Print

Print


Hi Andreas,

yes, the thing was, that a (dCache) pool accidentialy did not get an 
AAAA assigned, when it got deployed. So, in some sense, the behaviour 
was a good smoke test to notice the issue ;)

On the way, I had been thinking, if/what things to consider might be 
there, when there are more complex constellations coming, with 
IPv4-only, IPv6-only and dual-stack nodes trying to interact with each 
other through redirects or such more 'indirect' paths...

Cheers,
   Thomas

On 26/11/2020 10.43, Andreas Petzold (SCC) wrote:
>      Hi Thomas,
> 
> your question sounds a bit like a mix between dCache and SLAC xrootd ;-)?
> 
> Of course the experts should comment, but let me add my 2 cents.
> 
> If you are talking about SLAC xrootd, I don't see a reason why a 
> transition between v6 and v4 would not work.
> 
> The dual-stack client connects to the xrootd manager/redirector via v6 
> and gets told where to connect to access the file. If that _new_ 
> connection is made via v4 or v6 should not matter. It is always a new 
> connection for the client and that's the beauty of the protocol (IMHO).
> 
> How dCache handles a situation like this when you have an xrootd client 
> connecting to a dCache xrootd door via v6 and you have dCache pools on 
> v4, I don't know.
> 
>      Cheers,
> 
>          Andreas
> 
> 
> On 25/11/2020 14.10, Thomas Hartmann wrote:
>> Hi all,
>>
>> would it be in principle possible, that the xrd client reastablishes a 
>> conenction on IPv4 after the initital connection got established over 
>> IPv6?
>>
>> E.g.,
>> a dual-stack client connects to a dual-stack redirector/door and the 
>> connection is established on v6. The requested file is only available 
>> on a IPv4 node and the redirector points to the pools DNS, that only 
>> resolves for A but not AAAA.
>> The redirect fails as the non-resolved v6 ends up as an invalid 
>> redirect URL.
>>
>> I guess, that client/server do not have much chance, after the IP 
>> flavour got negotiated on the network stack, or?
>>
>> Cheers,
>>    Thomas
>>
>>
>> ########################################################################
>> Use REPLY-ALL to reply to list
>>
>> To unsubscribe from the XROOTD-L list, click the following link:
>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
> 


########################################################################
Use REPLY-ALL to reply to list

To unsubscribe from the XROOTD-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1