Skip to content

Latest commit

 

History

History
39 lines (27 loc) · 1.33 KB

stale-file-handles.md

File metadata and controls

39 lines (27 loc) · 1.33 KB
Raw

"Stale file handle" errors on NFS mounts

Symptoms

chef-client runs fail with a message like the following:

 * directory[/path/to/mountpoint] action create[2018-05-09T13:13:47+00:00] INFO: Processing directory[/path/to/mountpoint] action create (gitlab-nfs-cluster::client line 11)


    ================================================================================
    Error executing action `create` on resource 'directory[/path/to/mountpoint]'
    ================================================================================

    Errno::EEXIST
    -------------
    File exists @ dir_s_mkdir - /path/to/mountpoint

Possible checks

To make sure the problem is stale file handles, attempt to ls the path. It should generate an output like the following:

ls /path/to/mountpoint
ls: cannot access '/path/to/mountpoint': Stale file handle

Resolution

Stale file handles are simply solved by remounting the affected path. If you can afford it, the simpler way is to reboot the machine with something like:

echo "s" > /proc/sysrq-trigger # Attempts to sync disks attached to the system
echo "u" > /proc/sysrq-trigger # Attempts to unmount and remount all file systems as read-only
echo "b" > /proc/sysrq-trigger # Reboots the kernel without first unmounting file systems or syncing disks attached to the system.