123

I've got an entry for an external harddrive in my fstab:

UUID="680C0FE30C0FAAE0" /jgdata ntfs noatime,rw

But sometimes this drive isn't plugged in at boot time. This leaves me half way through a boot, with a prompt to "Continue Waiting, press S or press M" but no keypress has any affect at this stage (including Ctrl-Alt-Delete, not even caps-lock).

Short of writing a script to check the output of fdisk -l, how can I mount this drive at boot time only if it is present? It would be handy to have an fdisk entry for this drive, so I can just type mount /jgdata instead of needing a device name.

Improve this question

5 Answers 5

Reset to default
120

You're all on the right trail. I've found a way that is a little more clean and better form.

The correct option to add in fstab is nofail, however, it needs to come after auto. If you change it to noauto, it will never mount during boot. If you add nobootwait to the bootloader, you could potentially miss something serious, such as mounting the partition before fsck finishes checking it, or not mounting a partition that is used in the boot process.

After making the above change, The system will start normally (and mount the volume) if the device is plugged in while the system is shutdown. It will also boot normally if the device is not present at boot time.

The only inconvenience is that if you connect the device while the system is running, depending on configuration (too many variables to test), the device may not mount immediately. This can be remedied with a simple mount -a or mount /specific_device or a reboot.

Improve this answer
5
  • 4
    Every example I can find online shows nofail before auto. Is there any documentation for the correct order?
    – Hackeron
    Apr 7, 2017 at 18:30
  • 'man fstab' doesn't give the precision of a needed order (checked in centos 7 and Ubuntu 14.04). Though, it says "nofail do not report errors for this device if it does not exist." I'm wondering if this will help not waiting the timeout at boot if nfs volume is not accessible.
    – tisc0
    Dec 10, 2017 at 18:34
  • I omit auto altogether. My use case is to allow servers with failed data disks to complete boot.
    – dannyman
    Nov 27, 2019 at 21:39
  • does not work for encrypted file systems?
    – user2660966
    Oct 28, 2021 at 7:03
  • Regarding the timeout, perhaps refer to this answer as it includes configuring the timeout option rather than waiting an additional 90 seconds for a device to be inserted.
    – s3dev
    Aug 14, 2023 at 15:54
32

I had the same issue - I've done one extra step

If you use the nofail option in /etc/fstab, the system will look for your disk (and partition) at boot time. If the device is plugged, the filesystem will be mounted. If not, the boot will continue as normal.

See arch wiki: https://wiki.archlinux.org/index.php/Fstab

Example

UUID=XXXXXXXXXXXXXXX    /myhdd ntfs  auto,nofail,noatime,rw,user    0   0

I've tried to boot the system with and without the device plugged, and it works ok.

What I've not achieved is to automount when disk is plugged after boot (when isn't plugged at boot). I must use mount -a as root to mount all again.

Improve this answer
0
27 
/dev/xvdh1 /myfs xfs defaults,nofail,x-systemd.device-timeout=30 0 0

worked for me.

  • nofail - Do not report errors for this device if it does not exist.
  • x-systemd.device-timeout=30 - Boot will take 30 seconds longer if device does not exist (default is 90 secs).

Source: https://wiki.archlinux.org/index.php/fstab#External_devices

Improve this answer
2
  • This is the best and simplest answer. Note that this is the new syntax for newer versions of the OS that use systemd, replacing the old nobootwait syntax
    – cxrodgers
    Sep 21, 2018 at 21:09
  • 1
    This worked for me. Key was adding the device-timeout; just using nofail the boot would indeed fail due to udev timeout. (probably unique to USB devices)
    – BobHy
    May 24, 2019 at 21:48
14

does the noauto option let the boot process continue?

it doesn't automatically mount if present, but it does get it known if present so a simple mount /jgdata works...then a scripted mount /jdata wouldn't need an output check, just catch the error and keep booting

edit: upon some further reading bootwait is probably a more correct option to pass...(usually used for network shares that might not be present until later in the boot process, but it might still cause a hang, idk)

and the mount script could be added like so: https://stackoverflow.com/questions/2062543/running-a-script-with-the-help-of-grub-and-menu-lst

Improve this answer
2
  • Thanks for the answer. Just to clarify, the options are to be added to the fstab line, correct? (EDIT: Nvm, googled it. Looks like nobootwait is what I need.)
    – Jeremy
    Nov 24, 2010 at 4:56
  • 1
    I've used the nobootwait option in fstab, works perfectly, thanks.
    – Jeremy
    Nov 24, 2010 at 5:04
2

The recommended way to mount during the boot is instructing their system through the fstab file. Looking at your Ask, I could see you are almost there, lacks only the instruction that sets the device to use automount options, allowing your system to mount the device when its available.

So, just rewrite the line in your fstab to be like below:

# <file system>           <dir>       <type>    <options>         <dump> <pass>
UUID="680C0FE30C0FAAE0"   /jgdata      ntfs      user,auto,rw       0     0

After change and save it, try to mount it by hand:

$ sudo mount -a

It's important to note that:

  1. you need to be sure about the device's UUID. UUIDs are generated by the make-filesystem utilities (mkfs.*) when you create a filesystem.
  2. Those <options> needs to be written following a very specific format, separated by commas but no spaces after each comma. Be careful with this ;-)
  3. I'm not sure if this will work smoothly because you are trying to automount a NTFS filesystem, that are handleable using NTFS-3G utilities. While my instructions are supposed to work correctly, I never tried automount NTFS before. So, if its failed, I recommend you to look at this Mounting Windows Partitions for alternative uses of NTFS.

Thanks!

Improve this answer
4
  • i think he gets most of what you said, but it's hanging when the disk isn't detected. that's why i suggested a workaround to the "proper" way
    – RobotHumans
    Nov 23, 2010 at 5:02
  • Thanks for the answer. FYI the sudo blkid command is a good way to find out UUIDs.
    – Jeremy
    Nov 24, 2010 at 4:44
  • I have just tried that exact line in fstab. When the HDD is plugged in, the system boots normally. But, during boot, when the external HDD is not plugged in, it still prompts me to Wait/Skip/Manualy Recover. Since this prompt doesn't time out, if I'm not at the terminal, the box will not boot if the external HDD is not plugged in.
    – Jeremy
    Nov 24, 2010 at 4:53
  • You need to add the nofail, option.
    – kkron
    Dec 30, 2015 at 20:06

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .