Home
release-team@conference.openafs.org
Friday, February 1, 2019< ^ >
Room Configuration
Room Occupants

GMT+0
[13:45:29] meffie joins the room
[13:57:42] <meffie> good morning.
[13:59:45] <kaduk@jabber.openafs.org/barnowl> Getting an early start, eh?
[14:01:45] <meffie> yes
[14:03:02] <kaduk@jabber.openafs.org/barnowl> I had hoped to get an earlier start than I actually did :(
[14:03:04] <meffie> working out of our local co-working space today.
[14:03:37] <meffie> and was able to get here before 9, even with the snow today :)
[14:03:58] <kaduk@jabber.openafs.org/barnowl> :)
[14:05:54] wiesand joins the room
[14:06:25] <mvita> present
[14:06:26] <wiesand> good morning, sorry for being late
[14:06:55] <mvita> "fashionably late"
[14:07:05] <kaduk@jabber.openafs.org/barnowl> That just gives me more time to review review review
[14:08:25] <wiesand> I got around to work some on 1.8.3pre1 after the last meeting, and I think we're at least close
[14:08:43] <wiesand> Still have to check whether we're missing anything plnned to be include
[14:10:18] <wiesand> And fix whitespace in NEWS, at least
[14:10:56] <meffie> i only mentioned those to prove i read it :)
[14:11:12] <kaduk@jabber.openafs.org/barnowl> :)
[14:11:29] <wiesand> q.e.d. ;-)
[14:11:40] <meffie> thank you wiesand for doing the NEWS
[14:12:45] <mvita> reviewing NEWS as we speack
[14:12:48] <mvita> speak
[14:13:05] <wiesand> I'm sure they could be improved, in particular the "Assorted fixes" paragraph. Input most welcome.
[14:14:55] <wiesand> I think we should get 1.8.3pre1 out first. One of my mistakes leading to the delays was to attempt getting 1.6.24pre1 out at the same time. And then it was always to big to actually start :-(
[14:16:20] <kaduk@jabber.openafs.org/barnowl> ==mvita, thank you for doing the NEWS
[14:16:56] <wiesand> Would anyone be able & willing to smoke test the 1.8.3pre1 tarballs/srpm before we announce?
[14:17:32] <mvita> sure!
[14:17:57] <kaduk@jabber.openafs.org/barnowl> I suppose I could probably do that.
[14:18:38] <wiesand> Thanks. For 1.6.x I'm confident enough in my own smoke testing, but 1.8 is still a bit weak here.
[14:19:50] <mvita> I've been smoke testing rxgk-phase1
[14:20:09] <kaduk@jabber.openafs.org/barnowl> Ooh, exciting.
[14:20:12] <wiesand> Lovely
[14:20:13] <mvita> yes
[14:20:18] <mvita> it smokes a bit
[14:21:00] <mvita> I'll update the gerrits as I have reproducible results
[14:21:29] <kaduk@jabber.openafs.org/barnowl> many thanks
[14:21:54] <mvita> meffie:  I added preliminary rxgk support to afsutil
[14:24:08] <meffie> ok, great. (note, we are rapidly moving to ansble for setup stuff)
[14:25:12] <mvita> QUIT MOVING TEH GOALPOSTS
[14:25:17] <mvita> ;-)
[14:25:49] <meffie> heh, this was by popular demand.
[14:26:07] <mvita> I guess I'm not popular then.
[14:26:17] <meffie> me neither.
[14:26:29] <meffie> thanks for smoke testing rxgk
[14:26:36] <wiesand> We're not going to escape the ansible hype, let's face it.
[14:26:56] <meffie> true.
[14:27:35] <mvita> Ansible hype has not yet infiltrated my impregnable Basement Fortress of Solitude
[14:28:11] <wiesand> Take a bet whether/when it will?
[14:28:25] <mvita> Spiders, sure.  But no Ansible yet.
[14:28:48] <kaduk@jabber.openafs.org/barnowl> ansible has not infiltrated my ... rather vulnerable elevated bastion
of solitude.
[14:29:04] <meffie> cheyenne has been helping with the ansible now too :)
[14:29:32] <meffie> we are using it to spin up krb5 realms + openafs cells (for testing)
[14:30:05] <meffie> centos 7 only so far, but cheyenne is a gentoo fan, so is adding that too.
