rc.35 backward compatibility issues

[Carter Bullard]

Hey Philipp,
Yes, you are correct, the man record processing may seem  a bit weird,
and there is a very long story to explain the current state of
affairs, that result in client programs not seeing "INIT" man records.

However, I suspect that we should do something that doesn't generate
confusion.  So, I'll put in initial man record counting, and we'll  
see how
that goes.

I thought we fixed the [s|d]appbytes 2.x processing?  Do you have a  
small
set of records ( >= 1) that shows the error?

And, I'm not aware of the srcid getting screwed up, so if you have a set
of records that show that, and what the number is suppose to be, that
would be most excellent.

Carter


On Nov 18, 2006, at 12:40 PM, Philipp E. Letschert wrote:

> When reading 2.0.6 logfiles with ra 3.x I noticed two oddities:
>
> - some of the 'man' records are not read at all, so the number of  
> records is not
>   in sync with the output of racount. It looks that the missed ones  
> are the
>   first entries, that are generated when argus starts and creates  
> or appends to
>   a logfile.
>   Probably as a result of this, most of the 'srcid' entries are  
> screwed up.
>   There are addresses like 5.0.0.66, 5.112.0.66, 5.168.0.66 and so on.
>
> - when reading 2.0.6 files with ra 3.x the entries for loss, sloss  
> and dloss are
>   always zero, even when loss was recorded
>
> I can live with that - just to let you know...
>
>