Home
release-team@conference.openafs.org
Friday, March 22, 2019< ^ >
Room Configuration
Room Occupants

GMT+0
[12:12:01] meffie joins the room
[12:22:45] <kaduk@jabber.openafs.org/barnowl> Sorry folks, I don't think I can stay awake for the meeting (just flew
to Europe and didn't sleep on the plane).
I got a lot of code review done on the flight, though (for
rxgk-phase1)
[12:31:54] <meffie> good day ben, no worries. would you have time before monday afternoon (west coast time) to review the slides for hepix? (sorry i did not finish them sooner:(
[12:33:12] <meffie> great news on rxgk-phase-1. looks like that stack of patches is getting smaller.
[13:00:36] wiesand joins the room
[13:02:52] <meffie> good afternoon wiesand.
[13:03:05] <wiesand> good morning mike
[13:04:11] <meffie> just looking at 13533. merged already. i am curious why i dont see that bug.
[13:04:58] <meffie> (i use make dest on the master branch every day. must be some different conditions.)
[13:05:22] <wiesand> do you still install the kauth stuff?
[13:06:38] <mvita> sorry for my tardiness
[13:06:52] <meffie> no
[13:08:33] <meffie> (and the only thing in pod3 is AFS.ukernel.pod.)
[13:09:14] <meffie> (the patch does no harm, just curious)
[13:09:55] <meffie> good morning mvita.
[13:09:57] <wiesand> me too ;-)
[13:10:31] <mvita> good morning
[13:10:44] <wiesand> good morning mark
[13:11:03] <mvita> hello Stephan
[13:11:09] <wiesand> I don't have very much on the agenda today
[13:11:27] <mvita> okay, we can just gab then
[13:11:46] <meffie> ben is out too (see the scrollback)
[13:11:48] <mvita> whose turn was it to bring the donuts?
[13:12:22] <wiesand> yours, iirc.
[13:12:29] <wiesand> besides 13533 I also pushed 13534, addressing another user report
[13:12:33] <meffie> i can report some good feedback on 1.8.3pre3, see inbox.
[13:12:49] <wiesand> seen it, thanks a lot
[13:14:06] <wiesand> I have no news on 1.6.x yet, but hopefully I'll be able to turn to that soon.
[13:14:07] <meffie> still waiting to see if we get any more feedback, but some people are out.
[13:15:21] <meffie> also, today trying to finish slides for the report for hepix on monday.
[13:15:52] <wiesand> that's very timely ;-)
[13:16:19] <meffie> i'm trying to follow the basic format from last year: https://github.com/meffie/talks/blob/master/hepix-2018-openafs-rel-team/hepix-2018-openafs-rel-team.md
[13:16:59] <meffie> yeah, sorry for the lateness. somehow i keep getting distracted :)
[13:18:02] <wiesand> i was quite serious - i never manage to get the slides done several days before the presentation
[13:18:54] <wiesand> format looks good
[13:19:52] <wiesand> when doing the timelines, blame the recent slowness on the release manager
[13:20:06] <meffie> ok, thanks. there's a pdf for the actual presentation.  i'm just gathering info for what has changed in the 1.6 and 1.8 releases, and what is in the works.
[13:21:09] <meffie> i can reuse the slide "upcoming 1.6.23" and call it 1.6.24 ?
[13:21:46] <wiesand> sure ;-)
[13:21:49] <meffie> :)
[13:22:09] <wiesand> but mention at least the two fixes submitted by Jeffrey
[13:22:23] <meffie> ok, thanks.  also, i dont remember which linux versions we support in which releases, i'll have dig that up.
[13:22:34] <wiesand> since I figure he'll bash us for having those released yet ;-)
[13:22:48] <wiesand> not havinf
[13:22:51] <wiesand> having
[13:23:23] <meffie> looking forward to the feedback :)
[13:24:12] <wiesand> linux versions should all be in NEWS
[13:24:41] <wiesand> I see there's quite a bit of progress on #134904
[13:25:00] <meffie> ok, thank you. i'll summarize the NEWS. good idea.
[13:25:11] <wiesand> But I yet have to read up and look at the changes
[13:26:43] <wiesand> NB 1.6.24 should also get most of the other applicable changes from 1.8.3
[13:28:37] <wiesand> the last topic I had in mind for today is another one recently discussed on the list: binary distribution
[13:29:15] <meffie> the request for rpms?
[13:29:29] <wiesand> yes, that one
[13:29:32] <meffie> ok.
[13:29:51] <wiesand> it's correct that I haven't uploaded rpms for quite a while
[13:30:08] <wiesand> but then this is the first time since that anyone asks about them
[13:30:10] <meffie> sna continues to build rpms for el6 and el7. i've been building them with mock.
[13:30:45] <wiesand> but you don't have the "official" openafs signing key, do you?
[13:31:02] <meffie> they are unsigned and built from the git tags.
[13:31:14] <mvita> hrrm
[13:31:35] <mvita> it would be nice if they were signed by SNA or openafs
[13:31:56] <wiesand> Stephen also reliably builds EL6 and EL7 rpms - and he has the key
[13:31:59] <meffie> yes, we do sign the macos packages with an SNA key
[13:32:23] <meffie> (because that's otherwise mac is unhappy)
[13:32:29] <mvita> aye
[13:32:33] <meffie> er, s/that's//
[13:32:41] <mvita> that day is coming for Linux and Solaris kmods, too
[13:33:00] <meffie> yes
[13:33:07] <wiesand> right
[13:33:44] <meffie> (also cheyenne has a koji setup now and we will be using that for builds at some point soon)
[13:34:00] <mvita> (not to conflate binary signatures with package signatures)
[13:34:28] <wiesand> but installing an rpm w/o either inspecting every single bit in there or trusting the builder and verifying his signature is sufficient to run the highest risk - no need for a malicious kmod
[13:35:04] <wiesand> (they are related though)
[13:36:02] <wiesand> now, the reason why I haven't uploaded binaries for a while is that it's pretty tedious
[13:36:27] <wiesand> they are not available at the time of announcement
[13:36:59] <wiesand> Stephan builds a module for each and every EL kernel, typically oldest first
[13:37:14] <meffie> i could look into how we sign our other rpms and see if we could do the same for vanilla openafs rpms.
