On 2005.07.21 05:40:06 +0000, Mike Christie wrote:
>
> Wang, Zhenyu wrote:
> >
> >misunderstood the meanings... this's wrong, sorry.
> >I'll send my updated data digest patch later, after resolving the unalign
> >issue.
> >
>
> zhen - I forgot to mention for your data digest patch, keep in mind that
> someone can be writing to the data's memory while you are calculating
> the digest or while you are sending it :( (look over the discussion on
> linux-scsi from when sfnet was being reviwed long long ago).
bad...in which path it'll try to write the mem? Doesn't this also apply to
header digest? I really did see some of those digest errors on target side,
but don't know why...
> To not have
> to fail the IO when this happens you will want to try and use
> DID_IMM_RETRY. For tape you may not want to retry the IO though.
currently, with iet, initiator gets tcp connection fail when target detect
digest error. then we'll reconnect and try again, during which we have
chances to send the right digest. maybe we have no chance to use DID_IMM_RETRY
now?
> To
> complicate matters you also may not want to use DID_IMM_RETRY if the
> casue of digest mismatch is due to a transport problem. We had cases
> where something went bad on the target and we kept gettting back bad
> digests but we kept retrying the the IO on the same path (DID_IMM_RETRY
> does not take FAILFAST into account).
Do we need some stats things?
thanks
-zhen
-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
|
