[00:03:30] --- reuteras has become available [00:48:50] --- dev-zero@jabber.org has become available [00:48:54] --- dev-zero@jabber.org has left: offline [01:32:10] --- Simon Wilkinson has become available [01:33:56] I'm back ... [05:20:47] --- SecureEndpoints has left: Replaced by new connection [05:34:33] --- dwbotsch has left [05:36:01] --- dwbotsch has become available [05:38:07] --- SecureEndpoints has become available [05:38:25] how was life without wifi? [05:38:52] Interesting. I'm not enjoying the re-entry. 18,000 or so emails. [05:39:16] just discard them [05:39:21] can't be anything important [05:39:22] In the process of doing so :) [05:39:36] Anything interesting happen while I was away? [05:39:43] not really [05:39:46] Besides the continuing collapse of the world economy, that is. [05:39:53] I got a puppy [05:40:12] Cool. I think you said you were going to just before I left. [05:40:16] he is sleeping next to me at the moment [05:40:25] It's a slippery slope, you know... [05:40:28] he came home on saturday [05:40:49] Cool. Unplugged any machines for you, yet? [05:40:56] nope [05:41:25] he has been really good [05:41:33] Excellent. [05:42:03] crate training is going well [05:42:40] How old is he? [05:42:54] about 10 weeks. Jan 1 is his birthday [05:43:29] That's handily easy to remember ... [05:43:52] july 4th weekend will be his first show [05:44:00] --- stevenjenkins has left [05:44:03] the breeder has high hopes for him [05:44:08] --- reuteras has left [05:44:34] --- reuteras has become available [05:44:48] Ah, he's pedigree then? [05:44:53] yes [05:45:20] he decends from a line of champion beagles from Finland and Poland [05:45:43] Cool. [05:45:49] --- abo has left [05:45:57] he has dumbo ears [05:45:57] --- abo has become available [05:46:58] --- stevenjenkins has become available [05:49:24] ok, i guess i remembered that simon was coming back today, but, hi simon! [05:50:00] Hi Derrick! [05:50:15] how was the trip? [05:51:05] Pretty amazing. [05:51:38] any notable highlights? [05:51:46] I think we were both glad that Singapore was our last stop, tourism fatigue was starting to set in, but the whole thing was really good. A great set of places, and nothing that similar to anything else. [05:52:18] any restaurants Derrick and I should visit? [05:52:46] Thailand is amazing. Milford Sound in NZ is stunningly beautiful, and standing under a waterfall there is an experience that can't really be described. [05:52:47] or pubs. or attractions. [05:53:02] In which country, or any of them? [05:53:21] i've heard good things about thailand. [05:53:34] name the restaurant, pub or attraction and we will go to the appropriate destination [05:53:42] Neil loved Thailand [05:53:45] Yes. Thailand is amazing. You want to get off the beaten track though. [05:53:49] and Vietnam [05:54:14] apparently in Thailand Neil had some issues with skate boarding [05:54:48] the fascists held him down? [05:54:49] The Five Islands Restaurant in Koh Samui was pretty good, as was Tongue Thaied in Bangkok. [05:55:18] what a horrible pun :-) [05:56:29] Cafe 21 in Sydney, and Indochine in Singapore also good. [05:57:36] you are making me wish i had some time and money [05:57:46] We also ate a fair bit in the food markets in Bangkok, and the hawker markets in Singapore. [05:58:55] we get our hofbrauhaus today, so having german beer is as close as i will get to travel for a while i think [05:59:02] I took about 1500 photos, some of which will be online soon. [05:59:22] The revelation of the trip was that Australian lager is actually very nice. It's just the stuff they export that sucks. [06:00:34] they keep the good stuff for themselves [06:00:47] Indeed they do. [06:00:54] "Foster's. Australian for marketing. WE wouldn't drink it..." [06:21:33] puppy on keyboard is not a good idea [06:31:03] --- Simon Wilkinson has left [06:50:21] --- Simon Wilkinson has become available [07:49:25] --- dwbotsch has left [07:49:26] --- dwbotsch has become available [07:49:29] --- dwbotsch has left [07:49:45] --- Rrrrred has become available [08:43:55] user-interface question on disconnected mode: what is wrong with saying 'fs discon -mode offline'? [08:44:34] As in "Why doesn't that work for me", or "why do we need a GSoC project when we've got that command" ? [08:44:51] why doesn't that work for me? [08:45:00] trick question: you're clearly online, so why are you trying to lie? [08:45:03] Error setting discon mode "offline" fs: Invalid argument. [08:45:21] cachemanager not compiled with disconnected? [08:45:31] That would be my guess. [08:45:35] tx. [08:45:50] thats possible..I'm trying to help someone else out, and I'm unsure how he built & installed t hings. [08:51:27] Make sure he's running the latest code, too. Otherwise things won't work so well ... [08:56:23] 1.5.57... [08:56:27] works good.. [08:57:39] Cool. Anything older than that is missing some pretty critical fixes. [08:57:59] Let me know if you find any problems... There's a big stack of things in RT I'm gradually working my way through. [08:58:32] --- reuteras has left [09:39:26] Jeff's brought this up in RT, but I haven't seen anything on the lists... [09:40:44] RXAFS_Rename() can change the FID of the file being renamed. This breaks