[01:26:28] --- cclausen has left [03:49:17] --- dev-zero@jabber.org has become available [03:49:19] --- dev-zero@jabber.org has left: offline [05:16:34] --- cclausen has become available [05:30:30] --- sxw has become available [05:30:44] --- Jeffrey Altman has left: Replaced by new connection [05:30:46] 1.4.11pre2 is broken for UKERNEL on Linux. Dull. [05:33:06] i didn't announce it, so that's kind of hard. [05:33:46] if you grab the new tarballs in afs they should work [05:34:05] New since this morning? [05:36:46] --- sxw has left [05:37:01] new in the last 38 minutes. [05:59:04] --- mmeffie has become available [06:17:10] --- Jeffrey Altman has become available [06:20:56] you can build Linux and Windows in the same tree. The build processes are independent [06:44:25] --- dev-zero@jabber.org has become available [07:09:15] --- deason has become available [07:34:24] --- Simon Wilkinson has left [07:35:56] the Win XP hot fix http://support.microsoft.com/kb/971421 is now published [07:59:23] --- sxw has become available [08:05:44] > that certainly did not use to be the case No, it's pretty much always been the case. Before Microsoft, I think the best support we got was from HP and Sun; note that we are in direct competition with one of the latter's products [08:07:55] > What you _should_ be able to do is to be in a different path, and do > ../../path/to/afs/configure Which, as you point out, is broken. But what _does_ work is creating a shadow tree of symlinks and building in that. [08:08:39] > Sadly, the changes to add versioning information to the XML > documentation broke this. It was broken for a long time. If it was fixed recently, before the XML thing, it's news to me. [08:09:01] Is it just the version.xml stuff that's breaking objdir builds, or is there other, older, breakage? [08:10:15] It's possible it was fixed and I didn't notice, but it was broken long before the XML thing. I don't recall why; all I know offhand is that this is one of the few packages I build with a symlink tree instead of a proper objdir build. [08:12:14] BTW... > this is a stupid question The only stupid question is one to which you already know the answer. Except in a courtroom, where it's the other way around. [08:12:57] --- tkeiser has become available [08:14:27] --- deason has left [08:14:58] --- deason has become available [08:18:48] --- stevenjenkins has left [08:22:18] --- stevenjenkins has become available [09:09:42] --- sxw has left [09:12:49] --- tkeiser has left [09:33:25] --- Russ has become available [10:01:03] --- dev-zero@jabber.org has left [10:26:48] --- dev-zero@jabber.org has become available [10:26:52] --- dev-zero@jabber.org has left: offline [10:30:09] --- sxw has become available [11:56:54] --- dev-zero@jabber.org has become available [11:56:57] --- dev-zero@jabber.org has left: offline [12:00:46] --- cg2v has become available [12:02:47] --- cg2v has left [12:35:53] --- sxw has left [13:26:38] --- dev-zero@jabber.org has become available [13:26:38] --- dev-zero@jabber.org has left: offline [13:42:48] feedback, changes to http://www.dementia.org/twiki/bin/view/AFSLore/DemandAttach welcomed. [13:47:17] That's an edited version of an internal document passed to me... [13:47:52] some of the options are confusing to me [13:48:05] does "+int" means that it takes a positive integer? [13:48:26] and the hard limits for the commands enforced by the code should be listed on the document [13:52:27] tx; I see some other issues there [13:55:03] changed. [13:55:22] I'm a bit hesitant to document that on the wiki -- the man page is the right place for that. [13:55:28] ie, hard limits. [13:55:46] ok [13:55:53] but it's a wiki, so feel free to disagree and put them in. :) [13:55:59] I'd say that recomended values should proably be the defaults [13:56:06] but I guess that isn't up to me [13:56:28] I'm not 100% sure how to express that column..it's what a certain DAFS user runs.. [13:56:54] oh, I see [13:57:01] so its not a general recomendation [13:57:11] its a site specific optimization [13:57:56] I'd