[oi-dev] SSD-based pools
Udo Grabowski (IMK)
udo.grabowski at kit.edu
Mon Oct 13 23:24:00 UTC 2014
Check the details of the checksum error with 'fmdump -eV'. If you see
a lot of errors where the wrong checksums are all zero, you probably
have the same problem as I had with my mirror rpool (a unknown bug).
But I've not heard of this occuring on RAIDZ before.
On 14/10/2014 01:12, Andrew M. Hettinger wrote:
> Ok.
> I've now tried to run this on another machine (no hardware in common, save the
> SSDs), with
> interlopers, still generating checksum errors. I'm using iozone to load it up
> for testing (I was
> initially asked to do this to demonstrate how much of an improvement we could
> expect).
>
> I was wondering if you guys had any more thoughts or could do a similar test?
>
> Andrew Hettinger
> http://Prominic.NET | Skype: AndrewProminic
> Tel: 866.339.3169 (toll free) -or- 1.217.356.2888 x. 110 (int'l)
> Fax: 866.372.3356 (toll free) -or- 1.217.356.3356 (int'l)
>
> "Schweiss, Chip" <chip at innovates.com> wrote on 09/30/2014 07:49:44 AM:
>
> > From: "Schweiss, Chip" <chip at innovates.com>
> > To: OpenIndiana Developer mailing list <oi-dev at openindiana.org>
> > Date: 09/30/2014 07:58 AM
> > Subject: Re: [oi-dev] SSD-based pools
> >
> > On Mon, Sep 29, 2014 at 6:14 PM, Andrew M. Hettinger <AHettinger at prominic.net
> > > wrote:
> > Bob Friesenhahn <bfriesen at simple.dallas.tx.us> wrote on 09/29/2014
> > 05:57:26 PM:
> >
> > > How would ZFS know if the data stored is "incorrect" from the user's
> > > perspective?
> > >
> >
> > Presumably because the checksum is wrong.
> >
> > Exactly, if the data is returned incorrect from the SSD ZFS will
> > detect it via checksum. It will then rebuild from the raidz1
> > parity, if that fails it will return a data read error.
> >
> ....
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 5285 bytes
Desc: S/MIME Cryptographic Signature
URL: <http://openindiana.org/pipermail/oi-dev/attachments/20141014/ed651b07/attachment-0005.bin>
More information about the oi-dev
mailing list