File system loss due to software error or human error

From: Schepers, Jan (Jan.Schepers@atosorigin.com)
Date: Fri Apr 19 2002 - 05:36:28 EDT


Hello folks,

I am trying to do some probability calculations of the loss of a file system
on a RAID 1,3,4 or 5 configuration.
It is relatively easy to calculate the probability of the loss of such a
File System due to a double disk failure (Garth A. Gibson, Redundant Disk
Arrays: Reliable, Parallel Secondary Storage, MIT Press, Cambridge,
Massachusetts, 1992.).

However a file system can amongst others also be lost by software errors or
human errors.
I'm trying by means of your response to get a feeling for:
0 probability per file system
        - how often does it happen
        - how many file systems
0 the impact
        - on how many users
        - for how long
        - size of the FS
0 the measures to prevent this from happening.

It is trivial that causes for human error are unintentional "rm -r *" or
"newsfs" on the wrong special device.
Software errors may appear to have a more theoretical character, but have
you never been hit by "holes" in files or directories with missing or double
entries just because of a driver with an software error.

I will summarize but respect the privacy of those who indicate it.

regards,

Jan Schepers



This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:48:38 EDT