- Wednesday, August 6, 2014

openafs@conference.openafs.org

Wednesday, August 6, 2014< ^ >

Room Configuration

Room Occupants

GMT+0
[01:18:42] deason leaves the room
[02:27:32] mvita leaves the room
[05:52:01] mvita joins the room
[06:28:10] mvita leaves the room
[08:05:57] nwf joins the room
[09:09:42] wiesand joins the room
[13:43:06] mvita joins the room
[13:47:46] meffie joins the room
[13:52:17] kaduk joins the room
[13:54:16] mvita leaves the room
[13:57:23] mvita joins the room
[15:24:53] meffie leaves the room
[15:36:12] kaduk leaves the room
[15:39:44] meffie joins the room
[16:08:09] wiesand leaves the room
[16:23:28] deason joins the room
[17:14:08] meffie leaves the room
[17:14:08] meffie joins the room
[17:50:17] stephan.wiesand joins the room
[17:58:53] kaduk joins the room
[18:00:12] <kaduk> Is aklog -setpag supposed to work on RHEL6 with recent-ish 1.6.x?
[18:10:01] <deason> "supposed to". it's never surprising if it breaks, but as long as it's feasible we can try to fix it
[18:10:36] <kaduk> (I'm trying to decide whether to mention this as a possible cause of failure to the buy with the samba question on the list.)
[18:14:22] <stephan.wiesand> Hasn't it been broken for about a decade? I saved "no longer works, forget about it", but it was so long ago that I don't recall any details. I remember FNAL had a problem with it and found a workaround, but that's it.
[18:14:54] <kaduk> google found me a post from simon in ~2007 saying that there was code in 1.5 to use a linux feature to set the session keyring of the parent process.
[18:14:55] <ballbery> it varies. I think it broke when PAGs were done with groups on linux and re-fixed when we switched to keychains?
[18:17:58] <kaduk> In unrelated news, I'm not sure I understand meffie's comment on 11347
[18:19:01] <deason> oh yes, sorry, I didn't see that
[18:19:09] <meffie> well, the afs_trace.h file generated is vastly different without -v 2
[18:19:29] <deason> sure, -setpag is of questionable working-ness and it's certainly a possible cause of why that guy's seeing weird behavior
[18:19:31] <kaduk> I thought -v 2 only affected the C file...
[18:20:35] <meffie> $ diff -u /tmp/afs_trace.h-{original,without-v-2} | wc -l
351
[18:20:43] <kaduk> main() seems to only set use_msf=1 when version==2
[18:21:23] <meffie> there's the another option that affects only the c file.
[18:22:35] <kaduk> I continue to fail to see such code in my working tree.
[18:22:53] <Jeffrey Altman> if !use_msf provides a extern function definition instead of a CPP macro for the initialize_*_error_table()
[18:23:32] <Jeffrey Altman> see block around line 400
[18:23:53] <kaduk> Ugh, sorry.
I think there were other places in the tree that would be affected, then.
[18:24:00] <meffie> diff -u /tmp/afs_trace.h-{original,reverted} | wc -l
0
[18:24:51] <meffie> i'm pretty sure -v 2 is only used here.
[18:25:24] <meffie> no worries, i've been burned by this before.
[18:25:24] stephan.wiesand leaves the room
[18:25:45] <Jeffrey Altman> whether -v 2 should be used there is a different question but the parameters to generate the .c and .h files should be the same. Leave it up to compile_et to decide what applies and what doesn't
[18:26:21] stephan.wiesand joins the room
[18:27:51] <meffie> that does seem to be safer.
[18:35:56] <Jeffrey Altman> http://gerrit.openafs.org/#change,11360 is the revert commit for master
[18:36:56] <kaduk> Thanks.
[18:37:37] <stephan.wiesand> And I abandoned 11359. Sorry.
[18:38:06] <kaduk> Hmm, I didn't get mail about the new change for review.
[18:38:52] <Jeffrey Altman> perhaps Gerrit doesn't send one when the revert is generated by Gerrit
[18:39:42] <kaduk> Oh, that could be it.
[19:00:30] stephan.wiesand leaves the room
[21:23:14] mvita leaves the room
[21:24:45] mvita joins the room
[22:40:58] meffie leaves the room
[22:41:58] Jeffrey Altman leaves the room
[22:42:16] Jeffrey Altman joins the room
[23:09:48] kaduk leaves the room