[14:30:06] <wiesand> Is this on github somewhere?
[14:30:54] <meffie> yes, i've not mentioned it yet on the mail lists, but will. it's at https://github.com/openafs-contrib/ansible-openafs
[14:31:59] <meffie> it does not support 1.6.x yet. the ubik bugs to be fixed in 1.6.24 are blocking.
[14:32:28] <meffie> it's difficult to setup the databases in 1.6.x with those ubik bugs.
[14:32:47] <kaduk@jabber.openafs.org/barnowl> I can see how that would be the case, yeah.
[14:33:00] <meffie> (sorry for the ansible detour...)
[14:33:31] <meffie> i wish openafs was as easy to setup as krb5 :)
[14:33:47] <meffie> we need a kdb_util
[14:34:43] <kaduk@jabber.openafs.org/barnowl> "but openafs is as easy to setup as krb5 -- I can do it by hand
without consulting documentation"
<-- not the best person to ask
[14:35:23] <meffie> heh
[14:35:50] <kaduk@jabber.openafs.org/barnowl> (actually, I do always have to check what order the arguments for the
dafs bnode are)
[14:36:11] <meffie> me too.
[14:36:56] <meffie> (again, sorry for derailing the meeting !)
[14:37:17] <wiesand> np
[14:37:36] <wiesand> Pushed NEWS PS2, with nits fixed.
[14:37:41] <kaduk@jabber.openafs.org/barnowl> Thinking about automatic setup reminded me that 11629 is still sitting
unloved; I should push an update
[14:38:35] <meffie> oh... yes that would be helpful.
[14:38:50] <meffie> maybe for 1.8.4 :)
[14:38:52] <wiesand> (to say it once more, I do appreciate comments not only on mistakes but also on style, wording etc. - don't be shy here please)
[14:39:54] <wiesand> starting the 1.8.4 wish list right now :-)
[14:40:00] <kaduk@jabber.openafs.org/barnowl> Continuing to jump around from topic to topic, 13454 lacks the
buildbot "Verified +1" due to the solaris builder's hiccup.  But I am
failing to actually pull up a log to see what happened; does anyone
else feel comfortable applying the +1 based on their experience, the
behavior of that buildbot on other changes, and the other reports?
[14:43:01] <kaduk@jabber.openafs.org/barnowl> Stephan, have we entered the "master" section of the meeting yet or
did you have other topics for the stable branches?
[14:44:57] <wiesand> The former, unless anyone else wants to bring up "stable series" topics
[14:45:19] <wiesand> NB Solariss build failure:
[14:45:21] <wiesand> ir2hf: error: Out of memory
cc: ir2hf failed for /srv/buildbot/openafs.org/solaris11_x86/build/src/rx/rx.c
*** Error code 1
make: Fatal error: Command failed for target `rx.lo'
Current working directory /export/home/buildbot/openafs.org/solaris11_x86/build/src/libuafs
*** Error code 1
make: Fatal error: Command failed for target `libuafs'
Current working directory /export/home/buildbot/openafs.org/solaris11_x86/build
*** Error code 1
make: Fatal error: Command failed for target `build'
Current working directory /export/home/buildbot/openafs.org/solaris11_x86/build
*** Error code 1
make: Fatal error: Command failed for target `all'
program finished with exit code 1
elapsedTime=619.959820
[14:45:48] <kaduk@jabber.openafs.org/barnowl> Mike, is 13456 to be abandoned, given Andrew's comments?
(I perhaps merged a similar change recently without thinking too hard
:( _
[14:46:13] <mvita> re: 13454 looking
[14:46:39] <meffie> regarding 13454.
[14:46:52] <meffie> this was my fault.
[14:48:25] <meffie> i had turned off --enable-checking on the builders when we were trying to get things working, and then 2 warngings slipped thro to master when the 'pthreaded-xstats' was merged.
[14:49:05] <kaduk@jabber.openafs.org/barnowl> Ah, quite understandable.
[14:49:05] <mvita> could be the invalid —xregs=float ?  still looking
[14:49:13] <meffie> then on wed i turned up --enable-checking in too many places, including solaris, over zeals
[14:49:30] <meffie> over zealously
[14:49:43] <mvita> sorry, gotta step away for a bit… brb
[14:50:21] <meffie> now; we have --enable-checking *only* on recent linux builders. ubuntu 1804, debian 9, and the most recent fedora.
