[00:21:35] --- dev-zero@jabber.org has become available [00:37:22] --- abo has left [00:37:33] --- abo has become available [00:54:25] --- kaj has become available [01:58:22] --- haba has left [02:14:35] --- abo has left [02:14:53] --- abo has become available [02:34:12] --- pod has left [02:34:31] --- pod has become available [02:43:53] --- abo has left [02:44:48] --- abo has become available [02:46:20] --- Jeffrey Altman has left: Replaced by new connection [03:03:48] --- jaltman has left: Disconnected [04:01:30] > assuming we can determine that we cannot receive callbacks from a file > server, should we stop serving data from that file server? how would we know we couldn't, ever? [04:18:49] --- abo has left [04:19:01] --- abo has become available [04:29:32] --- abo has left [04:30:02] --- abo has become available [04:59:09] --- haba has become available [05:39:48] --- meffie has become available [05:40:02] --- meffie has left [05:49:51] --- meffie has become available [05:56:04] --- jaltman has become available [06:13:39] --- haba has left [06:13:59] --- haba has become available [06:20:00] --- haba has left [06:20:40] --- haba has become available [06:32:57] --- haba has left [06:51:34] --- abo has left [06:52:28] --- abo has become available [07:23:20] --- deason has become available [07:24:54] one thought. we could create a new RPC that queues a callback response at some time in the future. if the expected callback does not arrive, then we can assume there is a firewall / NAT blocking the port. [07:25:30] --- jaltman has left: Disconnected [07:54:59] --- abo has left [07:55:34] --- abo has become available [07:57:18] --- Simon Wilkinson has left [08:02:34] --- Simon Wilkinson has become available [08:05:30] --- Simon Wilkinson has left [08:06:26] --- Simon Wilkinson has become available [08:25:06] --- kaj has left [09:29:57] --- dev-zero@jabber.org has left [09:36:53] --- Russ has become available [10:18:24] --- reuteras has left [10:22:52] > what we should do is degrade into a mode where callbacks expire almost > immediately Instead of degrading into a mode where you can get your data, but it might be stale, we should degrade into a mode where you cannot get your data? That sounds dumb to me. [10:24:43] And no, if a fileserver is up but a client is behind a NAT, the client should not degrade into a mode where it attempts to make the fileserver not be up. [10:25:03] Well, it would degrade into a mode where the client caches ~nothing. [10:25:20] Like the Linux kafs client :) When that ships widely, fileservers are going to be sad. [11:36:16] Anyone know off the top of their head whether a GSSAPI acceptor identity created by an initiator using one name type can be used by an acceptor expecting a different name type? [12:34:39] --- jaltman has become available [12:36:17] --- meffie has left [12:56:50] --- deason has left [12:57:01] --- deason has become available [13:00:06] --- abo has left [13:00:17] --- abo has become available [13:00:18] --- abo has left [13:29:47] --- deason has left [13:30:14] --- deason has become available [13:41:50] --- deason has left [13:42:09] --- deason has become available [14:03:08] --- haba has become available [15:27:56] --- haba has left [15:29:01] --- haba has become available [15:46:45] --- deason has left [16:02:38] --- mdionne has become available [17:53:21] --- deason has become available [18:19:47] --- deason has left [18:46:07] --- deason has become available [19:33:35] --- mdionne has left [21:59:47] --- reuteras has become available [22:30:54] --- deason has left [22:39:32] --- steven.jenkins has left [22:47:28] --- steven.jenkins has become available [22:47:45] I'm seriously tempted to propose running GNU cppi across the entire source, although this is another one of those things likes fixing whitespace. [22:48:03] But for folks making significant changes to part of AFS, cppi will fix the indentation of C preprocessor conditionals for you. [22:49:35] --- kaj has become available [22:50:21] --- steven.jenkins has left [22:51:48] --- steven.jenkins has become available [23:12:59] --- kaj has left [23:21:50] --- Russ has left: Disconnected [23:25:04] --- haba has left [23:26:13] --- haba has become available