Joey's Notes: Using the Automounter
By Joey Prestia
In a Windows networked environment, you can go to any machine, log in to the system, and have access to your home directories and files on the server. This seems like a convenience until you realize that in doing this, the server and the client machines are eating up resources by keeping all network shares hard-mounted at all times. The Linux solution to this is the autofs daemon known as the automounter.
In the past, Linux networks would have various types of Network File System (NFS) servers being utilized in different ways: some for network backup, some exporting users' Network Information Service (NIS) mapped directories, and some exporting other network resources. One of the problems was that booting a system that used traditional fstab-based mounts while the NFS server was down could be a long process: it would take a long time to come up as it waited for the server to time out. Similarly, when you have two servers, each mounting directory hierarchies from the other, and both systems are down, both may hang as they are brought up (this is called server-server dependency). The automount facility gets around these issues by mounting a directory hierarchy from another system only when a process tries to access it.  If a client attempts a hard mount and the server is unavailable, then the Linux kernel treats the process as it would if a local drive's file system was unavailable. The user can do nothing until the server becomes available; the process is in an un-interruptible sleep until the server is back on line. Often, the process can't even be killed, and a "ps aux" command will reveal a 'D' state.
[ For anyone not familiar with "ps" output, 'D' means "Uninterruptible sleep (usually IO)". For more information, please see the "PROCESS STATE CODES" section in "man ps". -- Ben ]
Automounting is the process where the mounting and unmounting of file systems is done automatically by the autofs service. If the filesystem is unmounted, and a user attempts to access it, it will be automatically mounted or re-mounted. Using the automounter saves on resources by mounting a directory only when you need to enter it, then unmounting it automatically when you don't need it anymore. This is usually done after a default timeout (usually 300 seconds) that is specified in the /etc/sysconfig/autofs file, the main configuration file for this service. The other configuration files are located in the /etc directory and are called auto.master, auto.misc, auto.net, and auto.smb. The last two are executable scripts.
Let's say we have some NFS directories and our company's employees are only trained in use of the GNOME desktop environment; they could use the automounter if you were to set up the configuration files and copy them to their machines. They would be able to point and click their way around, and the remote file systems would do the right thing, invisibly. Using the automounter is not restricted to the console.
- Verify that /etc/sysconfig/autofs options will accommodate your particular needs (i.e., browse mode and timeouts)
- Modify /etc/auto.master as appropriate
- Create any /etc/auto.files that were referenced in /etc/auto.master
- Start or restart the autofs service
- Change into the appropriate directory
Let's check out the configuration files. (Note: this guide was written using RHEL 5.1, so other distributions may differ slightly in some aspects.) The main configuration file, "/etc/sysconfig/autofs", has the timeout for mounts defined as 300 seconds (or five minutes) of inactivity, and the default browse mode set to "no"; this prevents shares that are not being directly referenced from being automounted. There are lots of other configurable options for different setups, so it is worth your time to read through this file just so you know what is in there. Look particularly closely at any uncommented directives.
'auto.master', contains three uncommented lines:
/misc /etc/auto.misc /net -hosts +auto.master
The first line means that the configured devices in auto.misc will be mounted in the "/misc" directory; the second one means that the exports from the network hosts will be mounted in "/net". '+auto.master' is a reference to an external NIS or NIS+ master map. Entries in this file consist of a mount point followed by the path to a file which defines the details and the options of what will be mounted where. You can call the files anything you want, or whatever suits the need. The following entries would be valid:
/remote /etc/auto.somewhere_else /mnt/nas /etc/auto.nas /mnt/backup /etc/auto.backup
We would then create the file that we referenced in the auto.master and define the mount details in it.
It is also useful to consider mounting the users' remote home directories in a non-standard location (e.g., "/rhome/username" or "/export/home/username".) This would be considered a violation of the FHS (Filesystem Hierarchy Standard), since the mount point is located in the root of the filesystem; on the other hand, it's only a temporary directory, created and destroyed by the autofs daemon as necessary. The advantages are obvious: you get to have access to the content of your "/home/username" directory as well as the remote one, and you get to keep the use of the /mnt mount point for other devices or shares.
