Subject: Re: XS_SELTIMEOUT
To: Jason R Thorpe <thorpej@wasabisystems.com>
From: Matthew Jacob <mjacob@feral.com>
List: tech-kern
Date: 02/20/2002 17:31:28
> On Wed, Feb 20, 2002 at 04:07:02PM -0800, Matthew Jacob wrote:
>
> > Are there any objections to making the XS_SELTIMEOUT case the same as
> > XS_TIMEOUT? (i.e., if there retries, retry the command that failed instead
> > of just returning an EIO)?
> >
> > There could be a number reasons that selection failed (temporarily)- and this
> > is just for SPI (regular old SCSI). For Fibre Channel, this could happen a
> > lot.
>
> Seems fine to me ... but wouldn't XS_SELTIMEOUT be a good reason to
> automatically config_detach() a device?
*After* retries, yes. This allows you to set a policy about retries before
giving up (e.g.,
options SDRETRIES 100
)
in, say, very noisy SAN environments where disks go away a lot (but reappear
just about as quickly).
-matt