Re: Raid Stripe Size.....

>8 drive RAID 10 works very well.
>chances of losing a second disk on a mirrored pair are fairly remote.

I recently had a scary one (on hp-ux, but a raid issue).

13 drive RAID 5, one hot spare.

I go to startup db, get a corrupt block one on controlfile. Copy over another, same issue... can't shutdown, have to abort... run orapw, get interpreter "/usr/m/pa20^ÙL/dld.sl" not found exec(2): could not load a.out

Hmmmm, start looking around and find things buried deep down directory trees like
--wxrw---t+ 46477 814614889 9308519 6755399441062032 Oct 21 1921 SOL_C2R4.HTML -rws---r-x+ 27748 1949254172 1953787657 5797666986988568015 Jun 14 2033 SOL_C2R8.HTML -rw-r--r-- 1 oracle dba 7146 Jan 30 2002 SOL_C3E0.HTML

Turned out an impatient manager had replaced 2 disk drives, not waiting for the parity to finish rebuilding between the 2.

The scary part was the resemblence between this and the old ripper virus. It could have just as easily not hit the control files and how would I know what was corrupted in the db?

jg

--
@home.com is bogus.
"We are not our customers' accounting departments."
http://www.signonsandiego.com/uniontrib/20050303/news_1b3mkt.html