- Wednesday, May 13, 2015

release-team@conference.openafs.org

Wednesday, May 13, 2015< ^ >

Room Configuration

Room Occupants

GMT+0
[13:50:02] Marc Dionne joins the room
[13:59:50] wiesand joins the room
[14:00:10] <wiesand> Hello
[14:00:20] <Marc Dionne> hi Stephan
[14:01:17] <wiesand> Found your linux commit: de71ad2 :)
[14:02:00] <wiesand> Are there any new obstacles regarding linux?
[14:02:28] <Marc Dionne> no, everything looks ok for 4.1, with the commits in gerrit
[14:03:03] <wiesand> Great. So “all” it takes is to get those merged...
[14:03:25] <wiesand> 4.1 ETA is in ~4 weeks?
[14:03:58] <Marc Dionne> roughly, we're at rc3, and 7-8 RCs is typical
[14:04:36] kaduk joins the room
[14:04:54] <wiesand> Time to get out 1.6.12pre2...
[14:05:11] <Marc Dionne> some more vfs churn probably to come for 4.2, but i haven't looked at the details yet
[14:05:30] <wiesand> Oh well. Thanks for the heads up.
[14:07:29] <wiesand> I guess there’s not much more to say about 1.6.12. Those changes listed in the agenda (and I’ll copy the list into the minutes) need review and to be merged.
[14:07:40] <Jeffrey Altman> yes
[14:07:54] <wiesand> So, on to “1.8” ?
[14:08:02] <wiesand> (Hi Ben ;-)
[14:08:57] meffie joins the room
[14:09:25] <kaduk> Hi :)
[14:09:56] <kaduk> I was trying to push a new revision of 6947 before we got to this point of the meeting, but I guess I'm too slow
[14:10:56] <kaduk> I think we can go forward with not handling FPE and PIPE via softsig, and take some variant of Mike's bosserver patch.
[14:11:19] <kaduk> I didn't get much done last week since I was busy getting married.
[14:11:30] <meffie> CONGRATS!
[14:11:36] <Marc Dionne> woo.  congratulations
[14:11:40] <wiesand> congratulations!
[14:12:11] <wiesand> (and yes, it’s a reasonable excuse for not getting much done ;-)
[14:12:19] <kaduk> Thanks, all!
[14:13:05] <Jeffrey Altman> congrats!!!
[14:13:17] <wiesand> so, next SP for 6947 coming... anything else?
[14:14:08] <meffie> we should now have a beer in Ben and spouse's honor?
[14:14:18] <kaduk> We should look at 11865 and maybe Jeff's alternate suggestion in a comment on it.
[14:14:37] <kaduk> (And maybe start converting things to use opr's softsig as well.)
[14:14:39] <wiesand> Mike: sounds good, and here’s the right time of day for that...
[14:15:08] <Jeffrey Altman> no beer (or tequila shots) for me.  I have to drive to the airport shortly
[14:15:42] <meffie> sorry i did not have time to push the alternative suggested by Jeff. i can do that tonight.
[14:16:36] <Jeffrey Altman> the reason for my comment on 11865 is to preserve the existing behavior which is at present a one step process to disable the restricted mode.  Send a signal and then restricted mode is disabled.   There is no need to issue a bos rpc as a second step.
[14:16:40] <wiesand> Will that change BosConfig too?
[14:16:48] <meffie> yes
[14:17:19] <meffie> not, tested yet tho.
[14:17:24] <kaduk> I am not sure that checking for a file's existence "all the time" is a great design, either, but it's not like we are writing something from scratch and can make it perfect
[14:17:50] <Jeffrey Altman> Its not all the time, its only when restricted mode is on but yeah.
[14:18:43] <Jeffrey Altman> My preference would be an ipc that can be guaranteed to come from the same host and can authenticate the identity of the caller as root
[14:20:00] <wiesand> Should that actually be root, or the EUID of the running bosserver process?
[14:20:02] <kaduk> mumble named pipe mumble
[14:20:36] <meffie> heh
[14:20:50] <Jeffrey Altman> The EUID of the running bosserver process is correct but for openafs that is always root
[14:21:22] <wiesand> One can dream...
[14:23:09] <meffie> hrm.. ok, i have some ideas.
[14:24:59] <meffie> can we assume the bos client is on the server?
[14:25:20] <Jeffrey Altman> unfortunately not
[14:25:24] <meffie> rats
[14:25:41] <Jeffrey Altman> anything that uses network sockets can be spoofed
[14:26:07] <Jeffrey Altman> It needs to be an out of band signal.   The file system provides that.
[14:26:14] <meffie> well, i was thinking bos could use a non-network ipc for '-mode 0'
[14:27:01] <Jeffrey Altman> The benefit of my suggestion is that it works and doesn't significantly change the existing behavior.     We could provide a command that creates the file for the user and at some point in the future replace the use of the file with something else
[14:27:28] <Jeffrey Altman> If its an ipc then the command executing it must be on the same machine
[14:27:39] <Jeffrey Altman> well, depending on the ipc
[14:28:57] <meffie> ok, at least we are in agreement the SIGFPE should go, and the softsig patch is ok wrt that signal, correct?
[14:30:03] <Jeffrey Altman> I think it is ok for FPE to be ignored by softsig provided it isn't used by us
[14:31:33] <meffie> ok, thank you
[14:32:30] <Jeffrey Altman> time to head to the airport
[14:32:58] <wiesand> Have a safe trip.
[14:33:04] <Jeffrey Altman> once again congrats to Ben.  I will have a drink on the plane
[14:33:14] <kaduk> Thanks!
[14:33:23] <kaduk> Have a safe trip
[14:33:43] <wiesand> Anything else to discuss today?
[14:35:09] <kaduk> I don't think we have anything
[14:35:36] <wiesand> Fine. Thanks a lot everyone!
[14:35:45] <wiesand> Bye
[14:35:47] wiesand leaves the room
[14:35:47] <meffie> thanks, and congrats
[14:36:22] Marc Dionne leaves the room
[15:03:04] sorry, i was off applying for a loan
[15:05:13] <kaduk> Oh, man, is Daria sending as admin messages again???
[15:05:13] <kaduk> hmm, but I can't send, now?
[15:06:07] i am? barnowl never shows me my own messages
[15:06:15] oh, now it does, and i am
[15:06:24] the answer is kick the jabberd
[15:07:54] done
[15:08:31] shadow@gmail.com/barnowl0396B1CB joins the room
[15:09:02] <shadow@gmail.com/barnowl0396B1CB> hi?
[15:09:06] <shadow@gmail.com/barnowl0396B1CB> ok, back
[15:12:38] meffie joins the room
[15:12:45] <meffie> hi
[15:34:47] meffie leaves the room
[16:04:16] kaduk joins the room
[16:05:05] <kaduk> Ah, I was going to hold off on bouncing the jabberd until I got a functional jabber client back, but that works, too.
[20:54:07] kaduk leaves the room