[00:08:34] --- haba has become available [02:00:50] --- Russ has left: Disconnected [02:23:29] --- phalenor has left [02:27:24] --- phalenor has become available [04:05:53] --- Simon Wilkinson has left [04:38:36] --- sxw has left [05:55:59] --- jaltman/FrogsLeap has left: Disconnected [06:03:13] --- sxw has become available [06:04:35] --- phalenor has left [06:04:39] --- phalenor has become available [06:20:25] --- sxw has left [06:20:29] --- sxw has become available [06:35:52] --- sxw has left [06:35:53] --- sxw has become available [06:51:35] --- haba has left [07:11:28] --- haba has become available [07:31:10] hello, is there a time-line for the bpw mock run? will we receive login info? [07:31:47] Oh, I'd forgotten all about that. [07:31:56] heh [07:32:00] When was it supposed to be happening? [07:32:32] jeff's mail said today, but i didnt see a time or any login information. [07:33:50] Ah, I see I have also failed to give him slide decks by Sunday 5th [07:34:10] Without 8 hours on a plane, how am I supposed to find time to write my slides :) [07:36:29] --- haba has left [07:37:42] if the bpw mock run is happening today, i won't be there. but i thought it was postponed. hm. [07:37:59] --- summatusmentis has become available [08:05:25] sxw: thanks for "Use new form of cmd_AddParamWithOffset", i forgot to push mine. [08:05:48] tis fine [08:07:25] Okay, so this is rather broken. If we hit a NACK fast-restart event in RX, instead of setting the nack'd packets for immediate retransmission, we flag every other packet in the queue as if its never been transmitted. [08:12:01] --- deason/gmail has become available [08:18:04] if I have an afs vldb server that is offline for a week or more, once it comes online, does it automatically sync to the existing vldb servers or is there a command to cause it to sync up? or should I just restart its vldb server process? [08:18:52] it should sync automatically [08:20:56] I don't believe that the amount of time that it has been offline should be an issue. I think it should automatically sync up. [08:20:56] You can check whether it is doing so with udebug. [08:20:57] If you find that it isn't syncing, then a) restart it, and b) send an email to openafs-bugs@openafs.org reporting it as a bug [08:24:07] 'with udebug' hmm, must find the docs on that [08:24:25] I'm not saying I know it isn't, I just didn't want to have inconsistent vldb infoz ;-) [08:24:39] 'man udebug' [08:24:51] man: no entry for udebug in the manual. [08:25:25] Hmmm. prehistoric AFS? [08:25:37] 1.4.7 [08:25:42] I have the bin just not the man [08:25:50] guess I should look at why man pages aren't installed [08:25:54] http://doc.openafs.org/Reference/1/udebug.html [08:25:55] found the man on the web [08:26:11] http://docs.openafs.org/Reference/1/udebug.html [08:28:21] amazing, I am much reassured [08:28:26] db versions are the same [08:28:36] excellent [08:29:05] hopefully soon I'll get around to upgrading the openbsd port to something more recent. *sigh* where does the time go? [08:30:11] into the past [09:56:57] --- rra has become available [10:22:02] --- jaltman/FrogsLeap has become available [10:38:09] --- sxw has left [11:04:14] Any thoughts on my kdump heisenbug? [11:28:42] --- jaltman/FrogsLeap has left: Disconnected [11:53:09] --- steven.jenkins has left [11:54:24] --- steven.jenkins has become available [11:54:53] --- meffie has left [12:04:28] --- meffie has become available [12:33:54] --- jaltman/FrogsLeap has become available [15:08:05] --- Simon has become available [15:20:04] --- Simon has left [15:26:04] --- deason/gmail has left [16:17:51] So, uh, I have a file here that was written from a pre4 client and read from a "pre1" (+ patches; linerva) client ... that starts with a block of NULs and shouldn't. [16:18:34] Oops, "had", now. [17:07:07] --- Simon Wilkinson has become available [17:07:31] kaduk: On which platform? [17:08:01] And I don't suppose you know how many NULLs? cachesize? chunksize? [17:08:08] pagesize? [17:09:13] And is the file corrupt on the server, or just on the client that read in back in? [18:12:09] --- ktdreyer has become available [18:27:51] --- rra has left: Disconnected [18:37:45] 'od -c' went from line 0000000 to 0010000, all NULs. I don't think I actually managed to check from a different client; I attempted to, then got distracted, and then overwrite the file (.xsession-errors) by accident. Both clients linux; I'm not sure what OS the server (phlegethon.mit.edu) is. [18:46:41] --- Russ has become available [19:58:29] --- jaltman/FrogsLeap has left: Disconnected [20:01:44] --- jaltman/FrogsLeap has become available [22:53:34] --- lars.malinowsky has left [23:04:57] --- reuteras has become available [23:20:24] Assuming that's octal (which I think it is from od), that's a page of NULs. Which is very boring, because it suggests there's a problem in the page copy logic someplace. [23:22:59] What would need to know, to start tracking this down, is whether the file is corrupt on the server or not. If it happens again, can you let me know? [23:24:10] Yup. [23:24:20] Oh, and what cache are you running - disk or memory? [23:24:32] And anything funky (like cache bypass) turned on in the client? [23:25:09] Though now that I think about it, I think we've had a user report something like this occasionally; a flushv always clears it. [23:28:17] Disk cache; no bypass or funky stuff that I know of. [23:39:28] A flushv clearing it would be consistent with the pagecopy code being at fault. [23:55:52] --- Simon Wilkinson has left