[01:03:44] --- abo has become available [01:05:58] --- Russ has left: Disconnected [01:21:10] --- abo has left: Lost connection [02:01:53] --- abo has become available [02:34:29] --- abo has left: Lost connection [03:15:26] --- andersk has become available [04:11:58] --- abo has become available [04:15:54] --- abo has left: Lost connection [05:14:07] --- abo has become available [05:15:19] --- abo has left: Lost connection [05:15:22] --- abo has become available [05:15:32] --- abo has left: Lost connection [05:30:47] --- abo has become available [05:41:24] --- abo has left: Lost connection [05:50:50] --- abo has become available [05:52:57] --- abo has left [05:53:27] --- abo has become available [06:00:17] --- abo has left: Lost connection [06:02:47] --- abo has become available [06:03:06] --- abo has left: Lost connection [06:28:48] --- jaltman has left: Disconnected [06:52:08] --- abo has become available [06:52:09] --- abo has left: Lost connection [07:23:19] --- deason has become available [07:52:03] --- abo has become available [07:54:30] --- mho has become available [07:56:14] --- abo has left [07:58:22] --- mho has left: Lost connection [08:08:23] --- meffie has left [09:07:28] --- abo has become available [09:13:52] --- abo has left: Lost connection [09:39:48] --- mho has become available [09:42:46] --- mho has left [09:58:04] --- abo has become available [10:00:34] --- abo has left: Lost connection [10:04:13] --- abo has become available [10:04:57] --- abo has left: Lost connection [10:24:06] --- jaltman has become available [10:49:17] --- rra has become available [11:01:17] --- jaltman has left: Disconnected [11:01:30] --- jaltman has become available [11:12:37] --- abo has become available [11:20:34] --- jaltman has left: Disconnected [11:20:46] --- abo has left: Lost connection [11:45:19] --- abo has become available [11:50:38] --- abo has left: Lost connection [12:02:27] --- abo has become available [12:06:12] --- abo has left: Lost connection [12:18:23] --- abo has become available [12:19:53] --- abo has left: Lost connection [12:51:33] Is there going to be a release announcement for 1.5.76? I've been holding off on new Debian packages until that comes out. [12:53:12] --- jaltman has become available [13:35:28] > Is there going to be a release announcement for 1.5.76? uh, crap [13:48:56] --- meffie has become available [13:55:59] --- geekosaur has left [13:56:05] --- geekosaur has become available [14:19:57] --- geekosaur has left [14:23:17] --- geekosaur has become available [14:43:45] --- abo has become available [14:43:45] --- abo has left [14:44:32] --- abo has become available [14:53:32] --- abo has left: Lost connection [15:03:24] --- abo has become available [15:05:08] --- abo has left: Lost connection [15:25:37] --- abo has become available [15:26:38] --- abo has left: Lost connection [15:39:00] --- abo has become available [15:39:35] --- deason has left [15:43:23] --- abo has left: Lost connection [15:46:38] --- abo has become available [16:22:21] --- abo has left: Lost connection [16:36:49] --- kula has left [17:10:39] --- abo has become available [17:16:42] --- abo has left: Lost connection [17:26:56] --- abo has become available [17:28:19] --- abo has left: Lost connection [18:01:39] --- shadow@gmail.com/owl60DE4E08 has left [18:02:16] --- shadow@gmail.com/owl1BF2FE0F has become available [18:10:50] --- kula has become available [18:19:28] --- rra has left: Disconnected [18:40:31] --- dwbotsch has left [18:40:49] --- dwbotsch has become available [18:45:00] Hm. Sure would be nice if PAGs could be cgroups. [18:49:17] gpl only? [18:54:16] --- deason has become available [18:54:59] I thought cgroups were controllable even from userspace [18:56:12] i suppose we could add a userspace helper. my "gpl only" was a question tho [18:58:22] IIRC they were controlled from userspace via writing to something in a virtual filesystem... which I would think you could still do via the normal kernelspace file i/o mechanisms but I dunno [18:58:34] "in the worst case" anyway; I'd hope there are better interfaces to it [18:59:07] --- Russ has become available [19:01:06] No, the problem is that cgroups aren't really generalized; they're more like multi-specialized. The cgroup mechanism has to know, at kernel compile time, about every subsystem that's going to use cgroups. There has to be a line in a particular header for each subsystem, which controls sizing of a per-group fixed-size array containing the subsystem private data pointers and the enum that indexes that array. And, each subsystem has to provide certain callback interfaces. [19:01:27] Also, the cgroup-related kernel APIs aren't exported _at all_, so not only do you have to be in-tree, you have to be compiled-in. [19:01:56] It's unfortunate, though, because they're very nearly exactly PAGs. [19:06:42] I thought you could do it by cgroup name, not needing a subsystem... but I haven't really looked at it so I'm probably wrong [19:08:06] it sounds like kafs stands a chance at using it sometime, though, even if we can't [19:08:13] but don't keyrings solve this for us anyway already? [19:59:20] > do it by cgroup name do what? cgroups don't have names; heirarchies do. but for AFS to use cgroups, we need private data in which to store the PAG, possibly the iteration operations, and the notify-on-last-deref facility, via the kernel API, and none of the kernel API is exported. [20:10:24] --- dwbotsch has left [20:10:56] --- dwbotsch has become available [20:17:58] --- phalenor has left [20:27:59] --- phalenor has become available [20:51:24] --- jaltman has left: Replaced by new connection [20:51:25] --- jaltman has become available [21:13:05] --- jaltman has left: Disconnected [21:13:18] --- jaltman has become available [21:15:06] --- abo has become available [21:19:04] --- abo has left: Lost connection [21:20:51] --- abo has become available [21:21:39] --- abo has left: Lost connection [21:22:30] something like the equivalent of mkdir /dev/cgroup/AFSPAG:1234 cd /dev/cgroup/AFSPAG:1234 echo $NEWPAGPID > tasks cp /some/openafs/helper release_agent ? [21:22:55] '1234' to be just interpreted as a pointer to the private data payload somewhere in libafs [21:23:28] you do still need the notify_on_release flag turned on by the kernel, though, so eh [21:23:41] I might be misinterpreting the usage of these things entirely anyway, making what I said moot [21:24:47] --- abo has become available [21:31:59] --- abo has left: Lost connection [21:33:13] --- abo has become available [21:34:07] --- abo has left: Lost connection [22:06:11] --- abo has become available [22:06:24] --- deason has left [22:06:27] --- abo has left: Lost connection [23:29:40] --- abo has become available [23:31:24] --- abo has left: Lost connection [23:48:15] --- abo has become available [23:49:59] --- abo has left: Lost connection [23:59:35] --- abo has become available