Re: [PATCH 0/4] iser, new implementation

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Sun, 17 Oct 2010 16:06:13 +0200
Alexander Nezhinsky <alexandern@xxxxxxxxxxxx> wrote:

> On 10/14/2010 03:07 PM, FUJITA Tomonori wrote:
> > On Wed, 13 Oct 2010 15:31:08 +0200
> > Alexander Nezhinsky <alexandern@xxxxxxxxxxxx> wrote:
> > 
> >> + added sending NOP-IN. While not mandatory in iscsi, it is quite important 
> >> in iSER (specifically in IB case). When a connection is broken, IB does not 
> >> automatically generate error events if there is no operation to
> >> fail.
> > 
> > I plan to implement this for tcp too.
> > 
> >> Thus it is quite possible to have stale connections which will never clear,
> >> if the target has nothing to send (obviously, nothing will be received
> >> from now on). Sending NOP-IN periodically guarantees that such connections
> >> will be removed, as the send op will fail with an unrecoverable error.
> >> Sending NOP-IN is enabled by default. To disable, invoke tgtd with 
> >> "--iser nop=off". 
> > 
> > The feature should be supported for tcp. I think that configuration
> > per target is good enough.
> > 
> >> Uses a timer: do we need a centralized timer management in tgtd? 
> > 
> > usr/work.c doesn't work for you? isns code uses it.
> > 
> > When you finish bidi, I'll merge this.
> > 
> > btw, please use a different subject per mail next time.
> 
> Ok, i'll remove the current NOP-IN code and implement it using 
> usr/work.c primitives. 

If necessary, please send patches to improve user/work.c. I think that
you need a feature to cancel a task at least.
--
To unsubscribe from this list: send the line "unsubscribe stgt" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html


[Index of Archives]     [Linux SCSI]     [Linux RAID]     [Linux Clusters]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]

  Powered by Linux