Hi,
What happens I think is that lcg-cr command are running queries on the
LCG_GFAL_INFOSYS defined BDII, even if you provide a config file (a
modified /opt/edg/var/etc/edg-replica-manager/edg-replica-manager.conf
file that will use a so called "stub file").
So : if your SE doesn't publish info in the BDII lcg-* commands are
looking at, they will not find the desired SE in the MDS and give you
this "nice and easy to help debug" error message :)
This makes testing a little bit harder : you might have to use a "test
BDII"...
Anyone, please correct me if I'm wrong.
Regards,
Frederic Schaer
Mattias Wadenstein a écrit :
> Greetings,
>
> now we should have the HPC2N site here up and running and ready for
> proper testing (with one exception, see later). Since the underlaying OS
> on the worker nodes is Debian (with the rh73 tar wn package
> installed), it
> would be helpful to get some applications tested too for possible library
> dependancies, before running it in production.
>
> The one outstanding issue that I've found is with the lcg datamanagement
> tools, "lcg_cr: Invalid argument". Now, this is documented in the
> problems
> FAQ, but none of the diagnosises seems to apply.
>
> LCG_GFAL_INFOSYS is, as far as I can see, defined correctly. And the
> hostname does resolve, and respond. And both manual tests and the edg-rm
> testscript indicate that it is a working SE.
>
> dteam001@i023:~$ lcg-cr -v --vo dteam -d ibelieve-i.hpc2n.umu.se
> file:///etc/motd
> Using grid catalog type: edg
> lcg_cr: Invalid argument
> dteam001@i023:~$ ping ibelieve-i.hpc2n.umu.se
> PING ibelieve-i.hpc2n.umu.se (130.239.81.6): 56 data bytes
>
> /Mattias Wadenstein
>
|