Hi
Just to be clear, I was not objecting to Ricardo reporting things the
way he did at all. I was just a) surprised that a gridftp change was
made between 2.2 and 2.3 and b) it was running in the first place! The
natural question was to ask which behavior was correct -- running or
not?
For those interested, the reason we missed it was the move to quattor.
gridftp was never chkconfig'd on in LCFGng-driven LCG, because it was
started by the (in)famous and inappropriately named globuscfg object.
Easy to miss if you forget and believe the naming. This is one of the
reasons accurate naming is important.
J "sort of like data management" T
On Mar 7, 2005, at 21:15, Ricardo Graciani wrote:
> Hi Stephen,
>
>>
>>> Another source of problems is connectivity with RB's,
>>> thus I use
>>> the above commands to upload a script to each CE supporting LHCb
> (and
>>> publishing a queue in "Production") and checking the connectivity
> via
>>> globus-url-copy commands.
>>
>> I think CEs always talk to the gridftp server on the RB, not vice
> versa,
>> so you should be checking outbound gridftp connections rather than
>> inbound.
>
> Yes but to check for several RB's in a timely fashion it is
> much
> easier to upload a script and execute it. For that I use the CE gridftp
> server.
>
> As I said, it can be done by other means (ie sending a command
> that downloads the script and a second command that executes it). But
> since gridftp servers were supposed to be running it was easier the
> other way around.
>
> If it is finally decided that CE's should not run gridftp
> servers it is fine with me, I will change my test. For the moment this
> is part of an LCG CE configuration and I think I must report if I find
> a
> site were it is not the case. The initial mail was directly to Nikhef,
> and Jeff was the one that open the discussion, which I find the
> appropriated way thing should go.
>
> Now like in other similar cases we need a way to get to an
> agreement and implement it in future releases.
>
>
>
> Ricardo
>
>>
>> Stephen
|