This is the file that you create for your custom automounts. The auto.files would have three columns in them:
key mount options location or device
The "key" is the subdirectory under the mount point that you listed in the auto.master file. As an example, if in our auto.master we had the following:
/misc /etc/auto.misc /net -hosts /mnt/remote /etc/auto.remote +auto.master
We would then create, for example, the /etc/auto.remote file with our desired options:
# key mount options location or device records -rw,soft,intr server1.example.com:/srv/nfs/medical/records
Once we restart the autofs service, changing to the /mnt/remote/records directory will create the demand for the exported resource server1.example.com:/srv/nfs/medical/records to be mounted on /mnt/remote/records. It is important to note that if you use host names (rather than IPs) and don't have DNS available, you will need an entry in /etc/hosts that maps the hostname to the IP address.
The following example uses wildcard substitution. Placing an asterisk in the key field and an ampersand at the end of the location or device path field will map the end of the location path (directory structure and files) to whatever is found at the location. In the example below, the entire remote home directory structure located on server1.example.com:/home/username would be recreated when the user attempts to log in. By using this technique, a user could log into any machine and have their home directory mapped appropriately.
# key mount options location or device * -rw,soft,intr server1.example.com:/home/username/&
The /etc/auto.misc file contains a lot of sample configurations for automounting various types of devices; to use one of these, merely uncomment a specific line in this file and adjust the appropriate device accordingly. But just as different machines have different hardware configurations, different distros may have different setups so a little customization will probably be necessary.
# This is an automounter map and it has the following format # key [ -mount-options-separated-by-comma ] location # Details may be found in the autofs(5) manpage cd -fstype=iso9660,ro,nosuid,nodev :/dev/cdrom # the following entries are samples to pique your imagination #linux -ro,soft,intr ftp.example.org:/pub/linux #boot -fstype=ext2 :/dev/hda1 #floppy -fstype=auto :/dev/fd0 #floppy -fstype=ext2 :/dev/fd0 #e2floppy -fstype=ext2 :/dev/fd0 #jaz -fstype=ext2 :/dev/sdc1 #removable -fstype=ext2 :/dev/hdd
Most distros now automatically mount USB devices under the /media directory and several of the devices listed above are fading away - but there is still some use for this file. It can be configured in a multitude of ways, e.g. to automount partitions that you don't need to have mounted at all times (backup devices are just one example). A good way to get familiar with this file is to come up with a configuration scenario and put it into practice. For example, try mounting a USB flash drive with the automounter and setting up different configurations until you get familiar with it. Remember to always back up the original file before making any changes.
Suppose we want to be able to automount the NFS share on the network server
instead of, say, manually typing
"mount -t nfs
192.168.0.254:/var/ftp/pub/directory/subdirectory /mnt".) In this case, we
would just need the hostname of the machine with the exported filesystem
(if there is an entry in DNS) or the IP address of the machine. The
"showmount -e hostname" or "showmount -e IP_address" command can be used
to discover what directories a host has exported by NFS. Example:
"showmount -e 192.168.0.254" will display the exports for that host.
The auto.net file is an executable script which searches for available NFS shares to mount; the NFS service must be running for the auto.net script to be able to connect to remote shares. To invoke this ability, you need to ensure that the autofs and NFS services are running:
[root@station17 ~]# service autofs status automount is stopped [root@station17 ~]# service nfs status rpc.mountd is stopped nfsd is stopped rpc.rquotad is stopped
Since they are stopped, we start the service with the "service" command, and ensure that they stay on through a reboot by using the "chkconfig" command on Red Hat based systems.
