedora, is because you want to make it compatible, not because you want to make it different, I don't think that is acceptable
21:22:23 <tdawson> I agree with keeping Fedora and EPEL together, but that really isn't the debate at this time.
21:23:29 <tdawson> At this point, they have agreed to take the Provides out at this time.  Because current configurations clearly are not compatible enough with singularity.
21:24:01 <tdawson> So, at this point, I think we're all in agreement that this is good.
21:24:03 <carlwgeorge> should we also give the guidance about keeping fedora and epel the same, and deferring to fesco for appeals to add the provides back later?
21:24:42 <tdawson> Well, the pull request is for rawhide, so I assume they are doing the same for Fedora and EPEL
21:25:37 <tdawson> The big question I have is, do we want to discuss what "Compatable enough" is?  Or do we kick that footfall to a future meeting?
21:26:28 <smooge> Going from past experience.. discussions like that are better saved after holidays
21:26:31 <carlwgeorge> i made my pr to rawhide as i was advocating for the same thing in both fedora and epel
21:26:39 <smooge> family gatherings are going to be tough enough
21:27:03 <salimma> smooge: I thought that's Thanksgiving :P
21:27:04 <salimma> but yeah
21:27:19 <salimma> let's keep this light
21:27:23 <tdawson> I agree with salimma's comment about reviting this, possibly with a new ticket.  And possibly even getting Fedora/FesCO's comments on what consitutes "Compatible enough".
21:27:37 <carlwgeorge> if we're going to recommend doing the same thing in fedora and epel, we don't even need to have the "compatible enough" discussion
21:28:11 <carlwgeorge> the only reason this was an epel ticket was because he wanted to diverge
21:28:28 <tdawson> carlwgeorge: Yep ... hense bringing the Fedora people in, whichever are the correct ones.
21:29:38 <carlwgeorge> any other votes before we close out the issue?
21:29:57 <smooge> i vote we close out the issue
21:30:00 <carlwgeorge> lol
21:30:03 <tdawson> So, are we in agreement that the current pull request - https://src.fedoraproject.org/rpms/apptainer/pull-request/2 works?
21:30:07 <salimma> yeah
21:30:16 <carlwgeorge> why not https://src.fedoraproject.org/rpms/apptainer/pull-request/1?
21:30:39 <carlwgeorge> i'm not seeing a difference other than the changelog
21:31:01 <carlwgeorge> it seems he opened it to do the same thing but keeping the provides, then commented out the provides, so they're effectively the same at this point i think
21:31:29 * carlwgeorge looks closer to see if he missed anything
21:31:32 <salimma> yeah, I prefer accepting the original PR. though... we can't really dictate which one to merge, can we
21:31:49 <salimma> generally even if I need to rework someone's PR I still give them credit, so... this stinks a bit
21:32:51 <salimma> oh the runtime one is different
21:32:57 <salimma> Carl removes the Obsoletes, this one keeps the obsoletes in
21:33:01 <carlwgeorge> sif-runtime?
21:33:37 <salimma> on singularity-runtime
21:33:57 <salimma> and the maintainer also adds more blurb in the Summary stating this package 'is formerly known as Singularity'. that part is fine I guess
21:34:00 <carlwgeorge> oh, that one didn't move did it
21:34:04 <carlwgeorge> fixed
21:34:09 <salimma> see https://src.fedoraproject.org/rpms/apptainer/pull-request/2#_1__28
21:34:31 <tdawson> There is also wording differences in some of the comments.
21:34:32 <salimma> while Carl's remove that Obsoletes: https://src.fedoraproject.org/rpms/apptainer/pull-request/1#_1__28
21:35:04 <carlwgeorge> singularity-runtime obsoletes (an old thing unrelated to this discussion) restored to it's original spot
21:36:13 <tdawson> Honestly, I don't care which.
21:36:59 <carlwgeorge> i'm happy to merge this and https://src.fedoraproject.org/rpms/singularity-ce/pull-request/1 and push them out in the same bodhi update, or another proven packager can merge them
21:37:12 <carlwgeorge> ideally they'll go out at the same time so the sif-runtime conflicts works as intended
21:37:56 <tdawson> carlwgeorge: Considering the package maintainer has been having this disucssion with us ... I think it would be best if they did the merges and builds.
