Checking the file system every 22 boots

[Ioannis Vranos]

Bob McConnell wrote:
> 
> It is necessary. About one in five passes it finds a few orphaned inodes 
> on my systems. You can adjust the frequency by changing a value in 
> /etc/fstab. But if your system stays up most of the time, there is also 
> a maximum days between runs buried somewhere. So when I took down a 
> server that had been up for 200 days to install additional drives, all 
> of the partitions on the existing drives were checked when it booted 
> back up.


If I recall well, under Scientific Linux/CentOS/(Red Hat EL) x86, no
thorough fsck is made. Under versions 4.x if the system powers off
uncleanly it recovers by using the journal and it asks you if you want
to perform a thorough check with a time-out of some seconds, after which
it proceeds without thorough check. I think something like this happens
with versions 5.x x86.

Since ext3 is very robust in its default "ordered" journalling mode, why
a thorough check is enforced every 22 boots?