[14:51:09] <meffie> since those are the only ones with recent enough gcc to actually find warnings.
[14:51:56] <kaduk@jabber.openafs.org/barnowl> It looks like I can choose between:
kaduk@prolepsis:~$ gcc --version
gcc (Debian 8.2.0-14) 8.2.0
and
kaduk@prolepsis:~$ clang --version
clang version 7.0.1-4 (tags/RELEASE_701/final)
[14:52:12] <meffie> also, for some reason the sol11 builder is failing intermittently with "memory" errors, which may mean it's just low on disk space ? i'll check, since that one is an SNA builder.
[14:52:25] <kaduk@jabber.openafs.org/barnowl> I think my builds tend to default to gcc on this machine, though.
Which is maybe unfortunate, as I at least used to like clang's
warnings better.
[14:53:25] <meffie> i can add a deb builder with clang, or maybe cheyenne can add a gentoo builder with clang :) gentoo tends to be bleeding edge.
[14:53:59] <kaduk@jabber.openafs.org/barnowl> That would be nice, if Cheyenne is willing
[14:55:38] <kaduk@jabber.openafs.org/barnowl> In other master news, I  left some comments on the solaris kstat
change
[14:55:44] <meffie> i'll rebase 13454 to force a rebuild, to check the solaris builder.
[14:56:02] <kaduk@jabber.openafs.org/barnowl> I think I was also supposed to look at 13429/13430/13431; is that
right?
[14:57:26] <meffie> yes
[14:57:51] <kaduk@jabber.openafs.org/barnowl> Okay.  I guess I will probably get some quality review time in, in
between superb owl ads.
[14:59:06] <kaduk@jabber.openafs.org/barnowl> I see that Mike resuscitated pthread-bozo as well (thank you!).
I'll go through and remove my -2s if nothing else...
[14:59:40] <kaduk@jabber.openafs.org/barnowl> And I feel bad that rxgk-stage1 is not getting more love from me
either.
[14:59:42] <meffie> yes, thanks for merging the pthreaded-xstats, on to the next one. :)
[14:59:55] <kaduk@jabber.openafs.org/barnowl> But, in addition to all that, are there  other things I should try to
focus on?
[15:00:23] <meffie> it would be nice to have rxgk-phase1 on a master branch soon, so i have something to say at hepix 2019 :)
[15:01:01] <kaduk@jabber.openafs.org/barnowl> Well, we have some time to work on that, I  guess.
[15:01:30] <kaduk@jabber.openafs.org/barnowl> Is the rxgk-phase1 smoke any particular flavor that you can tell?
[15:01:49] <meffie> well, it's progress.
[15:03:47] <meffie> the pthread bosserver stack is largish :(
[15:04:09] <kaduk@jabber.openafs.org/barnowl> It's quite big, yes.  bozo has not really received much in the way of
regular maintenance
[15:05:19] <kaduk@jabber.openafs.org/barnowl> Any other topics for this meeting?
[15:06:25] <meffie> nope, oops, it's 5 after
[15:06:35] <kaduk@jabber.openafs.org/barnowl> Yeah, I was just noticing that we'd passed the hour
[15:09:44] <wiesand> Fine, let's adjourn. Thanks a lot everybody! And stay tuned for my request to smoke test preliminary 1.8.3pre1 tarballs and have a look at the draft announcement etc. (ETA "early next week).
[15:09:55] <kaduk@jabber.openafs.org/barnowl> Yes, thanks everybody!
[15:10:13] <kaduk@jabber.openafs.org/barnowl> Especially Stephan for getting us closer to 1.8.3pre1 -- that NEWS
file had lots of good stuff in it
[15:10:25] <wiesand> :)
[15:11:23] wiesand leaves the room
[15:16:25] wiesand joins the room
[15:20:19] <wiesand> hmm, just looking at the buildbot verdict on the NEWS change (which can hardly be the culprit): solaris11_x86 failure looks like there's something wrong with the vm again, but the ubuntu1804-amd64 failure might be a worry?
[15:21:02] <meffie> looking at the sol11 box, maybe it's low on memory?