21:38:24 <salimma> yeah, I think give them a sidetag but ask that they don't push the Bodhi update, but we handle that part
21:38:30 <carlwgeorge> they just can't submit a bodhi update that includes both as they're not maintainers of each others' packages
21:38:45 <salimma> right. one of us can do that
21:38:57 <carlwgeorge> they don't build against each other so i don't think the side tag is necessary
21:39:17 <tdawson> Wow ... this discussion just changed to something completely different fast
21:39:35 <smooge> i agree with tdawson on this... and I am not following where things are going
21:39:36 <carlwgeorge> sorry about that, we can put a pin in this and pick it up in #epel after the meeting
21:39:55 <tdawson> No ... what are we suddenly talking about?
21:40:16 <carlwgeorge> coordinating the merging and release of two pull requests
21:40:21 <tdawson> Why?
21:40:52 <salimma> to make sure we're not favoring one singularity implementation over another?
21:40:57 <salimma> but yeah no need for a side tag
21:41:04 <carlwgeorge> so the changes to conflict on sif-runtime happen at the same time.  it's not part of the ticket discussion, and my apologizes for going down the rabbit hole.
21:41:45 <tdawson> Does this really mater?
21:42:41 <carlwgeorge> so the package properly conflict with each other, yes.  but we can move on for the meeting.
21:42:58 <tdawson> Is someone planning on doing a mass update as soon as these hit?  I'm not really seeing a problem ... but ya ... moving on.
21:43:35 <tdawson> #topic Old Business
21:44:07 <tdawson> This last thing was supposed to happen at last weeks meeting.  I just wanted more than one LGTM before I merged this ...
21:44:15 <tdawson> Tweaks to retirement policy documentation
21:44:17 <tdawson> https://pagure.io/epel/pull-request/213
21:44:20 <smooge> ah sorry
21:44:28 <tdawson> Not a problem.
21:44:52 <tdawson> My finger keeps hovering over the merge button on it, but I never push it.
21:45:16 <smooge> LGTM
21:45:23 <smooge> push it
21:45:24 <tdawson> Thanks
21:45:34 <smooge> doooo it anakin
21:46:04 <tdawson> And done ...
21:46:37 <smooge> Good.. you have taken your first step
21:46:48 <smooge> sorry wrong meeting
21:47:03 * tdawson starts breathing in his facemask.
21:47:15 <tdawson> I think that's all the old business ...
21:47:22 <tdawson> #topic General Issues / Open Floor
21:47:30 <smooge> so i had one thing..
21:47:38 <tdawson> Sure
21:47:57 <smooge> i have been working on better counting of older epel systems before 8
21:48:07 <smooge> so EPEL-7 (durh).
21:48:08 <tdawson> Ooohhh
21:48:52 <smooge> the best method would be to add something to epel-release which mimic'd countme :). But because that might also be the worst method.. I have been coming up with something else
21:49:15 <jonathanspw> .hi
21:49:16 <zodbot> jonathanspw: jonathanspw 'Jonathan Wright' <jonathan@almalinux.org>
21:49:30 <tdawson> Hi jonathanspw
21:49:52 <jonathanspw> Sorry I'm late :)
21:50:41 <smooge> the current counting method just says 'if your ip has asked for an epel-7 repodata, we count you once per day'. That gives a number around 3.2 million per day.
21:51:16 <tdawson> Not a problem ... smooge was just telling us how the dark side can get us better epel 7 numbers.
21:51:17 <smooge> However I went trawling through the data and realized that over 50% of all EPEL-7 requests per day come from around 30k ips out of the 3.2 million
21:51:42 <salimma> company IPs?
21:52:00 <tdawson> Firewall of China?
21:52:02 <smooge> doing some rough guessing math, it comes out that we are undercounting by 10
21:52:31 <smooge> no the top 4 are google, microsoft, and amazon
21:52:40 <tdawson> So ... it should be 32 million?
