[collectd] Different OS for server and client

[Wallace, Ann]

On 2/4/10 8:53 AM, "Florian Forster" <octo at verplant.org> wrote:

> Hi Ann,
> 
> On Thu, Feb 04, 2010 at 08:11:06AM -0800, Wallace, Ann wrote:
>>> Do you, by any chance, still have the "config.log" file available?
>> 
>> I do not but I can rebuild to get the config.log if that will help. Is
>> there something in the log file I should be looking for?
> 
> that'd be great. This log contains how the configure script thinks that
> doubles can be converted to the x86 representation, which is the
> representation used in the network protocol. You can look for the string
> 
>   checking if doubles are stored in x86 representation

This string in is not in either of the config.log files. Either is the SPARC
endianflip string. The only thing I could find along the same lines is
checking for "checking for endian.h" Linux x86_64 Yes SPARC No.
> 
> which *should* result in "no" on SPARC and in "yes" on x86 and x86_64.
> 
> For SPARC I think the following test,
> 
>   checking if endianflip converts to x86 representation
> 
> should result in "yes".
> 
>>> Could you capture a couple of packets for me?
> 
>> 07:33:26.311630 IP xxx.nike.com.34821 > 10.10.111.10.25826: UDP, length 918
> 
> I was hoping to see in what way the doubles are present in the network
> packets. For this, I'd need a binary "PCAP" file containing the captured
> packets in their entirety. The following command should capture 20
> collectd packets sent from host "xxx" and store them in the file
> "/tmp/collectd.pcap" (interface eth0 is assumed):
> 
>   tcpdump -c 20 -i eth0 -s 0 -w /tmp/collectd.pcap host xxx and port 25826
> 

I will send this to you directly.

Thanks for all your help.

Ann