[OmniOS-discuss] [smartos-discuss] SuperStorage Server 6047R-E1R36L SAS Question

Keith Wesolowski keith.wesolowski at joyent.com
Mon Nov 11 17:03:42 UTC 2013


On Mon, Nov 11, 2013 at 03:05:00PM +0100, Tobias Oetiker wrote:

> We are looking at purchasing a new box. According to
> https://github.com/joyent/manufacturing/blob/master/parts-database.ods
> it seems that Joyent is using
> 
> Supermicro SuperStorage Server 6047R-E1R36L with 7K3000 (and maybe
> soon 7K4000) disks.
> http://www.supermicro.com/products/system/4u/6047/ssg-6047r-e1r36l.cfm
> 
> We asked our system integrator for an offer on these boxes and he
> said, that he has several in the field and they wer cool, BUT he
> has seen problems with the on-board LSI 2308 and the backplane, and
> that he would recommend installing 5 extra LSI SAS controllers to
> directly connect each sas disk to a controller port ...

That's not how this system works.  The backplanes have SAS expanders and
as far as I know there is no way to get an alternate backplane.

The only reason to want to bypass the expanders is if you are using SATA
devices.  We don't use them in this system, and do not recommend them.
We've had no expander-related issues in our SAS systems.  The only
systems we use SATA devices in are the ones with expanderless 16-bay
backplanes (Richmond-A/Richmond-C).

Anything more useful will require actual data as to what "problems" have
supposedly occurred and under what circumstances.  A crash dump, if a
panic occurred, is essential.  Error messages, anything at all.
Otherwise this feels a lot like an attempt to sell you 4 extra HBAs and
a giant unmanageable wad of internal cables.

> Is this the reason that joyent lists all the firmware versions and
> one should make sure to use exactly these versions ?

Yes.  Never deviate from known-to-work firmware revisions unless you
encounter a bug that is specific to this version and *known* to be fixed
in a different one.


More information about the OmniOS-discuss mailing list