[01:01:15] --- Russ has left: Disconnected [01:21:47] --- abo has left [01:21:59] --- abo has become available [01:49:35] --- sxw has left [01:51:10] --- Simon Wilkinson has left [05:29:52] --- Jeffrey Altman has left: Replaced by new connection [05:30:37] --- Jeffrey Altman has become available [06:00:02] --- reuteras has left [06:27:59] --- cclausen has left [07:03:44] --- deason has become available [07:45:46] --- matt has become available [07:46:19] Interesting discussion about timelines for afs3-stds. However, it is not echoed to the list, and that is a problem. Tkeiser finds 30 days insufficient, as well. [07:46:30] He hasn't sent that to the list either. [07:49:07] dafs fileservers broken: I was under the impression that bugfixes had been submitted on several tickets? [07:49:49] well, there's still other tickets [07:49:54] "broken" would be subjective [07:50:28] (quoting) You see, I suspect that in my world, dafs wouldn't even be on 1.5 yet. [07:51:07] All judgments at that level are going to be subjective, but sure. [07:51:54] But in fact there was a list of critical fixes for DAFS. [07:52:31] the real problem with DAFS is that we do not know what is still broken [07:52:54] how does it break? [07:53:38] (I agree I do not. I believe tkeiser, sjenkins, adeason, mmeffie have a much clearer idea.) [07:54:03] there is a list of known issues in RT. I'm more concerned about the unknown issues. [07:54:39] --- stevenjenkins has left [07:54:40] over the last year several orgs that have volunteered to test 1.5 deployments (either with or without dafs) have given up because there were just too many issues. [07:55:45] there have been failures with new cell deployments, with mixed deployments of 1.5-dafs and 1.4.x, and 1.5+dafs [07:56:06] so, more a 1.5 issue than a dafs issue [07:56:29] the changes in 1.5 are dafs [07:56:44] k [07:56:55] dafs alters the structure of the code in significant ways even if it is not being used. [07:57:58] I think looking for reasons to push DAFS back might be unwise. I think we should be looking for ways to formalize workload testing. [07:58:01] --- stevenjenkins has become available [07:58:26] no one is suggesting that dafs be pushed back. without dafs there is little justification for 1.6 [07:58:59] test? [07:59:20] ok. The biggest issue I hit regularly with DAFS rw vols is asserted fixed. But that was the tip of the iceberg. Testing. [07:59:30] jeff altman - all the of known DAFS issues are already in RT [07:59:42] ie, there haven't been any 'known but private' issues for several months now. [07:59:46] the only documentation that has been provided for DAFS is doc/arch/dafs-fsa.dot which is hardly sufficient to understand what the code is supposed to be doing [08:00:11] Jeff is worried about unknown unknowns. So the issue is elevating confidence through demonstration. [08:00:15] steven: I'm not suggesting that there are issues known to SNA that are not known to OpenAFS [08:00:22] ah, ok. np then. [08:00:32] However, there are according to the tickets submitted patches that SNA has that OpenAFS does not [08:00:56] there have been a few, but those are getting straightened out and submitted. [08:01:07] time frame? [08:01:25] --- deason has left [08:01:31] some of the relevant parties are busy atm; I'll see if Andrew Deason has updates. [08:01:45] --- deason has become available [08:01:47] I know he's been getting those upstream [08:01:47] on one hand folks want 1.6 to ship soon. and yet the community's hands are tied behind its back [08:02:30] nobody is going to work on addressing issues for which there are fixes already available [08:02:55] there is a list in RT. it may or may not be complete. if there are more items broken they are not listed there [08:03:11] steven: why not post all available patches, and just note what each intends to do, and the level of validation each has or hasn't received. [08:03:30] --- abo has left [08:04:09] --- abo has become available [08:04:25] If a patch does not exist, but a design writeup does, I'd suggest post that. [08:07:32] there are some tickets for which a four line design suggestion is posted. there is no indication however of whether this design is something that is being worked on or not. [08:09:08] Ah. 'being worked on b jerry' (or not claimed) would make it more useful. [08:10:42] some of the tickets represent performance issues that are not show stoppers for a release. I put 124488 and 124489 into that category. [08:11:28] The rest however are really problems. the deadlocks between services may be an architectural issue that requires a redesign [08:11:43] However, its really hard to tell considering the total lack of documentation [08:14:25] I'm following up w/some folks now. will get an update as soon as I can. [08:14:56] would you like replies on tickets indicating they're being worked on? I could probably do that for a few [08:15:04] deason - that would be great. [08:15:21] you're probably best placed for that atm. Let me fw an email I just sent... [08:16:02] err, to be clear, I meant being worked on by me [08:16:45] Andrew: I think you can assume that at the moment no one from anyone other than SNA is working on DAFS [08:17:15] If it is not being worked on by you or Tom, it is not being addressed [08:18:43] that would still require me to find out what Tom is doing :) [08:19:58] you are probably in a better position to do so than anyone [08:20:04] (other than Tom) :-) [08:22:04] I may have accidentally volunteered myself [08:22:06] "oops" [08:27:22] > so, more a 1.5 issue than a dafs issue [08:27:30] disagree. [08:29:19] oh, owl was confused, sorry, i see yinz went further already [08:34:07] as an aside, if 1.6 is ever going to happen, dafs needs to become tested and known to be stable. i figure i should be a sucker, so my test fileserver is now running dafs. [09:23:01] --- Derrick Brashear has left [09:24:04] > Interesting discussion about timelines for afs3-stds. The timeline for afs3-stds is spelled out in the charter document. I'm still waiting for Simon to decide whether to issue a draft-4 to address Derrick's comment about checking to see whether candidates consent to be nominated. Then I'll shout from the rooftops that we want to