Dear All,
Paul Trepka pointed me in the direction of the ip_nat_ftp module (which
wasn't loaded) - I haven't had chance to test it yet but I'm pretty sure
that this is going to fix our NAT issue. I say this because Paul Trepka
and (Simon George at) RHUL both have srmcp working with NAT, both
replied off-line and both gave me the impression that it "just kind of
works out of the box".
What I said about GridFTP is still true (parallel transfer are always
active on the sending side - but with single streams either side can be
active). And this may still cause problems if you have to go through,
say, a University firewall. But with regards to our particular NAT
problem I'm pretty sure Paul is on the money.
regards
Jon
Winnie Lacesso wrote:
> On Wed, 26 Sep 2007, Greig A Cowan wrote:
>
>>> And rfrm if it matters was also broken: No route to host.
>>> (srm-advisory-delete worked but that does not unregister the file -
>>> lcg-del does)
>>>
>> So no RFIO access was possible at all? We'll need to test this out.
>>
>
> That was before JonW made a partial fix in experimenting with srmcp
> problems. Probably fixed now since JonW has made several fixes.
>
> Will test it soon, once gpfs storage is rebuilt (test storage hardware
> got taken away) & WN & UI tar install restored / rebuilt.
>
|