Re: Antwort: Re: [Linux-decnet-user] fal problem (but Phase V)

[Patrick C. <pa...@pa...>]

Thanks for the reply. I'm now rather confused. My current kernel (2.2.16 with 
the patch I last sent you) works fine to my VMS 7.2 DECnet/Phase V system.

That's ot to say I disbelieve you, I certainly do. Your symptoms are consistant
with my reasons for issuing the optfil patch in the first place. Of course
when I issued that patch it was because I was seeing those problems with
my phase V system. But now I'm not :-(

This needs some careful thought and possibly some tcpdumps. If you have chance
could you please send me a tcpdump of the problem communication from VAX to
Linux and a similar dump showing communication between VAX and another VAX.
Please do tcpdump -x -s1600 as the packet decoding for DECnet is not very good!

Just not sending optfil when communicating on some directions is not an option
I'm afraid because it's not always obvious what the direction is at that
level. There must be something else, more fundamentally wrong in the packets.

If anyone else on the list is using Phase V, please could you post your
experiences here so we can compare. Thanks,

Patrick


On Thu, Jun 15, 2000 at 06:56:41PM +0200, hei...@po... wrote:
> 
> 
> 
> Hi Patrick,
> 
> we applied your patch and than we tested the communication.
> 
> Thats what we found
> 
> before your Patch
> 
> initiator
> VAX                    VAX   -----> Linux      problem
> VAX                    VAX  <-----  Linux      OK
> 
> Linux                   VAX  ------> Linux      OK
> Linux                   VAX <------  Linux      OK
> 
> with your Patch
> 
> initiator
> VAX                    VAX   -----> Linux    OK
> VAX                    VAX  <-----  Linux    OK
> 
> Linux                   VAX  ------> Linux   OK
> Linux                   VAX <------  Linux   problem  ***
> 
> 
> the copy of files initiated by the VAX works fine now, but ....
> 
> the copy of files from Linux to the VAX initiated by the linux box failed
> 
> 
> 
> We have the same symptoms like we had before we applied the patch but reversed
> in direction.
> 
> 
> Following attachments are for the marked (***) case.
> 
> We recorded some information on the VAX (the last few lines are interesting
> starting with "Connect request received ..."
> 
> (See attached file: net.log)
> 
> 
> And I started the dncopy on the linux box with -v -v -v -v
> (See attached file: dncopy_send.output)
> 
> It looks like the optfil is  necessary when you are active sending a file and
> you have to omit it when you passive receiving a file.
 

patrick