adopt this thing and would people please come over to afs3-stds to talk about it if they have any comments. -more- [09:24:59] The discussion about possible timelines for revving some RPC's was not entirely echoed to the list, because I've been doing other work since then (well, and sleeping). I'll make my comments there soon. [09:26:49] > without dafs there is little justification for 1.6 really? I was under the impression there was a fair amount of stuff in 1.5 by now that never made it into 1.4, because the stable series is supposed to actually be stable. I can't imagine that disconnected and dafs are the only things in 1.5. [09:42:23] --- dev-zero@jabber.org has become available [09:42:28] --- dev-zero@jabber.org has left: offline [10:34:38] --- Russ has become available [10:47:27] --- Derrick Brashear has become available [10:55:27] > I can't imagine that disconnected and dafs are the only things in 1.5. [10:55:31] there's not a whole lot else [11:05:35] --- dev-zero@jabber.org has become available [11:05:39] --- dev-zero@jabber.org has left: offline [11:12:28] So, we haven't been doing any development? [11:13:10] a lot of what's there is [11:13:12] bug fixes [11:13:17] build system fixes [11:13:25] documentation [11:13:46] in, like, 3 years? [11:13:53] new OS support [11:14:03] workarounds for what linux has removed [11:14:23] hey, i'm just telling you. diff it yourself if you don't believe me [11:14:46] i could give you a cynical take, but i don't think it would be fair or helpful [11:31:45] --- Derrick Brashear has left [11:47:15] a great deal of what we have done over the last three years is code quality improvements. [11:48:36] of course, all of the Windows work (minus the afs redirector) is on the 1.5 branch. that includes all of the 64-bit support, byte range locking support, large file support, Vista/2008/Win7 support, TB cache file size support, etc. [11:49:43] Most of the rest of the improvements have been pulled to 1.4. Usually in a restricted form because 1.4 is stable and we do not want to make non-functional code changes just for the sake of code readability or prototyping. [11:50:06] cache bypass is in 1.5, connection clones -was- there [11:51:07] For example, as the prototyping work on 1.5 took place whenever we found a location where a pointer was being misused or parameters were being incorrectly passed we applied appropriate fixes to 1.4 [11:51:32] Matt, sure, but those items are not sufficient justification from a functionality perspective to issue 1.6 [11:51:39] (sure) [11:51:40] and disconnected is not ready for general use [11:51:59] so if DAFS is not included there is little justification for issuing 1.6 at all. [11:52:00] I hope that an rxk5 protocol document appears shortly, rxk5 development has tracked 1.5 (not trying to start more talk about rxk5 until that protocol doc appears) [11:56:16] now I want to have 1.5 become stable so that Windows has a development series once again [11:58:24] does cache bypass work on anything other than Linux? [12:19:22] not yet. there is solaris code that would shorten time to add that. support for freebsd should be manageable, but doesn't exist yet [12:20:21] (i.e., UM cache bypass supported solaris and Irix, but the behavior was quite different, and there's some extra readahead logic to add) [12:33:27] i'm trying to help a user with vista (when i don't have access to a vista system myself): how do you configure the default cell? i'm reading the release notes from the site, and don't find anything on configuring the default cell if you download the exe installer. the user reports that "going to Advanced -> Configure AFS Client -> General -> Cell Name and changing the Cell Name from "openafs.org" to "umich.edu"" is giving her an error that she doesn't have authorization to do this. she c [12:34:10] she might have have authorization. you need to be "Administrator" to make changes to that area of the registry [12:34:20] she might *not* ... [12:35:12] the default cell can be set during the installation process or changed via the HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters key "Cell" value [12:35:22] see Appendix A of the release notes [12:35:41] when you say 'you need to be "Administrator"', does that mean "you have to be logged in as the account "Administrator"', or something else? [12:36:13] yes, you need to be logged in as the Administrator (or a member of the Administrators group) or start the process using "Run as Administrator" [12:36:24] i've touched vista literally something like three minutes in my life. i'd rather not try to take the user through editing the registry if at all possible.... [12:36:32] okay, i'll pass that along to her as a suggestion. [12:36:46] --- deason has left [12:36:55] thanks. [12:37:00] --- deason has become available [12:37:02] --- stevenjenkins has left [12:38:39] --- deason has left [12:39:37] Brant is working on a better tool. the "afs_config.exe" tool simply doesn't work well on Vista. Not to say that it worked well on anything else. [12:40:35] --- deason has become available [12:40:55] --- stevenjenkins has become available [12:41:21] yeah. and it's a shame that locally we have no offical support for installing the openafs client, and no real windows folks in our group. [12:41:55] to clarify --- i presume that if you went the "Run As Administrator" route, what you'd want to run is afs_config.exe. true? [12:42:02] yes [12:42:08] good. thanks. [12:53:30] --- Derrick Brashear has become available [13:35:12] --- asedeno has left [13:35:31] --- kaduk@mit.edu/owl has left [13:37:31] --- matt has left [14:07:07] --- cclausen has become available [14:08:58] --- asedeno has become available [16:29:26] --- stevenjenkins has left [17:43:09] --- edgester has become available [18:33:08] --- deason has left [18:46:20] --- edgester has left [19:11:32] --- cclausen has left [19:30:03] --- deason has become available [21:39:04] --- matt has become available [21:39:12] --- matt has left [22:08:43] --- kaduk@mit.edu/owl has become available [22:13:47] --- deason has left [23:24:19] --- reuteras has become available [23:30:50] --- dev-zero@jabber.org has become available [23:30:55] --- dev-zero@jabber.org has left: offline