Well, I wasn't able to get any useful information from the logs...but the reason I failed may be as interesting as any information I could get. I added --debug in my central configuration database and pushed that out to all the workstations. Then I poked around to see which workstations were exhibiting this behavior.
None of them had any information about autofs in syslog. Odd, I thought, so I checked what was running. autofs hadn't been started with --debug. I tried restarting autofs -- and it failed to kill the current automounter, with varying symptoms (some systems failed outright, others claimed to have succeeded but didn't). I tried unmounting some stuff in /bulk by hand, and got "device is busy"; a user was actively accessing an automounted directory. So: the clients where autofs was mounting the wrong directory appear to be exactly the ones where an attempt was made to restart autofs, but where the attempt failed because the automounted directories couldn't be unmounted. Could that be relevant at all? BTW, just unmounting the incorrect mount point and letting autofs remount it seems to correct the problem. Daniel -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
