argus-clients-3.0.2.beta.10 segfault when reading argus 2 data

[Robert Kerr]

On Fri, 2009-07-31 at 00:22 +0100, Carter Bullard wrote:

> To clarify, you are saying that ra* programs from argus-clients-3.0.2,
> reading from a remote argus-2.0 server, generate output files that are
> corrupt? But, argus-2.0 files that are generated by the argus server,
> for the same time period, are readable by argus-3.0 clients?

Yes, essentially. I'm not 100% whether the files are corrupt or the
files are perfectly valid but there's a bug in the clients that stops it
being read.

> Can you try an experiment?  What happens when an argus-3.0 client
> reads argus-2.0 and converts that to an argus-3.0 output file, and then
> reads that output file?  Problems?  If so, could you share the original
> argus-2.0 file, along with a brief description of how to recreate the  
> fault?

Yes, the same problem occurs when repeating the test this way round. So
there must be something in the original argus 2 datastream that argus 3
can read but not correctly write out.

I can't really share this file as it's from a production network, but
the problem is very reproducible - I'm not sure this system has produced
a single readable file. I will see if I can either reproduce in a lab or
with an anonymised version of the same data.

> What version of argus-clients-3.0 are you running?
> Have you tried argus-clients-3.0.2.beta.10 code?

This is on 3.0.2.beta.10 but I don't think it's a beta 10 bug, 90% sure
it happened with the stable version too.

-- 
 Robert Kerr