Home
release-team@conference.openafs.org
Friday, December 27, 2019< ^ >
Room Configuration
Room Occupants

GMT+0
[15:44:59] meffie joins the room
[15:45:17] meffie leaves the room
[15:46:46] meffie joins the room
[16:59:58] <meffie> good morning/evening
[17:06:56] wiesand joins the room
[17:07:30] <wiesand> good evening; sorry for being late
[17:08:47] <meffie> no problem.
[17:10:21] <wiesand> Not that much news from my side today - frankly, I let my vacation take precedence.
[17:10:41] <meffie> nice! hope it was a good holiday!
[17:10:56] <wiesand> It was and still is.
[17:11:42] <meffie> we had some time off here too. most people are out for the holidays.
[17:11:50] <wiesand> This night will be the first one I'll not be alone around here.
[17:12:50] <wiesand> That's fine. I'm actually surprised not to be the only one in this meeting.
[17:14:17] <wiesand> Swapped holiday/new year wishes with Jeffrey, and in return I received a warning that our current master may have problems with files > 2 GiB
[17:14:51] <meffie> ok
[17:15:44] <meffie> looks like ben is away at the moment, but his jabber client is running.
[17:15:46] <wiesand> $ dd if=openafsz of=/dev/null bs=4M
dd: error reading 'openafsz': Input/output error
512+0 records in
512+0 records out
2147483648 bytes (2.1 GB, 2.0 GiB) copied, 9.25338 s, 232 MB/s
[17:16:34] <wiesand> openafsz is supposed to be 10GiB in size, and that's supposed to be an oenafs-master client
[17:16:55] <wiesand> any chance you could look into it?
[17:17:10] <meffie> interesting. yeah, we can look
[17:17:17] <wiesand> Thanks!
[17:17:25] <meffie> thanks for the report!
[17:18:57] <meffie> looks like it would be an easy test to add too.
[17:19:23] <wiesand> Yes. Not quite cheap though.
[17:19:48] <meffie> yeah
[17:20:24] <meffie> is the file in the local filesystem or in /afs ?
[17:20:34] <meffie> i guess we should test both.
[17:20:51] <wiesand> But yes, well worth introducing if the resources can be spent.
[17:22:05] <wiesand> given "dd of=/dev/null" the source file is probably in /afs…
[17:22:21] <meffie> oh, yes, of course.
[17:22:45] <meffie> sorry for the dumb question.
[17:23:56] <meffie> i know there was another fix for master for the client. i think andrew already pushed a fix to gerrit.
[17:25:07] <meffie> yes, gerrit 13984 is a fix for master (master only)
[17:25:33] <meffie> it is for the "linux-native-mounts" patches.
[17:25:55] <wiesand> It would really be nice not to enter 2020 with a "large file" problem
[17:26:05] <meffie> yes
[17:26:46] <wiesand> has the linux-native-mounts series been merged on master? forgive me, I'm really supposed to know
[17:26:58] <meffie> some have, but some have not yet.
[17:27:17] <wiesand> ok, so maybe there's the culprit
[17:27:41] <meffie> there are only a few, so it is not hard to try by applying the patches manually.
[17:28:46] <meffie> maybe, there are more patches on master than the 1.8.x and 1.6.x branches.
[17:29:35] <meffie> we can look
[17:29:38] <wiesand> there certainly are ;-)
[17:29:58] <wiesand> they haven't tested 1.6 nor 1.8 - well, that's something I can promise to get done before next week's meeting
[17:30:09] <meffie> ok, thanks.
[17:31:13] <wiesand> Any other news?
[17:31:49] <wiesand> If not, Happy New Year!
[17:31:53] <meffie> looks like ben and andrew have been making progress on the rxgk-phase2 patches on master.
[17:32:06] <wiesand> Good.
[17:32:47] <wiesand> But I think that has rather mid-term impact. Though it *is* important.
[17:33:08] <meffie> cheyenne and i have been pushing patches to prep for ipv6 support to our github repos.  i have a version of "vos" that is address agnostic now.
[17:34:05] <meffie> the main thing is to separate the volume server rx connections from the volume "server numbers"
[17:34:14] <wiesand> Wow. Yes, ipv6 (supposed to be) improtant too.
[17:35:20] <meffie> we have not pushed to gerrit yet because we are rebasing and such as we find dead-ends.
[17:36:15] <meffie> there will be "phases" of commits, some just to fix bugs we found and to remove dead code, and other clean ups first.
[17:37:02] <meffie> then changes to avoid assuming serverNumbers are addresses, etc.
[17:37:59] <meffie> hopefully to be ready when ben has finished merging the rxgk-phase2 patches.
[17:38:01] <wiesand> Sounds like a plan. I'm not exactly looking forward to the resulting code skew, but…
[17:38:48] <meffie> yes, the hardest part is avoiding "unneeded" changes.
[17:39:05] <meffie> but there will be code skew.
[17:40:11] <wiesand> Keeping changes as small and as self contained as possible would help..
[17:40:51] <meffie> yes, but there will be many small changes in many files. no way around that.
[17:41:28] <meffie> it is a sort of "porting" project.
[17:42:13] <meffie> but the changes will be boring, so that is good.
[17:43:49] <wiesand> yeah, but if an incremental introduction is infeasible, that's kind of a show stopper
[17:44:25] <meffie> also, we have to be compatible as we good. so libraries we make have to have wrappers and such.
[17:44:39] <meffie> yes, we are doing this incrementally.
[17:45:09] <wiesand> ok
[17:46:09] <meffie> at this point we only support ipv4 so far, but making changes to not preclude ipv6 servers and clients in the future.
[17:46:34] <meffie> does that make sense?
[17:47:08] <wiesand> in my personal opinion, it does
[17:47:51] <wiesand> all existing afs sites should have plenty of v4 addresses
[17:49:42] <meffie> i dont think "plenty" is true. there are waiting lists now, and it gets harder all the time.
[17:51:18] <meffie> anyway, we can make incremental improvements.
[17:52:02] <wiesand> yes, v4 addresses are a scarce resource - but  I guess all sites running an AFS cell today have an allocated share of them that will last a while
[17:53:13] <wiesand> and new sites doomed to be v6-only are not all that likely to be desperate to whip up an afs cell
[17:53:52] <kaduk@jabber.openafs.org/barnowl> Oops, just remembered that I was supposed to be watching here an hour
ago.  No progress to report from my end; busy with holidays and a
touch of work stuff, and a bit of a head cold that slowed me down.
[17:54:02] <meffie> hello Ben!
[17:54:24] <meffie> oh, i hope you feel better soon.
[17:55:03] <kaduk@jabber.openafs.org/barnowl> I am definitely on the upswing; hopefully just another day or two left
[17:55:59] <wiesand> Being busy with holidays is definitely acceptable these days, as is being slowed down by a cold.
[17:57:27] <wiesand> Like I've been busy with being on vacation the last two weeks - despite all plans to use some of it to work on openafs.
[17:59:19] <meffie> as well you should!
[17:59:48] <wiesand> Adjourn?
[18:00:21] <meffie> yes, have a good week everyone. ben, get well soon.
[18:00:42] <wiesand> Yes, get well soon Ben!
[18:00:49] <wiesand> CU next week.
[18:00:52] wiesand leaves the room
[18:03:10] <kaduk@jabber.openafs.org/barnowl> Thanks everyone!
[18:07:48] meffie leaves the room