Argus seems to stop writing data that ra can read.

Carter Bullard carter at qosient.com
Wed Jun 18 09:34:26 EDT 2014


Great !!!
Carter

> On Jun 18, 2014, at 9:26 AM, elof2 at sentor.se wrote:
> 
> 
> Late response:
> 
> Very agreeable!
> 
> /Elof
> 
> 
>> On Tue, 8 Apr 2014, Carter Bullard wrote:
>> 
>> Hey Scott,
>> OK, sorry for the delay on this….  So we have in the clients a sanity check
>> for bogus metric values as we’re reading argus records from the input,
>> and these records that argus-3.0.7.5 is generating that have errant
>> packet and byte counts are causing us to reject the stream, and stop
>> processing.
>> 
>> I have a fix in the clients to realize the error, zero the errant
>> data and mark the flow record cause as ARGUS_ERROR, which puts an
>> “ERR” for the flow status.  I don’t like the reason we’re doing this,
>> argus generating bogus data, but it allows us to continue to process
>> the stream without propagating the corrupted record.
>> 
>> So if this is agreeable,  I’ll have this up on the server later today.
>> 
>> Sorry for the inconvenience.
>> 
>> Carter
>> 
>>> On Mar 27, 2014, at 1:57 PM, Unknown <s.a.mcintyre at gmail.com> wrote:
>>> 
>>> Hi Carter,
>>> 
>>> 
>>>> On 27 Mar 2014, at 22:47, Carter Bullard wrote:
>>>> 
>>>> Hey Scott,
>>>> This reads like a corrupt output file.   Any chance you could be running 2 argi, or same argus writing to the same output file twice ???
>>> 
>>> Nope, double verified that, definitely only one process writing, and as you probably noticed from my follow-up, using an older client version reads the server output file just fine.  I've gone ahead and stuck with the older client version for now.
>>> 
>>> 
>>>> 
>>>> Can you share one of the files that exhibits the problem??  Does the client just spin or does eventually terminate / die ???
>>> 
>>> Running a simple ra -n -r argus.out terminates fine, doesn't spin, just doesn't show the full contents.  Using the old ra() version shows the full contents.
>>> 
>>> I can get you a sample file, sure, let me know if you still want one or if the clue about different versions of the client was enough.
>>> 
>>> Cheers,
>>> Scott
>> 



More information about the argus mailing list