Occasionally, the problem of "stale nfs file handles" arises on an external drive. This is a strange error which I believe has to do
with a file changing without it's time stamp being updated. I have no idea which index or hash this offends, but it seems to occur with external USB drives on occasion, in such a way that prevents mounting the drive. What to do? My first thought was to run
fsck, but without mounting, we can't be sure of the "dev" name to provide
fsck, and neither can we run
fsck when the device with stale handles is mounted.
The best way to break this impasse is to use
udevmonitor or, if now under the dreaded "systemd",
udevadm monitor --environment --udev", to get the block device name. Then just verify it's unmounted, and
fsck it. Let's say we plug-in our our USB drive with stale NFS file handles, and it becomes labeled "/dev/sdc" by the system when it's detected...
SysV
# udevadmin monitor [plug-in device]
# umount /dev/sdc
# fsck /dev/sdc
SystemD
# udevadm monitor --environment --udev [plug-in device]
# umount /dev/sdc
# fsck /dev/sdc
No comments:
Post a Comment