Subject: Re: Raid reconstrution fails
To: None <netbsd-users@netbsd.org>
From: Greg Oster <oster@cs.usask.ca>
List: netbsd-users
Date: 03/04/2004 09:46:43
Christoph Kaegi writes:
> On 04.03-07:31, Greg Oster wrote:
> > > it says in the logs:
> > > 
> > > 
> > > -------------------------------------- 8< -------------------------------
> ----
> > > ---
> > > Mar  4 09:11:44 sstffw /netbsd: Rebuild: 0 0
> > > Mar  4 09:11:44 sstffw /netbsd: About to (re-)open the device for rebuild
> ing:
> > >  component0
> > > Mar  4 09:11:44 sstffw /netbsd: raid1: rebuilding: raidlookup on device: 
> comp
> > > onent0 failed: 2!
> > > -------------------------------------- 8< -------------------------------
> ----
> > > ---
> > > 
> > > What does that mean?
> > 
> > You can't rebuild onto "nothing".
> > 
> > > The other raidsets (raid0 and raid2) are ok.
> > 
> > For whatever reason, the component label for one of the components in 
> > raid1 has gone missing, and the autoconfig code was unable to find it
> > on reboot.
> > 
> > The output of 'raidctl -s raid0' and 'cat /var/run/dmesg.boot' would help
> > diagnose this further.
> 
> -------------------------------------- 8< -----------------------------------
> # raidctl -s raid0
[snip]

Ok, that one's fine..  So wd1 is alive.

> -------------------------------------- 8< -----------------------------------
[snip]
> Kernelized RAIDframe activated
> RAID autoconfigure
> Configuring raid0:
> RAIDFRAME: protectedSectors is 64
> RAIDFRAME: Configure (RAID Level 1): total number of sectors is 2097408 (1024
>  MB)
> RAIDFRAME(RAID Level 1): Using 6 floating recon bufs with no head sep limit
> RAID autoconfigure
> Configuring raid1:
> RAIDFRAME: Configure (RAID Level 1): total number of sectors is 2097536 (1024
>  MB)
> RAIDFRAME(RAID Level 1): Using 6 floating recon bufs with no head sep limit
> RAID autoconfigure

Oh.. Was expecting this to say something else, but I'm too used to 
the output of -current 
:-}

> -------------------------------------- 8< -----------------------------------
> 
> 
> Hm, upon further thinking: 
> 
> Do I have to raidctl -a /dev/wd1b raid1 and then raidctl -vF component0 raid1 
> ?

Yes... that will do what you want/need.

Later...

Greg Oster