When one runs /bin/pwd (using coreutils 5.97) on an autofs mounted
volume, it causes an fstat to be run on all autofs mounted volumes
on that automount's mount point. THis is not only a big waste but
if a NFS mount server down it causes /bin/pwd to hang. /bin/pwd
in perfectly working NFS mount should not hang because some other
mount is down.
I have an autofs mount point at /autofs/space. THe autofs map for this mount
point has over a thousand entries. If I am in /autofs/space/pgolland_001
for instance I do a 'strace /bin/pwd' I see a thousand lines like:
fstatat64(4, "meso_018", {st_mode=S_IFDIR|0555, st_size=0, ...},
AT_SYMLINK_NOFOLLOW) = 0
fstatat64(4, "47tuser_006", {st_mode=S_IFDIR|0555, st_size=0, ...},
AT_SYMLINK_NOFOLLOW) = 0
fstatat64(4, "bomba_036", {st_mode=S_IFDIR|0555, st_size=0, ...},
AT_SYMLINK_NOFOLLOW) = 0
fstatat64(4, "pimlico_004", {st_mode=S_IFDIR|0555, st_size=0, ...},
AT_SYMLINK_NOFOLLOW) = 0
fstatat64(4, "tadasana_001", {st_mode=S_IFDIR|0555, st_size=0, ...},
AT_SYMLINK_NOFOLLOW) = 0
fstatat64(4, "acumonk_003", {st_mode=S_IFDIR|0555, st_size=0, ...},
AT_SYMLINK_NOFOLLOW) = 0
fstatat64(4, "mody_004", {st_mode=S_IFDIR|0555, st_size=0, ...},
AT_SYMLINK_NOFOLLOW) = 0
fstatat64(4, "soulenoid_006", {st_mode=S_IFDIR|0555, st_size=0, ...},
AT_SYMLINK_NOFOLLOW) = 0
fstatat64(4, "madrc_003", {st_mode=S_IFDIR|0555, st_size=0, ...},
AT_SYMLINK_NOFOLLOW) = 0
until it hangs on one that is mounted but down at the moment.
--
---------------------------------------------------------------
Paul Raines email: raines at nmr.mgh.harvard.edu
MGH/MIT/HMS Athinoula A. Martinos Center for Biomedical Imaging
149 (2301) 13th Street Charlestown, MA 02129 USA
_______________________________________________
Bug-coreutils mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/bug-coreutils
