[ENBD] Doubled requests

Arne Wiebalck wiebalck at kip.uni-heidelberg.de
Tue Jan 13 05:19:16 MST 2004


On Tue, 13 Jan 2004, Peter T. Breuer wrote:

> I believe it's an accounting error. If it isn't, I would be very
> interested.

I don't think it's an accounting error. I am quite good at counting ;-)

> Yes - the server maintains a cache of requests seen in order to avoid

The cache contains the IDs of requests dealt with in the last time?
How large is it?

> the possiblilty, and replies using the old error value instead of doing
> the write again.

That would be nice for me ...

> It is quite possible that a timeout will occur and then the server
> sends the ack. This will indeed cause a request to be sent twice. The
> server should catch the second transmission. But I agree, the cache is
> probably too small, and the semantics needs tweaking so that it
> preferentially retains records of requests which took a long time to
> process, in case they are retransmitted.
>
> So the retransmit is real?

I am not sure whether it's retransmitted over the net. All I can say is
that from time to time I see too many requests arriving at my device.
I see this (aside from counting) by having "wholes" on my device: if a
block is written twice the effect on my resource is zero.
Maybe the server does writes mutiple times in certain circumstances?

The rate is about 1 in 100.000 requests.

> If there is a retransmit, kernel dmesg should say. It will record a
> "rollback", at least.

No messages in /var/log/messages or by dmesg.

> You can avoid them by not allowing rollback (show_errs=1).
>
> Very interesting.
>

My device seems to work fine if I write directly to it, but I will check
again ...


arne


**********************************************************************
 Arne Wiebalck                        Kirchhoff Institute for Physics
                                           Technical Computer Science
                                                  INF 227, Room 3.315
                                                   D-69120 Heidelberg

 e-mail: wiebalck at kip.uni-heidelberg.de
 web   : http://www.kip.uni-heidelberg.de    Tel.: (+49) 6221/54-9816
**********************************************************************




More information about the ENBD mailing list