On Mon, 7 Feb 2000, Brian Denheyer wrote:
> >>>>> " " == <[EMAIL PROTECTED]> writes:
>
> > i believe you should be fine just letting it do its thing on its own.
> > cat /proc/mdstat
>
> Thanks, I should have known that.
>
> Yes, that's exactly what I did. Everything has returned to normal.
>
> When a reset or other catastrophic even happens, I was under the
> impression that one of the disks had to be chosen as the "original"
> and a copy made to the other disk to resync them.
>
> Is this essentially what the re-sync daemon does ?
Under raid-1, there is always one disk that is designated as the 'master'
until that disk fails and the other becomes the master until a resync is
executed where it may or may not swap things back again. Raid is designed
to protect you against disk failures, not system failures.
If the power fails (or some catastrophic event), then /both/ drives will
be marked unclean on bootup (in effect, its marked unclean as soon as its
running, and marked clean if its shut down properly). In this case, the
raid system will probably choose the original master, even though neither
drive is 'good'. Ie... you'll /still/ get file system errors.
If you want to protect yourself against system failures, then you'll
probably want to couple raid with a transaction filesystem like Rieser.
(Dont ask me for details, I only know that its beta, and its Good :)
("`-/")_.-'"``-._ Chris Cogdon <[EMAIL PROTECTED]>
. . `; -._ )-;-,_`) Purecom Pty. Ltd.
(v_,)' _ )`-.\ ``-' 4/20 St. Huberts Rd. Carnegie. VIC. 3053.
_.- _..-_/ / ((.' Phone: 0413 670 979
((,.-' ((,/ fL