[13:37:21] <wiesand> so it can take many days before they are complete
[13:38:05] <wiesand> and then it's hundreds of packages, and rsyncing them from Edinburgh to Pittsburgh vie Berlin takes ages
[13:38:32] <wiesand> and then they don't show up on the release page, so that has to be done too
[13:38:57] <wiesand> so, with a single reuqest during the last 3 releases, is it worth the effort?
[13:39:16] <meffie> i think it takes less than a day for my mock builder to make all the kmods for el6 and el7.
[13:39:42] <meffie> i feel it would be nice, yes.
[13:40:06] <meffie> if we could sign them and automate the transfer
[13:40:31] <wiesand> and then the gatekeeper's decision not to provide binaries for new releases of either openafs or operating systems still stands
[13:40:39] <wiesand> affecting EL7 and 1.8.x
[13:41:28] <meffie> sorry, i dont understand.
[13:42:54] <wiesand> the gatekeepers decided that the project shall not provide binaries for EL7+, nor for OpenAFS 1.8+ releases
[13:43:24] <wiesand> this would have to be overruled by Ben (and accepted by the last gatekeeper)
[13:43:39] <meffie> ok
[13:44:46] <wiesand> I also feel no urge to upload a new kmod and redo the web page upon ever EL kernel update
[13:45:10] <wiesand> so maybe this could be limited to base packages + dkms?
[13:45:13] <meffie> yes, that's not a good use of time. it has to be automatic.
[13:46:01] <meffie> dkms seems like it could be an option.
[13:46:20] <meffie> perhaps this is a foundation discussion?
[13:47:23] <wiesand> yes, makes sense
[13:47:53] <mvita> no, I don't think so.
[13:48:18] <wiesand> ok?
[13:49:03] <mvita> meffie talked me into it out-of-band
[13:49:13] <mvita> <ouch, that smarts>
[13:49:17] <mvita> ;-)
[13:49:18] <meffie> well, i think it is worth the discussion.
[13:49:38] <meffie> to be clear, they were already talking about macos binaries.
[13:51:14] <meffie> perhaps they could fund the binary distribution work.
[13:52:27] <mvita> ^this
[13:52:49] <wiesand> Pascal?
[13:53:13] <mvita> ?
[13:54:04] <meffie> the language or the person?
[13:54:46] <wiesand> it was my first programming language
[13:55:24] <meffie> ah, one of mine too. but i'm not following :)
[13:55:51] <meffie> oh, mvita's "hat" symbol. i get it.
[13:56:00] <mvita> OH I just got it too
[13:56:01] <wiesand> type
  pNode = ^Node;
  Node  = record
          a : integer;
          b : char;
          c : pNode  {extra semicolon not strictly required}
          end;
var
  NodePtr : pNode;
  IntPtr  : ^integer;
[13:56:06] <meffie> lol
[13:56:07] <mvita> I don't know Pascal at all
[13:56:25] <meffie> (sorry my Pascal is a bit rusty:)
[13:56:47] <mvita> (mine is non-existent - if I ever knew it, all those neurons died long ago)
[13:56:48] <wiesand> so is mine - there was just that vague association ;-)
[13:57:42] <wiesand> i.o.w., I don't know what else to make of mark's "^this"  
[13:57:47] <mvita> sorry, " ^ this " is chat shorthand for " I agree with what was just said"
[13:58:02] <mvita> emphatically
[13:58:13] <wiesand> ah, my "chat" is non-existent ;-)
[13:58:19] <mvita> LOL
[13:58:46] <wiesand> ok, let's wrap it up
[13:58:55] <meffie> at any rate, DKMS bins seems reasonable at a minimum.
[13:59:07] <meffie> and would be an improvement.
[13:59:17] <wiesand> we won't get this sorted out today, I just wanted to point out the issues
[13:59:34] <mvita> thank you
[13:59:45] <meffie> yes, thanks for bringing it up. i was not sure what to respond on the list.
[14:00:01] <wiesand> NB there's another one: back then it was also ruled that we shall stop providing binaries using transarc paths
[14:00:34] <wiesand> so it has been EL6-only for years now anyway
[14:01:55] <wiesand> on automation: doing that requires a lot of care w.r.t. security
[14:02:45] <wiesand> I take good care of my g.c.o. admin password - handing it down to a robot feels wrong.
[14:04:07] <wiesand> There's a lot of responsiblity here. We really don't want to be the cuplrit for openafs sites becoming trojaned.
[14:04:42] <wiesand> [fullstop]
[14:06:32] <wiesand> Let's pick this up again next week, or maybe on the list.
[14:06:53] <meffie> ok, thank you.
[14:07:27] <wiesand> Feel free to voice your personal opinion, even if it clashes with mine.
[14:07:53] <wiesand> Anything else to discuss today?
[14:08:09] <mvita> no, nothing here
[14:08:18] <meffie> no thanks
[14:08:40] <wiesand> Let's adjourn then. Thanks a lot Mark & Mike!
[14:08:54] <wiesand> Looking forward to the HEPiX slides.
[14:08:56] <mvita> Thank you Stephan!
[14:09:42] wiesand leaves the room
[14:09:54] meffie leaves the room