Can I cancel fsck safely?

[nsk7even]

Hello,

due to an unbelievably concentrated combination of rare circumstances my main system is running a fsck on my main data partition of 1.6 TiB (1.4 TiB used) since 2 days now.

Can I cancel this or will it make things worse?

The Story:

• My X freezed due to whatever reason while gparted was shrinking this partition from the beginning
• I then checked via ssh that gparted still was running - it was
• After work I got no ssh connection anymore
• Did reboot via SysRq REISUB
• Everything booted up as usual and first I was happy
• But then I noticed that at least one directory was lost
• My main data was there so I did a backup
• Since the missing directory was only containing installed program data I thought everything is fine and I proceeded to work as usual INCLUDING writing additional data on this partition
• After next reboot fsck warned of corrupted partition
• I decided to let fsck do the check for maybe saving me the time to reinstall those programs
• At that time I forgot the newly written data to this partition which had no backup until now (digicam photos and videos)

What do you think, does fsck make things good or bad just now?
Does it correct all wrong entries with respecting the new data?
Or maybe it undoes all the gparted works now because maybe gparted was not able to cleanly close its modifications?

Any advises or experiences appreciated..

[seawolf167]

fsck is set to automatically scan and check every 30 boots or if a problem has been detected.

I personally would let it finish, though if you want to stop it, I would highly suggest immediately imaging your system with something like Clonezilla in case you end up needing to do a recover, then booting back up and starting a fresh fsck disk check or seeing if you can login and maybe check your syslog messages.

[bodhi.zazen]

due to an unbelievably concentrated combination of rare circumstances my main system is running a fsck on my main data partition of 1.6 TiB (1.4 TiB used) since 2 days now.

2 days is a long time. I would boot a live CD and run fsck manually to see what the problem may be.

Better to fix the problem, or if that is not possible, back up your data and re-format.

[nsk7even]

Thank you for your answer.

So fsck can proceed later when I interrupt it now?

The problem is, I can not read the output because it is scrolling too fast on the screen.
Therefore I can not tell what progress it is and how long it possibly will need to finish...

At the beginning it prints something like "Inode 744763?? ...."
Can I estimate the progress based on this number?

[nsk7even]

Ok, found out via Google that I can multiply the Inode Number with the Inode Size to get the number of Bytes.

Now the magical question: what would be the standard Inode size for a 1.6 TiB Ext4 partition created by Gparted or Palimpsest?
Which one I can not remember since I use both :/

[seawolf167]

So fsck can proceed later when I interrupt it now?

You can just re-run it from a Ubuntu LiveCD terminal no problems. See the fsck link I posted earlier for a how-to.

At the beginning it prints something like "Inode 744763?? ...."
Can I estimate the progress based on this number?

Not unless you are the drive manufacturer or have a bunch of info available to you on the drive specs.

[seawolf167]

Ok, found out via Google that I can multiply the Inode Number with the Inode Size to get the number of Bytes.

Now the magical question: what would be the standard Inode size for a 1.6 TiB Ext4 partition created by Gparted or Palimpsest?
Which one I can not remember since I use both :/

Default inode sizes:
128 bytes for ext3 format
256 bytes for ext4 format

[nsk7even]

OK thank you for those informations!

Assuming 256 bytes inode size and me reading the inode number correctly I calculated the progress to be at 17,7 GiB which means it would take another 78 days to complete. I am too impatient for that

Last question:
This is not my root, boot or other partition vital to the system - would it harm to cancel this now and then proceed normal bootup to finally start fsck from the running system afterwards?
Except the fstab automount there should be not much difference to a live cd or is there anything evil I forgot?

[nsk7even]

Something is wrong in my calculation ... just compared with my 2.05 TiB ext3 partition on my server that has an inode size of 256 as well.

nsk@server7even3:~$ sudo tune2fs -l /dev/md127 | grep Inode
Inode count: 137363456
Inodes per group: 8192
Inode blocks per group: 512
Inode size: 256

Comparing roughly the inode count, my main system passed the 50% a while ago, so my simple formula "inode count * inode size = bytes" is wrong ... I will do a little more Googling, maybe I have to respect the other inode values as well - maybe they are not written there because they look so beautiful...

[seawolf167]

You might just want to check to ensure the scrolling text doesn't indicate that it is repairing the file system and restart into a Live environment to check the drive's S.M.A.R.T. status and do a (hopefully quick) fsck from there.