Home
release-team@conference.openafs.org
Thursday, May 26, 2022< ^ >
kaduk@jabber.openafs.org/barnowl has set the subject to: openafs release team
Room Configuration
Room Occupants

GMT+0
[15:59:37] wiesand joins the room
[16:02:57] <wiesand> Moin
[16:03:01] <mvitale> Hi Stephan
[16:03:05] <wiesand> Moin
[16:03:20] Cheyenne joins the room
[16:03:21] <mvitale> we just finished another meeting
[16:03:22] <wiesand> Moin
[16:03:23] <Cheyenne> Hello all
[16:03:27] meffie joins the room
[16:03:39] <meffie> greetings.
[16:04:01] <meffie> gsoc 22 kickoff went very well imho
[16:05:50] <wiesand> Glad to hear that.
[16:06:26] <wiesand> Re 1.8. stable: I just merged 14926..14931.
[16:08:35] <mvitale> nice
[16:08:53] <kaduk@jabber.openafs.org/barnowl> yay
[16:09:13] <Cheyenne> +1
[16:10:38] <meffie> thank you wiesand!
[16:12:57] <wiesand> As the next step, I'm kind of throwing in the towel. Looking at Cheyenne's sack and the required modifications in there make me hopeless. Cheyenne: could you please submit this stack to openafs_stable-1_8_x? If at all possible, please base it upon the "easy" open FBSD changes (14878 14909 14910 14911 14920 14921 14919). AND I'd like those LINUX_ENV changes (topic: cleanup-LINUX_ENV) to go in at the same time as well. Is that doable?
[16:13:34] <Cheyenne> Yes -- it should be doable
[16:14:36] <wiesand> Thanks Cheyenne! And everybody else: that's quite a bunch, and will require quite a bit of your time to review.
[16:15:05] <meffie> ok, thanks Cheyenne.  I make some time to do reviews.
[16:16:22] <meffie> s/I/I will/
[16:17:08] <kaduk@jabber.openafs.org/barnowl> I will also make some time for reviews.  A unicast mail to me when the
changes are in gerrit would be appreciated in this case (the in-gerrit
review requests are too easy to get lost in the flood of mail from
gerrit).
[16:17:49] <wiesand> I can arrange that ;-)
[16:18:23] <kaduk@jabber.openafs.org/barnowl> :)
[16:19:03] <wiesand> And sorry if I sound a bit rude today. But then it was Ben who suggested I should twist arms ;-)
[16:19:20] <meffie> lol
[16:19:55] <wiesand> And that's what I have for today.
[16:20:15] <Cheyenne> not a problem, and glad to help
[16:20:25] <mvitale> Thank you Stephan
[16:23:39] <kaduk@jabber.openafs.org/barnowl> I got to look at some of the fallout from changing the glock/Rx
interaction (14184) that caused issues for UKERNEL and DARWIN (the
DARWIN one might still require some work, through no fault of Marcio)
[16:24:01] <kaduk@jabber.openafs.org/barnowl> And have been going through the linux-5.18 stuff this morning (but not
much progress to report yet)
[16:24:26] <meffie> was the the same fallout Cheyenne was looking at?
[16:24:32] <kaduk@jabber.openafs.org/barnowl> I do see the gcc12 and clang14 warning fixups that Cheyenne uploaded
yesterday, which will hopefully be quick
[16:25:05] <kaduk@jabber.openafs.org/barnowl> I'm thinking of 14963 and 14932 for the rx/glock fallout
[16:25:18] <meffie> ok, thanks
[16:25:25] <kaduk@jabber.openafs.org/barnowl> Cheyenne will have to speak to what he was looking at ;)
[16:26:00] <Cheyenne> Let me know if what I did for 14956 is what you are looking for in terms of the huge #if clause
[16:27:00] <kaduk@jabber.openafs.org/barnowl> eyeballing it, that looks about as expected
[16:27:20] <meffie> fyi, 14743 did fix the oops i saw on master when running cmdebug.
[16:27:25] <Cheyenne> For 14963 -- I was just trying to test afsd.fuse and it just failed immediately.  Seemed like a simple enough fix
[16:28:10] <kaduk@jabber.openafs.org/barnowl> Yeah, that was just Andrew and I forgot to look at ukernel at all, I
think.
[16:28:15] <Cheyenne> biggest problem was that the messages were not being displayed (I'll have to look at that later)
[16:28:48] <Cheyenne> I suspect that was the case
[16:30:53] <Cheyenne> Linux-5.19 will require some work (I haven't dug too deep, but it doesn't look too bad -- some more changes in the address space operations stuff)
[16:30:57] <kaduk@jabber.openafs.org/barnowl> The issue about 14932 is more complicated though, with me (I can't
quite speak for Andrew) just having not cheked enough call chains to
find affected locations.
[16:31:12] <wiesand> Nice :-/
[16:31:21] <kaduk@jabber.openafs.org/barnowl> ok.
Do we have timescale estimates for when linux 5.19 will become
pressing?
[16:32:09] <Cheyenne> It hasn't reached rc0 yet.. -- so I don't know.
[16:32:21] <kaduk@jabber.openafs.org/barnowl> ok.  "We have some time", at least
[16:32:43] <meffie> prob on the order of a few weeks?
[16:33:03] <mvitale> yes, it's usually 6 or 7 weeks per Linux release
[16:34:09] <Cheyenne> the changes at this point don't look too complicated (at least from the first glance)
[16:34:23] <wiesand> BTW Mike: Last year my registration fee for the workshop was refunded. This year, I had to block the card I used for registration (someone, probably in the US, went shopping with the card data) and it's no longer valid. If that's any hassle, just don't bother.
[16:35:54] <kaduk@jabber.openafs.org/barnowl> I haven't gotten around to reading the ToS for the payment processor
and registering yet :(
[16:37:14] <wiesand> The abuse started before I registered for the workshop, thus that wasn't the culprit.
[16:37:21] <meffie> sorry to hear about your card.
[16:38:08] <meffie> let me know if there is anything i can do.
[16:38:57] <wiesand> Rid the world of those frauds?
[16:40:17] <wiesand> Anything else to discuss today?
[16:40:31] <meffie> yes
[16:40:57] <meffie> for master, i have been looking at the rxgk-phase2 patches..
[16:41:15] <wiesand> ah
[16:41:40] <meffie> we have 14 patches in gerrit, and andrew has another 23 patches in github queued up for gerrit after those first 14 are merged.
[16:42:10] <meffie> i wonder if it is possible to have a target timeline when we can have an rxgk implementation merged on master?
[16:42:41] <meffie> security people want to kill des it seems.
[16:43:27] <wiesand> Am I hearing a "2.0 RSN" thundering?
[16:43:39] <kaduk@jabber.openafs.org/barnowl> Hmm, target timeline... maybe "christmas"?
[16:43:41] <meffie> maybe a 1.10?
[16:44:04] <kaduk@jabber.openafs.org/barnowl> We still need to make headway on the db format changes and upgrade
path, though.
[16:44:12] <kaduk@jabber.openafs.org/barnowl> (unless those are hiding in github and I didn't notice?)
[16:45:04] <meffie> yes, but i think on a different branch?
[16:45:06] <wiesand> Yeah, "Christmas" worked fine for many years ;-)
[16:45:39] <meffie> let me verify with andrew i'm looking at the correct branches :)
[16:46:06] <mvitale> I would like to take a look at those as well
[16:46:27] <meffie> my notes say: https://github.com/adeason/openafs/commits/adeason/rxgk-phase2
[16:47:48] <wiesand> Do we really need DB format changes?
[16:48:55] <mvitale> I think for key material, yes?
[16:49:06] <meffie> i think for some features of rxgk we do, but maybe not for a basic implementation to remove des.
[16:49:23] <mvitale> per-server keys, that is
[16:49:42] <kaduk@jabber.openafs.org/barnowl> The VL_RegisterAddrsAndKey RPC is used to indicate that a fileserver
supports rxgk
[16:50:05] <kaduk@jabber.openafs.org/barnowl> IIRC, in order for the RPC to succeed we need to actually be able to
store+use the key in question.
[16:50:38] <kaduk@jabber.openafs.org/barnowl> It's possible to build something that still assumes a single shared
cell-wide key and uses out-of-band configuration to determine when to
switch to rxgk.
[16:51:17] <meffie> yes
[16:51:31] <kaduk@jabber.openafs.org/barnowl> There is also some scope for prdb changes to support extended names,
but that is not as critical
[16:51:50] <meffie> yes, that can be done next christmas :)
[16:53:10] <meffie> ok, thanks for the discussion. yes it will be a lot of reviews, but at least we have an implementation to consider :)
[16:54:51] <wiesand> If we could possibly ´get by w/o DB format changes for the time being, I'd be in favour of that.
[16:55:41] <wiesand> Since we're approaching the top of  the hour: Adjourn?
[16:55:51] <meffie> yes, i think so. we want to change the DB format for other reasons, but on the wire crypto is more critical in my opinion. thanks
[16:57:33] <meffie> thanks for the discussion.  have a good weekend all.
[16:57:33] <kaduk@jabber.openafs.org/barnowl> Probably true.
[16:57:41] <kaduk@jabber.openafs.org/barnowl> And I don't have any other topics for today, so ok to adjourn
[16:58:06] <Cheyenne> Sounds good..
[16:58:11] <wiesand> Good. Thanks a lot everybody!
[16:58:17] <kaduk@jabber.openafs.org/barnowl> thanks everyone!
[16:58:17] <Cheyenne> have a good week everyone :)
[16:58:44] wiesand leaves the room
[17:00:06] meffie leaves the room
[23:26:31] Cheyenne leaves the room