Re: How to mark dbspace "down"

In article <5lh487$bjo@drn.zippo.com>, ibogorad@lhr-sys.dhl.com writes > >Hi folks, > >I am evaluating OnBar in Online 7.22, generally it seems OK - as easy as ontape >to use and in most cases it works. >However , if a non-critical dbspace is not down, it does not allow to do warm >restore - says - " Excuse moi, the dbspace is up and I am not going to run >restore >on it" > >Now , I have a non-critical dbspace (i.e dbspace which is not rootdbs and does >not contain logical or phys logs). It consists of one chunk, which is cooked >file >for simplicity. Mirroring is disabled in this instance. > >I create some dummy database in this dbspace, populate with some data and then >create level 0 archive ( onbar -b -w ). > >Then I do something nasty : say delete this chunk on Unix level ( yeap, just rm) >or corrupt it : banner " I'll be back.Schwarz"> /my/poor/chunk . > >After that I do all sort of things to assure Online that something bad has just >happened: force checkpoints, run onchecks, bring the instance down and up, etc. >Online marks the chunk down, but it thinks that the dbspace is perfectly well, >though this bleeding dbspace consists of only one chunk,which is terminally ill. > Online will - it is the chunks that get marked as down not the dbspace. At the physical level the dbspace is is the reserved pages in the root dbspace (PAGE_DBSP1 and PAGE_DBSP1) which map extents to chunks.... >Therefore I stiil cannot do warm restore for this dbspace, only full (cold) >restore >is possible. > ??? I have never come across this before. What error you do get when you try to do a warm restore? >Are there any sure ways to tell Online that a dbspace is down? > >All ideas will be appreciated. > >Cheers >Ilya > > -- David Williams