[15:22:45] <kaduk@jabber.openafs.org/barnowl> That ubuntu one looks familiar (Wformat-truncation), and we've fixed a
bunch of those "recently" on master.
[15:23:05] <kaduk@jabber.openafs.org/barnowl> So maybe pull up https://gerrit.openafs.org/12813
[15:24:29] <meffie> ah, yes, that old thorn!!
[15:24:30] <kaduk@jabber.openafs.org/barnowl> Also, I wonder if my buildbot woes are because I secretly need to
change apache to reverse-proxy https://buildbot.openafs.org to port
8010 instead of ... whatever the old one used to be on.  Or something.
[15:26:54] <meffie> ah, i've been meaning to ask. i do not have any visibility into the apache setup, and as far as i can see the buildbot config is correct.
[15:27:30] <meffie> so, i'm stuck getting the https://buildbot.openafs.org working.
[15:27:39] <kaduk@jabber.openafs.org/barnowl> (It looks like if I allow javascript over the http-not-s port 8010
connection, from a different machine, things look fine.  But  I have
this allergy to javascript over http-not-s...)
[15:30:06] <kaduk@jabber.openafs.org/barnowl> Well, buildbot-ssl.conf looks sane I guess:
<VirtualHost *:443>
    ServerName buildbot.openafs.org
    ServerAdmin openafs-gatekeepers@openafs.org
    
    SSLEngine On
    
    SSLCertificateFile /etc/letsencrypt/live/git.openafs.org/cert.pem
    SSLCertificateKeyFile
/etc/letsencrypt/live/git.openafs.org/privkey.pem
    
    SSLSessionCacheTimeout 28800
    SSLOptions +StdEnvVars
    ErrorLog /var/log/httpd/buildbot-ssl-error_log
    CustomLog /var/log/httpd/buildbot-ssl-access_log combined
    ProxyRequests Off
    ProxyVia Off
    ProxyPreserveHost On
    <Proxy *>
        Order deny,allow
        Allow from all
    </Proxy>
        ProxyPass / http://127.0.0.1:8010/
SSLCertificateChainFile
/etc/letsencrypt/live/git.openafs.org/chain.pem
[15:30:40] <meffie> http://docs.buildbot.net/current/manual/configuration/www.html
2.5.13.4. Reverse Proxy Configuration
[15:31:23] <kaduk@jabber.openafs.org/barnowl> Oh, I have to proxy websockets, I see.
[15:32:49] <wiesand> proxying websockets is fun :-/
[15:34:11] <wiesand> about to pull up 12813 - thanks for the hint
[15:35:48] <meffie> excellent! thank you wiesand
[15:36:12] <meffie> and thank you ben for taking a look at the apache config.
[15:37:42] <wiesand> er, looks like 12813 is already present on 1.8.x (as 12897)
[15:38:16] <kaduk@jabber.openafs.org/barnowl> uh-oh
[15:40:36] <wiesand> sorry, I have to go paint another wall
[15:40:42] wiesand leaves the room
[15:41:35] meffie looking at the ubuntu 1804 log...
[15:42:41] <meffie> vutil.c: In function ‘VWalkVolumeHeaders’:
vutil.c:860:34: error: ‘%s’ directive output may be truncated writing up to 255 bytes into a region of size 63 [-Werror=format-truncation=]
      snprintf(name, VMAXPATHLEN, "%s" OS_DIRSEP "%s", partpath, dentry->d_name);
                                  ^~~~
[15:45:35] meffie looking.
[15:46:42] <kaduk@jabber.openafs.org/barnowl> Hmm, looks like I will have to invoke server operations to figure out
if there's an easy way to do this.  So maybe no buildbot https for
today.
[15:57:52] <meffie> ok, thanks for looking into this. i've been meaning to ask for help.
[15:58:10] <kaduk@jabber.openafs.org/barnowl> Happy to help
[16:01:52] <meffie> i can reproduce stephan's build error on 1.8.x with --enable-checking
[16:04:42] <meffie> can confirm, we already have:
12897 4988628a2e Avoid gcc warning
[16:07:26] <meffie> 1.8.x is missing:
13274 2daa413e3e Avoid format truncation warnings
[16:11:15] <meffie> cherry-picked, local test build in progress
[16:16:00] <meffie> Change 13459 pushed to 1.8.x (master version is 13274)
[22:02:46] meffie leaves the room