Last thing I did before boot fail was Ctrl-Alt-Del out of boot process because I missed my first chance to re-direct the motherboard to boot from USB. So I did that, then next time I tried to boot the main SSD I ended with “You are in emergency mode…” Fortunately, the USB stick in question is a backup system, so I’m not in too bad a situation, plus I have a separate home partition that’ll preserve everything important if I have to re-install. However, the btrfs root partition doesn’t seem corrupted, as I can explore it when I boot from USB. What I don’t know is what files to check for corruption.
I’ve used linux for years but never had occasion to check logs; “journalctl -xb” from the emergency mode prompt yields a nearly 3k lines that mostly look fine, ending with “The unit plymouth-start.service has successfully enetered the dead state.”
About 30 lines prior, it says “Failed to get new runlevel, utmp update skipped.”
If anyone has experience fixing a problem of this nature, is it easier to just reinstall? And if not, where do I find the file that journalctl reads, and what should I ^F search for to explain where the hiccup is?
I normally tackle issues like this by searching for answers already given rather than asking for new ones, but I do that with an error message of some kind, and in this case I don’t have any unique terms to search for, so thanks for any pointers you can offer.
So I did that, then next time I tried to boot the main SSD I ended with “You are in emergency mode…”
The most common cause of this problem, is that a file system failed to mount.
I suppose a disk failure (SSD failure) is a possibility. I suggest you check the output of “df” to see what has been successfully mounted. Or maybe the output of “mount” which will tell you whether they are mounted read-only or read-write.
However, the btrfs root partition doesn’t seem corrupted, as I can explore it when I boot from USB. What I don’t know is what files to check for corruption.
Are you able to mount the various subvolumes when booting from USB?
I’ve used linux for years but never had occasion to check logs; “journalctl -xb” from the emergency mode prompt yields a nearly 3k lines that mostly look fine, ending with “The unit plymouth-start.service has successfully enetered the dead state.”
That plymouth line is unlikely to be a problem.
About 30 lines prior, it says “Failed to get new runlevel, utmp update skipped.”
That might be significant. Maybe check a few lines before that, if you can.
Thanks for the reply, nrickert. Sorry I haven’t had the kind of day that allowed much time for troubleshooting, but I can say per mount command that the only thing read-only is tmpfs on sys/fs/cgroup. However, when I escape out of the Tumbleweed bootsplash, I see the usual output full of green OKs going by too fast to read otherwise, until it gets to dev/mapper/xxx. For xxx, substitute text changing too fast for me to see, but this is not scrolling, just one line that keeps changing right before I hit emergency mode. Looking through journalctl -xb some more, I see “Dependency failed for Local File Systems” and “Failed to start File System Check on /dev/system/home” and “systemd-fsck0dev-system-home.service: Failed with result exit-code.” If I do have a system directory that’s not mounting, does that mean it’s time for a new drive? -GEF
OK, back in business, looks like there was corruption on my home partition (ext4), not the root partition (btrfs). I ran fsck on /home and agreed to all its suggestions, so now I can boot. It looked like the bits were plasma config files like kwinrc, so I may have lost some desktop settings. Is an occasional glitch like this nothing to worry about, or a sign that the cheapest SSD I could find was a little too cheap?
In order to decide this one, I recommend having smartd run in the background (maybe you have it installed and active already) or do a »smartctl -a /dev/yourSSDdevice« as root.
I found it a good idea to redirect each smartctl output into a separate file (I do it monthly with my backups) and keep those logs around. It’s informative to compare changes over time:
