> Install the autofs-debuginfo package and when you see this use gdb to > get a backtrace of the running threads. > > gdb -p <automount pid> /usr/sbin/automount > gdb> thr a a bt
will do. >> 1. is there a better way to do this? in the old sun days, it was like: >> >> user1 server:/disk/1:user1 >> user2 server:/disk/1:user2 > > When was this syntax used, I don't remember it? > And exactly how is it supposed to work? that was the original automount in sunos. i'm pretty sure that sun deprecated it as well and just treated the extra : as a slash. but it worked almost exactly like i want. a single nfs mount is made, no matter how many subdirectories were referenced. i can't recall exactly how the mounts were laid out to make it all happen. but it was extremely useful in the days when a system could only handle 32 or some really small number of nfs mounted file systems. that isn't such an issue today, but i still worry about it for a mail or web server that may touch 1000s of home directories at once so that is why i use the recursive automounts. it also makes df output much more readable. like i say, it has been working pretty well until recently (i think when i started using autofs v5). _______________________________________________ autofs mailing list [email protected] http://linux.kernel.org/mailman/listinfo/autofs