[root@station17 ~]# service autofs start Starting automount: [ OK ] [root@station17 ~]# chkconfig --level 35 autofs on [root@station17 ~]# service nfs start Starting NFS services: [ OK ] Starting NFS quotas: [ OK ] Starting NFS daemon: [ OK ] Starting NFS mountd: [ OK ] [root@station17 ~]# chkconfig --level 35 nfs on [root@station17 ~]#
The "chkconfig" arguments just ensure that this service will remain on through a reboot in runlevels 3 and 5. Now we can see if any network server exports are available. First, we'll go to /net; assuming that our server with the NFS export is "server1" and has a DNS entry, we could execute the following:
[root@station17 /]# cd /net [root@station17 net]# cd server1 [root@station17 server1]# ls data documents powervault rhome var [root@station17 server1]#
or, if we know the IP address of the machine, this would work:
[root@station17 net]# cd 192.168.0.254 [root@station17 192.168.0.254]# ls data documents powervault remote var [root@station17 192.168.0.254]#
The directory magically appears - but if we didn't know the machine's hostname or IP address, we would not be able to find it. The automounter does not list directories unless they have been accessed, so unless a user knows what to access they will not be able to blindly find it. This sure makes mounting NFS shares a lot easier - and you don't have to remember to manually unmount them!
As an exercise to see this work on our machine, let's execute the following commands.
[root@station17 ~]# echo "/var/www/html 127.0.0.1(sync,rw)" >> /etc/exports [root@station17 ~]# service nfs restart Shutting down NFS mountd: [ OK ] Shutting down NFS daemon: [ OK ] Shutting down NFS quotas: [ OK ] Shutting down NFS services: [ OK ] Starting NFS services: [ OK ] Starting NFS quotas: [ OK ] Starting NFS daemon: [ OK ] Starting NFS mountd: [ OK ] [root@station17 ~]# cd /net/localhost [root@station17 localhost]# ls var [root@station17 localhost]#
We've exported our /var/www/html directory as a read/write NFS export to our loopback IP address just to try this (in case you're not in a networked environment). Don't forget to remove the line in /etc/exports that we echoed into the file when you are done checking out the network mounting capabilities. Do note that typing "mount" at the command line will not show what is mounted by the automounter: you will have to execute a "cat /proc/mounts" to see what is currently mounted by the autofs daemon.
The default auto.smb file is also an executable script, and the use of it may require the Samba servername/username/password, or that a credentials file be specified for your particular setup. You can also use the guide included with your Samba documentation, as I did, to create your own auto.smb file; it follows the same format as the auto.misc file. If you want to automount Samba shares, you may want to just go ahead and create your own configuration file customized for your own needs. Just remember to back up the original file if you're not going to use the executable.
# automount points below /smb # This is an automounter map and it has the following format # key [ -mount-options-separated-by-comma ] location # Details may be found in the autofs(5) manpage # smb-servers supra_andreas -fstype=smb,username=andreas,password=foo ://supra/aheinrich supra_cspiel -fstype=smb,username=cspiel ://supra/cspiel phonon_andreas -fstype=smb,username=andreas ://phonon/andreas helium_cspiel -fstype=smb,username=cspiel ://helium/cspiel
The automounter will work with NFS exports, Samba, NIS, NIS+, LDAP, and various other devices, not to mention being scriptable and very configurable. In this article, I've just touched on the basics very briefly; don't forget to read the autofs(5) man page, since it has some interesting concepts. Also, be sure to consult it when setting up automounting, as well as any other distro-specific documentation when configuring the automounter to meet your needs.
- The Filesystem Hierarchy Standard
- Red Hat Documentation
- Michael Jang's book "Red Hat Certified Engineer Linux Study Guide", Fifth Edition
- Mark G. Sobell's book "A Practical Guide to Red Hat Linux" Third Edition
 Taken from Mark G. Sobell's book "A Practical Guide to Red Hat Linux" Third Edition, page 691
Joey was born in Phoenix and started programming at the age fourteen on a Timex Sinclair 1000. He was driven by hopes he might be able to do something with this early model computer. He soon became proficient in the BASIC and Assembly programming languages. Joey became a programmer in 1990 and added COBOL, Fortran, and Pascal to his repertoire of programming languages. Since then has become obsessed with just about every aspect of computer science. He became enlightened and discovered RedHat Linux in 2002 when someone gave him RedHat version six. This started off a new passion centered around Linux. Currently Joey is completing his degree in Linux Networking and working on campus for the college's RedHat Academy in Arizona. He is also on the staff of the Linux Gazette as the Mirror Coordinator.