Looking for non-existing swap partition during startup: A startjob is running

  1. no physical swap partition exits
  2. sudo swapoff -a
  3. /etc/fstab is OK

still, booting is delayed by 1:30 minutes due to this bug. Any advice? Thanks!

hi @ng0177,

I would double check your /etc/fstab file. Most of the time this bug is caused by an entry in the /etc/fstab that refers to the old uuid for your partition.

It saw a swap partition during install which was later removed i.e. blkid does not help wrt /etc/fstab. I probably should report it in bugzilla. It strikes me, though.

If you cat /proc/cmdline, do you see the non-existent swap partition UUID referenced there?

Yes, I am guessing that “resume=UUID=” is trying to access it. If so, what are the steps to remove the option? Thanks.

hi @ng0177,

Did you happen to install Fedora SilverBlue alongside another OS on your machine by chance? I ask because looking on the Ubuntu forums there’s an identical situation someone also experienced.

Yes, it lives alongside many other OS. However, all is clean after having removed the swap partitions. The root cause is probably the “resume” pointing to a non-existing UUID. How to remove it, just edit the file and that is it? Thanks.

@n0177 What is the output for sudo cat /boot/grub2/grubenv?

You might need to fix resume=<device> using rpm-ostree kargs --editor

rpm-ostree kargs --editor solves the problem (even w/o running grub-mkconfig) but sudo cat /boot/grub2/grubenv still shows resume = which is odd. Appreciate your comments.

1 Like

rpm-ostree (and Fedora 30 in general) manage kernel arguments quite differently from other OSs.

Suddenly, I cannot boot into Silverblue

Generating "/run...

Entering emergency mode.

Should I open a new thread? The problem might be related, tough.


You probably should, did you modify any kernel arguments at all other than removing the swap device?

see Entering emergency mode during boot

Any advice on a good piece of documentation that describes how Silverblue differs from Fedora? Thanks.