Re: drive controller Q?
On Wed, Sep 14, 2022 at 03:43:16PM -0400, gene heskett wrote:
> On 9/14/22 14:55, Dan Ritter wrote:
> > gene heskett wrote:
> > > Greetings all;
> > >
> > > Does anyone have experience with this controller card?
> > >
> > > https://www.newegg.com/p/14G-04YB-00003?Item=9SIB2XHHUE3880
> > >
> > > Specifically, whats my chances of moving an existing software raid10's 4
> > > Samsung 1T's to it,
> > > and then attaching 4 more 2T drives to it too, to create a separate 4T
> > > raid-10 for amanda?
> > >
If your existing software raid is mdadm, quite good. YOu're pinning quite a
lot on one card though - as mentioned, you may suffer bandwidth poverty :)
> > > Without any data loss if possible?
> > I don't have any experience with this one. I don't know what
> > SATA chipset it is using, but there aren't many that the kernel
> > doesn't already support.
> >
> > That said, it is a straight SATA3 board, not a RAID board, so
> > there will be no difficulty in moving mdadm, btrfs or zfs RAIDs
> > over to it.
> >
> > I would point out that you can't actually fit 16 x 3Gb/s worth
> > of bandwidth over one PCIe lane; if this is v1 PCIe, you have a
> > total of 250MB/s available. That's probably fine for four
> > spinning disks doing backup duty.
> >
> > Here's a 4-port model with named PCIe v2 support and a
> > recognizable SATA chipset, for slightly less money:
> >
> > https://www.newegg.com/syba-si-pex40064-sata-iii/p/N82E16816124064?Item=N82E16816124064
> >
>
> Can I make a 2nd raid 10 from two separate controllers? 2 drives on the mobo
> controller and 2 on the
> plugin controller? I'd think that could lead to mix-n-match problems given
> udevs penchant for
> shuffling drives.
>
mdadm should use internal blkid and shouldn't care if you shuffle drives.
Mixing and matching two controllers almost certainly won't work for other
reasons: there's a reason that add in cards are usually intended to
support one set of RAID.
> Too many options......
>
> Cheers, Gene Heskett.
> --
>
With every good wish, as ever,
Andy Cater
Reply to: