[00:01:15] --- jaltman has left: Replaced by new connection [00:01:16] --- jaltman has become available [00:06:26] --- jaltman has left: Replaced by new connection [00:06:27] --- jaltman has become available [00:10:54] --- jaltman has left: Replaced by new connection [00:10:55] --- jaltman has become available [00:22:27] --- Russ has left: Disconnected [00:35:44] --- jaltman has left: Disconnected [01:29:06] --- Simon Wilkinson has left [02:45:56] --- Simon Wilkinson has become available [04:32:29] --- jaltman has become available [05:00:41] --- jaltman has left: Disconnected [05:41:36] --- jaltman has become available [05:45:38] --- meffie has become available [06:03:28] --- meffie has left [06:03:32] --- meffie has become available [06:05:37] --- jaltman has left: Replaced by new connection [06:05:40] --- jaltman has become available [06:12:48] --- jaltman has left: Disconnected [06:12:57] --- jaltman has become available [06:55:27] --- Simon Wilkinson has left [07:16:47] --- deason has become available [09:08:57] --- jaltman has left: Disconnected [09:11:26] --- ezyang@mit.edu/barnowl has left [10:27:24] --- shadow@gmail.com/owl60747F95 has left [10:52:14] --- shadow@gmail.com/owlF2E50C3B has become available [11:02:31] --- rra has left: Disconnected [11:03:59] --- kaduk@mit.edu/barnowl has left [11:04:00] --- jaltman has become available [11:05:11] --- kaduk@mit.edu/barnowl has become available [11:11:30] --- jaltman has left: Disconnected [11:12:14] --- jaltman has become available [11:13:01] --- rra has become available [11:13:04] --- rra has left [11:13:45] --- rra has become available [11:50:16] Are there known common classes of bugs that cause things to stay in afs_rx_cv_wait for tens of seconds? [11:51:35] uh. that narrows it down only a little. which cv? [11:52:23] Dunno. (This is just what I frequently see at the beginning of a 'make buildworld' when it seems to hang and I press C-t to get info.) [11:53:18] sadly the string provided to CV_INIT is not used [12:45:01] Well, you'll be in CV wait if you're say, in rx_Read blocked waiting on data from the other end. [12:45:42] My machine is on the same ethernet switch as the fileserver. I am somewhat skeptical of a tens-of-seconds wait for the fileserver to give me data. [12:45:58] (But I don't discount that possibility entirely.) [12:46:34] what version is the server? if it answered too fast, it's possible [12:48:58] OpenAFS 1.4.8 built 2009-03-24 I seem to recall it has a couple of local patches that weren't in the debian package at the time. [14:04:57] --- jaltman has left: Replaced by new connection [14:09:37] --- jaltman has become available [14:24:12] --- jaltman has left: Disconnected [14:24:30] --- jaltman has become available [14:33:03] --- meffie has left [14:38:14] --- meffie has become available [15:28:32] During my afs_rx_cv_wait steps: freebuild# rxdebug localhost Trying 127.0.0.1 (port 7000): getstats call failed with code -1 [15:30:43] --- deason has left [15:52:15] uh, are you debugging a client or a server? [15:52:37] because port 7000? not so much client [15:55:30] Oopsie. [15:55:38] ok then. [15:56:00] (Needed to come up with the command in the 20-second window, since I wasn't thinking ahead.) [16:25:18] --- jaltman has left: Disconnected [16:33:12] --- cudave has become available [16:44:57] --- deason has become available [17:10:26] Hm, lockup with master+3219. Not sure if 3219 is relevant, though. Nothing in cmdebug; an ld process is stuck in afsslp. I may have to dump a kernel core to get a full stack. [17:12:43] Any other info to grab before resorting to the sledgehammer? [18:39:57] Okay, so I was sleeping forever in afs_MemRead. [18:43:51] --- rra has left: Disconnected [19:03:31] --- Russ has become available [19:57:00] --- jaltman has become available [22:21:28] --- deason has left [23:46:39] --- kaj@kth.se has become available