Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 11 May 2005 11:51:42 +0200
From:      Peter Orlowski <petero@itp.physik.tu-berlin.de>
To:        freebsd-stable@freebsd.org
Subject:   Re: State of gvinum RELENG_5 or RELENG_5_4 ?!
Message-ID:  <20050511095142.GA12629@ella.physik.tu-berlin.de>
In-Reply-To: <4280EE3C.8050505@jim-liesl.org>
References:  <8b6eae96050510071037d3b6e4@mail.gmail.com> <4280C277.4030904@jim-liesl.org> <27C1250C-9BFD-4FA8-876C-09DF5504B058@isncom.com> <4280EE3C.8050505@jim-liesl.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, May 10, 2005 at 11:24:12AM -0600, secmgr wrote:
> Gabor Esperon wrote:
> >
> >How reliable is the gmirror subsystem?
> 
> gmirror seems fine as best as I can tell.  I've been running it for a 
> few months on sata drives.

I had gmirror running on two IDE drives for system and two SATA
drives for user data. That was RELENG_5_3. When one of the SATA 
drives broke (later I got errors like 

ad4: FAILURE - READ_DMA status=51<READY,DSC,ERROR>  

on fscking) the whole system just stopped. I could switch
consoles but nothing else. Nothing in the logs either. I believe
it had lost all it's file systems.

After power cycling, both mirrors were found but marked as broken.  
Gmirror chose the defective SATA disk as the intact one and started
rebuilding the mirror from that - but fscking the mirror failed,
obviously, so I ended up with one disk that was broken, but marked intact
by gmirror and one disk the other way round.

I had to erase the gmirror metadata on the intact disk to get it
to work again. 

I'd say gmirror will save your data but it won't save you from some
downtime...

Greetings, 	Peter



-- 
Peter Orlowski			
Institut für Theoretische Physik
Technische Universität Berlin



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050511095142.GA12629>