[00:38:22] --- abo has left [00:55:07] --- abo has become available [00:55:24] --- jaltman has left: Replaced by new connection [00:55:25] --- jaltman has become available [01:11:29] --- kaj has become available [01:56:57] --- haba has become available [02:06:07] --- jaltman has left: Replaced by new connection [02:06:08] --- jaltman has become available [03:45:41] Crickey, the use of secLevel in pr_Initialize is a mess. [03:51:23] Oh, everything about pr_Initialize is a bit of a mess [03:53:20] Another nice bit is that it's supposed to be capable of being called multiple times from the same process. But if you do, it leaks a ubik client every time, if you're using any secLevel other than 0. [03:54:14] I suspect we knew that. [03:54:36] I think that's part of why the fileserver got its own copy of some of that [03:54:49] The problem is that fixing it is going to need a new function name, because it's going to break out-of-tree callers. [03:57:31] (I'm busy moving all of the code where we create client and server security layers into a set of functions in auth/authcon.c so that it's easier to add new security objects) [03:58:35] Yeah, some abstraction there would be good. [04:00:08] I, however, have been up for way way too long. So, having just deployed a salvager hack that should cause the rest of my salvages to take maybe 30 minutes instead of that many hours, I am going to send email and go to sleep. [04:00:20] Ouch, still up! [04:00:24] Sleep well, then [04:00:28] well, I broke a RAID array last night. [04:00:38] Ah. Sorry. [04:00:42] fortunately, in a way that let me recover nearly everything. [04:00:52] Not RAID5, then? [04:01:37] RAID6, actually. But what I did was delete RAID partitions from a live server instead of the server I was setting up. [04:01:56] Fortunately, there was a way to recreate them and keep the data. [04:02:20] So now you're just waiting on the salvager agreeing that all of the data is good? [04:03:16] Mostly. All the relevant partitions had to be fsck'd, just in case, but they came up clean. However, the fileserver took some 500 volumes offline during the time when the LUN's weren't there. So I have to salvage those. [04:03:28] Ouch. [04:04:32] Unfortunately, it's an inode server, which means that each time you salvage a single volume, the salvager has to enumerate all the inodes on the partition. My hack takes advantage of the fact that I _know_ the volumes I'm salvaging are not attached prior to the salvage, and thus the relevant bits of inode state don't change from one run to the next. So I squirrel away the inode list and reuse it in the next run. [04:05:02] Ah, and things become much much faster. [04:05:26] Well, I don't have to salvage them. A shell script is doing that, and has been since around midnight. It should finish sooner now, though; 4-20 minutes versus a couple of seconds is going to make a big difference. [04:05:36] This is one area where namei wins hands down [04:05:39] Yeh. [04:06:06] But I should let you go sleep. It must be almost day where you are? [04:06:21] 07:00 [04:06:33] Ouch. Sleep well! [04:26:27] --- jaltman has left: Replaced by new connection [04:26:28] --- jaltman has become available [04:35:42] I shall try. [04:59:27] I gave David off-list some names to buzz for Solaris and AIX packaging help. [05:28:48] --- jaltman has left: Disconnected [05:30:31] --- jaltman has become available [05:42:49] --- jaltman has left: Disconnected [05:46:37] --- jaltman has become available [05:47:47] jaltman - bug submission to MS re blank Domain in win7 logon provider is in progress [06:56:56] --- haba has left [07:06:02] --- deason has become available [07:21:42] --- jaltman has left: Disconnected [07:37:10] --- meffie has become available [08:08:41] --- summatusmentis has become available [08:24:54] --- jaltman has become available [08:30:30] great. once you have a PSS Incident number please open a ticket in the openafs-bugs rt with it. I can then push the microsoft devs to address it. [09:20:20] --- meffie has left [09:20:47] --- reuteras has left [09:35:52] --- kaj has left [09:38:04] --- kaj has become available [09:46:34] --- kaj has left [09:53:02] git reflog rocks [09:53:56] is it better than git flog? [09:54:34] Well, it's an alternative to flogging yourself when you discover you just typed the wrong thing in a rebase and messed up all your work. [09:55:49] i hurt myself with rebase a lot. i try not to use it. i should fix that [09:56:06] Yeh. It's wonderfully powerfully, right up until its not. [10:03:23] well, next time jake is around, i fixed his "old xcode" problem: http://gerrit.openafs.org/1180 [10:55:17] hey cool, I appreciate it [10:59:29] --- meffie has become available [11:04:52] --- Kevin Sumner has become available [11:07:59] --- kaj has become available [12:54:32] --- Kevin Sumner has left [13:18:45] --- jaltman has left: Disconnected [14:02:31] --- deason has left [14:03:30] --- deason has become available [14:10:31] --- deason has left [14:11:53] --- deason has become available [15:14:45] --- deason has left [15:15:05] --- meffie has left [15:23:27] --- mdionne has become available [16:37:39] --- Russ has left: Disconnected [16:55:00] --- Russ has become available [19:46:26] --- mdionne has left [23:40:05] --- Russ has left: Disconnected