A=open("A/file.txt"); rename("A/file.txt", "B/file.txt"); write(A, "my stuff"); [09:41:13] I suspect this is what breaks blogbench... [09:45:45] sxw - I have a crash dump I'm looking at right now. but I can't tell yet if it's disconnected or dynamic vcaches. [09:46:17] if it's disconnected, I'll let you know.. [09:46:25] ... or some evil combination of the two, I guess. [09:46:30] if it's dynamic vcaches, I'll continue strangling max.. [09:46:50] yeah, discon + dynamic vcaches works fine. but discon + *disabling* dynamic vcaches doesn't work so hot. [09:47:43] I need to look at vanilla 1.5.57 with discon enabled and re-test to get the baseline established. [09:48:03] Let me know if you find a disconnected bug. [09:48:08] (and do a bazillion things here at work..) [09:48:17] will do. I'll update here regardless. [09:48:29] I know that feeling. I'm giving a Kerberos tutorial next week that I really need to update my slides for ... [10:55:28] crash isn't your problem, sxw... [10:56:00] at least, not based on initial analysis & testing. we'll see what the next hr or so of digging into it brings - hopefully a fix. [11:54:32] --- Russ has become available [12:16:02] --- dev-zero@jabber.org has become available [12:16:07] --- dev-zero@jabber.org has left: offline [12:20:28] --- Rrrrred has left [12:21:14] Simon: the RXAFS_Rename() was originally discovered on Windows and I checked Unix to see if the same problem exists there. It does. [12:21:50] There is a related ticket reminding me to propose a revised RPC that returns the updated FIDs [12:23:06] SecureEndpoints: I think that the issue is that the fileserver can't change the FID, if we're going to have POSIX semantics. [12:23:19] If you change the FID, the client can't continue to write to a file that's open across a rename. [12:23:26] --- dwbotsch has become available [12:24:05] it can, but it's racy [12:24:40] well, it's racy and that assumes extended callbacks, even, to get to racy [12:27:41] Well, it's even messier if client A has the file open, and client B does the rename. [12:28:21] with extended callbacks, you can get that notification. otherwise i think it simply won't work [12:28:56] the problem we have today is that the FID changes, the callback is broken on the file server but not on the client. [12:29:00] --- stevenjenkins has left [12:29:01] Well, client A will get the same error as it would if the file had been unlinked, presumably. [12:29:12] My argument is that the FID shouldn't change. [12:29:18] client has no idea what the new FID is so can't patch the director or migrate the file handle [12:30:07] if there are hardlinks to the file then after the rename the new file is not the same as the old file [12:30:16] The client may still lose if the rename takes the file from a directory it can write to, to one where it can't, but not as badly as it does at present. [12:30:56] there's only so much that can be done inside AFS semantics. we should still do whatever we can [12:31:13] rename is implemented as a delete and insert. There was a recent discussion in the last week (#openafs? or -info?) asking why 'd' and 'i' are required for rename. That is why [12:31:54] a rename to a directory the client does not have insert permission to fails before the rename starts [12:32:35] I think if the file server can preserve the old FID, then it should. [12:32:46] --- stevenjenkins has become available [12:37:26] --- dwbotsch has left [12:39:03] sure but if the client can't count on the behavior it doesn't make much difference [12:40:39] --- dwbotsch has become available [12:43:53] Well, it means today's clients will be able to handle the rename of open files :) [12:44:15] they are still losing the callback and do not know it :-) [12:45:18] But, even when that bug is fixed, the client will then re-establish the callback and continue writing ... [12:48:40] --- tkeiser@sinenomine.net/owl has left [12:49:00] agreed [13:03:59] --- abo has left [13:04:47] --- abo has become available [13:34:03] --- dwbotsch has left [13:36:20] --- dwbotsch has become available [13:54:20] --- abo has left [13:54:32] --- abo has become available [14:31:11] Bouncing the conference server. Will be right back. [14:31:28] --- LOGGING STARTED [14:31:31] --- Russ has become available [14:32:21] --- dlc has become available [14:33:52] --- shadow@gmail.com/owl2775CD44 has become available [14:33:59] --- summatusmentis has become available [14:35:23] --- RedBear has become available [14:35:41] --- kula has become available [14:35:44] --- dwbotsch has become available [14:59:51] --- dwbotsch has left [15:00:07] --- Rrrrred has become available [15:37:18] --- stevenjenkins has become available [15:45:07] --- Simon Wilkinson has become available [18:15:58] --- Russ has left: Disconnected [18:32:39] --- Russ has become available [20:00:27] --- abo has become available [22:18:17] --- stevenjenkins has left [22:18:29] --- summatusmentis has left [22:18:59] --- summatusmentis has become available [22:46:33] --- stevenjenkins has become available [22:52:33] --- stevenjenkins has left [22:54:32] --- stevenjenkins has become available [23:02:03] --- stevenjenkins has left [23:20:26] --- Russ has left: Disconnected [23:28:12] --- stevenjenkins has become available [23:57:05] --- reuteras has become available