21:52:44 <smooge> yeah
21:52:48 <tdawson> wow
21:53:01 <salimma> ah, likely cloud instances then
21:53:23 <salimma> wait, how does the math work? (still jetlagged)
21:54:17 <smooge> so a lot of systems have a cron script which updates mirror data every hour. That means that a lot of ips would show up around 24 times per day per 1 system
21:55:11 <smooge> so I did a very simple ceiling(number_per_ip/24) and saw the number was around 10 times larger
21:55:37 <smooge> if I were to assume that every hit was a differnt system.. then the count of 3.2 million would need to be increased by about 24 :)
21:55:44 <smooge> sorry 240
21:55:53 <salimma> ah
21:56:20 <salimma> that's a lot of instances
21:56:42 <smooge> because there are about 750 million requests to the proxies per day
21:56:47 <salimma> what's the downside of trying to replicate how countme works?
21:56:49 <smooge> for epel-7
21:57:27 <carlwgeorge> countme is built into libdnf
21:57:37 <salimma> ah
21:57:46 <salimma> right, so I guess without deploying something to replace it client side, you can't do that
21:57:48 <smooge> 1) it would need to have epel-release add a cron job which once per week did a wget/curl request that said basically what countme does'
21:57:57 <carlwgeorge> re-implementing it outside of libdnf as a cron job sounds like a bad idea
21:58:14 <smooge> 2. it would also be adding something outside of what may have been initially in any security review of allowing epel at said sites
21:58:20 <carlwgeorge> way too easy to duplicate the cron job to manipulate the numbers
21:58:49 <smooge> carlwgeorge, as if any distribution would do that
21:58:57 <smooge> with the regular countme numbers
21:59:33 <salimma> phew, and look at the time!
21:59:34 <smooge> honestly I am not worried about manipulating the numbers as much as the 2nd.
21:59:50 <tdawson> Yep, that worries me more
22:00:06 <smooge> anyway.. that was all. I am working on it and hope to have better numbers next quarter
22:00:08 <carlwgeorge> right before the buzzer, i wanted to highlight one more simple thing here in the open floor
22:00:15 <carlwgeorge> i retired a bunch of epel packages that were in rhel at newer versions, so straightforward retirements with no announcement
22:00:23 <carlwgeorge> and two more that i did announce as it resulted in a few subpackages going away, but they were -doc and -test subpackages that nothing required
22:00:39 <tdawson> I saw that.
22:00:46 <tdawson> And thank you for doing that.
22:00:52 <carlwgeorge> full list for those curious https://paste.centos.org/view/e6a1c631
22:00:56 <pgreco> yeap, cleanup is good
22:01:12 <carlwgeorge> well, that includes tesseract-tessdata which i didn't do, the maintainer did after i pointed out the policy
22:02:12 <pgreco> I just have one extra comment that was discussing with salimma before the meeting. I picked up proxychains-ng because it was orphaned in fedora. if anybody wants to share it, please ping me
22:02:32 <salimma> ah yeah that
22:02:51 <salimma> right now pgreco and I are both admins, we're happy to add anyone
22:03:08 <pgreco> for both fedora and epel
22:03:24 <salimma> looks like the previous admin retired everything / became inactive otherwise, since there's no new bug that'd trigger the orphaning
22:03:54 <tdawson> Oh, and I have one extra thing, though I think everyone already knows it.  I hearby declare by my new dark side powers that we will have no meeting next week.
22:04:03 <smooge> Nooooooooo
22:04:13 <carlwgeorge> excellent
22:04:33 <pgreco> no work, no epel
22:04:34 <tdawson> Yessss ... give in to the holiday side ...
22:04:44 <pgreco> next week is full of nothing, looks good!
22:05:08 <tdawson> Anything else?  Cuz we've gone a little over.
22:05:43 <carlwgeorge> hope everyone has a good holiday break, see y'all next year
22:05:50 <jonathanspw> merry crimmus all!
22:05:57 <tdawson> Thank you all for a wonderful year full of EPEL.  I'll talk to you next year.
22:06:06 <smooge> May Krampus visit you all and all a good night
22:06:11 <pgreco> see ya next year
22:06:15 <smooge> sorrry wrong one
22:06:16 <pgreco> happy Festivus
22:06:25 <tdawson> #endmeeting