Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 26 Oct 2005 16:05:39 +0200
From:      Palle Girgensohn <girgen@FreeBSD.org>
To:        "Matthew N. Dodd" <mdodd@FreeBSD.org>
Cc:        hardware@FreeBSD.org
Subject:   Re: ida problems, one disk broken = system very slow
Message-ID:  <66AA0695B4793976B82E0A75@rambutan.pingpong.net>

next in thread | raw e-mail | index | archive | help


--On onsdag, oktober 26, 2005 09.41.34 -0400 "Matthew N. Dodd" 
<mdodd@FreeBSD.ORG> wrote:

> On Tue, 25 Oct 2005, Palle Girgensohn wrote:
>> A customer has a machine with four disks in RAID 10 using the ida(8)
>> controller on FreeBSD-4.11.
>
> ftp://ftp.jurai.net/users/winter/idacontrol.tar would solve a few of your
> problems on 5.x or better but getting it to work on 4.x requires patching
> and recompiling.

What kind of patching is required for 4.x? Any ideas or pointers?

>> Now one disk is broken. I have some questions:
>>
>> 1. How can I detect that a disk in a raid cluster is broken? It seems
>> natural to me that the raid driver would log info about a broken disk,
>> but I have not seen this happen with any raid controller driver.
>
> Most array drivers have a userland utility that allows inspection of
> array status.  The utility for ida(4) isn't in the base system yet.

Oh. A port would be great. :-)

Any ideas on something similar for the ciss(4) controller?

>> 2. The system is extremely slow and hardly usable right now. The
>> customer are still waiting for a replacement disk. Is there any way to
>> get the system to just ignore the broken disk instead of trying to use
>> it and fail. I get thousands of "ida0: soft error" in messages log. If
>> the system would realize that it had problems with the disk, and ignore
>> them, perhaps it wouldn't become unsable,
>
> Well, you could rate-limit the error messages, or comment out the message
> entirely.

You mean from /etc/syslog?  That is not the problem, it seems to be some 
kind of timeout for every error that is the problem, I'm not sure. CPU runs 
at 100% constantly, anyway, but mostly running postgresql and java, not 
syslogd.

>
>> 3. Will the array rebuild automatically once they insert the new disk?
>
> Depends on the BIOS setting.

OK, thanks.

/Palle

>
> --
> Lord, wheresoever this rifle goest, guide my hands upon it.
> Let a shot never be fired in anger or error.
> Bless it to your righteous service.
> In your mercy, Thine is the kingdom Amen.







Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?66AA0695B4793976B82E0A75>