Skipping e2fsck / fsck
fsck always force starts when I had planed to use the computer for something, so I want to abort it with Ctrl+C, and then later run it during a break or something.
But Ctrl+C does not cancel anymore. I have a file /etc/e2fsck.conf Code:
[options] Why? |
Is it perfectly OK with you if you crash the file system and must reload the machine?
If not, the first question should be "why does it need to check the file system?"! What kind of file system do you have? |
It says it does this fsck run every 30 days
> What kind of file system do you have? ext3 |
Quote:
tune2fs -c 0 /dev/sda2 If you wanted to set a file system to be checked by fsck only every 100 mounts when no error has been detected you would use: tune2fs -c 100 /dev/sda2 see: man tune2fs ---------------------- Steve Stites |
Let it run. If it's running, it's running for a reason - an error has been detected, unless it is a scheduled run as per post #4.
If you're seeing this a lot, then you need to solve the underlying problem (perhaps associated with the way you shut down the computer). |
If you move to EXT4 file systems you will find that a normal fsck rarely runs, and when it does it runs much faster.
Playing back journals into extents is really remarkably quick for the average case. Exception: in the (hopefully rare) case that you have media failure issues (bad spots appearing on disk) it can take a long time, but is slightly more likely to recover. If you see such events it generally means it is time to invest in a new hard drive, and SOON. |
Quote:
Quote:
Quote:
Quote:
But for an update I would need to make a backup Quote:
Although it never finishes :( |
Quote:
------------------------ Steve Stites |
Quote:
According to man tune2fs: Quote:
|
If during boot a mount command encounters a file system that was not unmounted cleanly at shutdown then the boot sequence runs fsck against that partition. I have had that happen occasionally, more frequently since systemd began controlling shutdowns, and fsck then runs at boot time and always clears the problem. I also have occasionally had an external drive not unmount cleanly causing the next mount to fail. In that case I run fsck manually against the external drive.
Running fsck occasionally against a cleanly mounted drive is supposed to fix errors that mount did not detect. I have never had that happen in seventeen years of Linux usage so I consider running fsck against a partition which has no reason to be suspected of errors to be a waste of time. -------------------------- Steve Stites |
Quote:
|
If it running on schedule bothers you, you can always set it to run on next boot and then reboot the machine. This resets the counter back so that your next SCHEDULED check would be 30 days out again.
Like the EXT4 developers (and every other file system developer I know) I would recommend scheduling a regular run. It does not have to be every 30 days. On SERVER class hardware every 60 or 120 days might be fine: depending upon the applications and use on that server. Laptops are notoriously more prone to errors. Recent SSD storage may change that, but we have not had enough time with enough subject machines for a proper sampling. |
FWIW, Red Hat no longer defaults to automatic scheduled fsck for filesystems, neither by time nor by mount count. Those automatic checks have a nasty habit of happening at awkward times ("I need to take the database server offline briefly for a quick reboot. OK?" ... "Oops!"). You can, of course, set up those automatic checks, but they are disabled by default when the filesystem is created.
|
Quote:
So I wanted to watch a video during breakfast and my laptop is like "no, you have to watch fsck till your breakfast is finished". Or in a train that is late, trying to look up the hotel's check-in policies and then you only see the fsck, and end up having to pay a 80€ late fee. |
A complete fsck of my laptop drive takes less than five minutes. What are you running on that takes fsck longer than breakfast? (Or do you eat VERY fast?)
|
All times are GMT -5. The time now is 11:05 PM. |