Forcibly Unmounting NFS Filesystems

forcibly unmounting nfs filesystems

available solutions

setup a temporary

“surrogate” nfs

server

In the example shown in Figure 1 the fuser(1M) command was able to kill the processes

holding the NFS filesystem busy, and thereby allow the filesystem to be unmounted.

However, in many situations the processes accessing a down NFS filesystem are sleeping

at an uninterruptible level in the kernel and are unable to receive signals, such as SIGKILL

or SIGTERM. In these situations, fuser(1M) will not be able to kill these processes and the

filesystem will remain busy.

Obviously the best solution in this situation is to get the real NFS server back in operation

so that these blocked processes can complete their transactions. Unfortunately, it isn’t

always possible to bring the NFS server back online quickly, or at least not in an

acceptable timeframe for NFS client users.

At times such as this, one possible solution is to setup a temporary “surrogate” NFS

server. In other words, locate a working NFS server system in your environment and

temporarily configure the IP address (or addresses) of the down server to this working

NFS server. By doing this, the NFS client systems will be able to get a response from a

live NFS server system. If an alternate NFS server cannot be located, another possibility

would be to take one of the blocked NFS client systems and enable NFS server-side

services on this system and then add the down NFS server’s IP address to this system.

Figure 2 below and Figure 3 on page 8 illustrate the various steps involved in setting up

a temporary “surrogate” NFS server. Each step in the example is numbered and

described in detail below.

Figure 2 - Setup a “surrogate” NFS server (part 1)