Hello,
I'm having fun and games with an upgrade to 1.8.0-12p4 too. Seems
naked srmcp fails in exactly the same way as Greig sees. Lcg-cp and
srmcp with explicitly set port ranges work. I didn't change the
dcachesetup file with the upgrade, and only upgraded the
srm/poolmanager node- not the gridftp door nodes- although all nodes
got a restart after the upgrade.
cheers,
Matt
On 31/01/2008, Greig Alan Cowan <[log in to unmask]> wrote:
> Hi all,
>
> I've upgraded the Edinburgh dCache to 1.8.0-12p4. Everything seemed to
> go fine, apart from the fact that I now can't use srmcp to copy files
> out. g-u-c and lcg-cp are fine. In the srmcp case it is becasue the pool
> is trying to connect to a port on the client machine that is outside of
> the defined GLOBUS_TCP_PORT_RANGE. In the guc and lcg-cp cases, it is
> the client that is connecting to the pool or door, which is fine. Anyone
> else seen this?
>
> If I run the client with the port range defined explicitely, it works:
>
> srmcp -1 -debug -globus_tcp_port_range=50000,52000
> srm://srm.epcc.ed.ac.uk:8443/pnfs/epcc.ed.ac.uk/data/lhcb/1201621959
> file:////tmp/test4
>
> Has there been a change in how the port range should be defined
> (50000,52000 or 50000:52000 or 50000 52000)?
>
> Cheers,
> Greig
>
|