[Linux-cluster] File system checking

[Scott Becker]

Does anybody know the best way to check that a filesystem is healthy?

I'm working on a light selfcheck script (to be ran once a minute) and 
creating a file and checking it's existence may not work because of 
write caching. Checking the mount status is probably better but I don't 
know. I've had full filesystems and once the kernel detected an error 
and remounted read only. Other times, when a drive in the raid array was 
slowly failing, it would hang on all IO for a spell.

If there's an existing source module or a script somebody is aware of 
that would be great.

    thanks
    scottb