[jcifs] New user questions

[Michael B Allen]

On Sun, 17 Mar 2002 01:33:35 -0600
Christopher R.Hertel <crh at ubiqx.mn.org> wrote:

> > > [Mike:  I'm missing something.  The second attempt has a TrnID of 2, not 1
> > >         like the first.  Shouldn't there be three tries, all with the same
> > >         TrnID?]
> > 
> > Are you sure thr TrnIDs are supposed to be the same? That may have been
> > true at one time but it's possible it's just wrong now.
> 
> Running Samba's nmblookup I see all three attempts (for a non-existant
> name) using the same TrnID.  Likewise using nbtstat on a W/98 system.
> 
> :)
> 
> > Regarding retries jCIFS does 2 tries and waits three seconds for each
> > response instead of 3 tries for 5 seconds each accoring to the spec. That
> > cuts the wait time in half but reduces the likelyhood of success by much
> > less than half (probably 0.0000000001% :~). The bottom line is 15 seconds
> > is ridiculous.
> 
> I agree that the 15 seconds is silly.  I would do something like:
> 
> Try
> [1 sec]
> Retry
> [2 sec]
> Retry
> [3 sec]
> 
> Still totals 6 seconds max.

I don't want to complicate the algorithm too much for what would be a
*very* small gain. If it's really an issue for someone they have complete
control over it with the retryTimeout and retryCount:

  jcifs.netbios.retryTimeout = 5000
  jcifs.netbios.retryCount = 3

>  If all three use the same TrnID then a reply
> to the initial query can take up to 6 seconds to show up and still be
> valid.

Agreed. I'll fix is some day but it's not a super high priority right now.

Mike

-- 
May The Source be with you.


#################################################################
#################################################################
#################################################################
#####
#####
#####
#################################################################
#################################################################
#################################################################