say making vlruthresh 25 hours by default would be good [13:58:11] wouldn't needlessly detach volumes that are only touched daily [13:59:53] would you want volumes only touched once per day to be detached or not? I don't know. [14:00:09] as in 'I don't know if there is a default that is reasonable'. [14:00:14] yeah, I guess that is true [14:00:33] reasons WHY the defaults were changed would be useful to know [14:00:38] if such data is available [14:00:39] * stevenjenkins nods. [14:01:09] I think having people test DAFS and work out usable defaults/cases would be helpful. [14:01:25] that's part of my offer (in openafs-devel) to help whomever wants to play with it. [14:01:40] hmm, "volume headers read" in the DAFS startup sequence seems slightly misleading to me [14:01:55] once we have a few cases, that column could get dropped from that table, then another section added which is 'common settings for common cases' [14:02:09] in what way? [14:02:18] you mean, like, we look for volume headers, right? [14:02:23] it doesn't read each individual one [14:02:33] (unless I am confused; which is possible) [14:02:48] having debian experiemtnal packages would help me [14:02:59] (or even better, Ubuntu 8.04 LTS packages so I didn't need to recompile) [14:03:58] oh. I didn't realize we don't produce pkgs for 1.5 except on windows & mac osx. [14:04:09] yeah [14:04:20] Windows packages work great! [14:05:11] "There is therefore a hard-limit on the number of times a volume will be salvaged which is reset when the volume is removed or the file-server is restarted." did you mean "moved"? in place of "removed"? [14:05:27] I mean... technically that's true, but the volume is rather... gone, at that point [14:06:01] deason - not sure. what you say sounds right, but I'll need to look at the code to see. [14:14:28] hm. probably easier to look at it empirically. [14:28:01] --- cclausen has left [14:36:41] --- brantgurga has become available [14:51:48] --- mmeffie has left [15:15:29] --- cclausen has become available [15:21:17] --- brantgurga has left [15:43:13] --- deason has left [15:44:03] --- brantgurga has become available [17:57:10] --- summatusmentis has left [18:21:52] --- Russ has left: Disconnected [18:41:02] --- rra has become available [18:48:43] --- summatusmentis has become available [18:49:46] --- mdionne has become available [18:54:56] any way for an average user to comment on tickets in RT? or do you need a real account for that? [18:55:28] Mail sent with the right magic subject line will be appended to the ticket. [18:55:44] I think that's true no matter who sends it. [18:55:53] I think I've tried that before with no success [18:56:23] Hm. Maybe it checks the from line. [18:57:15] I'll give it another try [19:07:10] it doesn't show up in the ticket, and there's no response to the mail either. maybe there's an approval queue somewhere.. It would be a useful feature. [19:09:30] Yeah... that's rather annoying. [19:15:03] so the comment I wanted to make was about RT#124991 - I didn't intend for that pid_task() bit to be part of the RCU patch, it was a quick fix so I could do test builds. As Simon point out in the ticket it's harmless and allows us to build on current 2.6.31-rc, so it's not all bad. just begs for something cleaner. [19:22:38] --- mdionne has left [19:51:21] is src/afs/afs_osi_gcpags.c broken? [19:51:44] from a current cvs pull [20:00:49] "broken" here meaning doesn't build, specifically: [20:00:50] /Users/jts/openafs/src/afs/afs_osi_gcpags.c:552:2: error: #endif without #if [20:56:36] --- deason has become available [20:58:32] mdionne: the email address you send from has to be part of the ticket [20:59:05] being a requestor works at least; probably some other fields, as well [21:06:19] --- Jeffrey Altman has left [21:14:19] --- deason has left [21:25:27] --- Jeffrey Altman has become available [21:25:42] --- Jeffrey Altman has left [22:04:16] --- Jeffrey Altman has become available [22:46:13] --- dev-zero@jabber.org has become available [23:26:34] --- rra has left: Disconnected