RE: timers

["Douglas Otis" <dotis@sanlight.net>]

Julo,

While remaining within the minority, there is a need to prevent overlapping
retry mechanisms.  IP networks introduce latency delays that warrant
considerations with respect to Fibre Channel E_D_TOV (Node Life) and R_A_TOV
(Fabric Life) values.  The total path delay between a source and a
destination FC port should be less than a value of 3 seconds (10 seconds at
a maximum) with normal Fibre Channel timeout values.  I requested additional
negotiation for a connection probe NOP-PING recommended at once every 10
seconds while idle and with a response (other timers) pending to avoid the
SCSI layer retry.  Perhaps you should expand these timer values to include a
retry after X and a quit after Y.  The probe could be retry after X and quit
after Y as well.  These time values should not be left open and eventually
recommended values should be found in the appendix.  With default timeouts
of TCP being 2 hours, other means of timeout should be offered.  What is a
short time anyway.  When will the target know the initiator has given-up?

As written:
   At any timer expiration (timeout) the initiator MUST resend the
   command or task management PDU with the restart bit set.

   Timers support recovery for gateways between an iSCSI transport and
   an unreliable transport (e.g., FCP).
...
   If all the connections of a session fail and can't be reestablished
   in a short time or if initiators detect protocol errors repeatedly an
   initiator may choose to terminate a session and establish a new
   session. It will terminate all outstanding requests with an iSCSI
   error indication before initiating a new session.  A target that
   detects one of the above errors will take the following actions:

      - Reset the TCP connections (close the session).
      - Abort all Tasks in the task set for the corresponding
      initiator.

Doug

> Dear colleagues,
>
> I did not hear anything about the timers for our gateway specialists and
> I've heard many of you screaming against them (truly only 1 or 2 but this
> is a VAST MAJORITY those days :)-).
>
> If nobody (except one) speaks for them I might be tempted to take them out
> completely.
>
> Julo