Hi,
I'm sorry I have not written such a tool. I just asked JP how could I do
that, so that I could write the tool when I needed it, if ever. That
moment hasn't arrived yet (actually our DPM is not really in production
yet). From JP comments, the procedure would be more or less:
- Use dpns_listreplicax to list the files on a given filesystem.
- For each file:
- Use dpm-replicate and specify a new file type (by playing
with --s_type and RD_ONLY flags, you could force the fs/pool where the
replicas go to).
- Remove the original instance with dpm_delreplica.
As you see, you can't do it just with CLI, it involves writing some C
(or maybe perl, if the perl API is already there).
So, passing the ball back to you, if you happen to write the tool
sooner, you could share it with us :)
Cheers,
Antonio.
Jeff Templon escribió:
> Hi Antonio, *,
>
> Thanks for the answer -- Jean-Philippe and Sophie helped me to
> understand the problem. My confusion was thinking that the GID of the
> pool needed to be the unix GID of the associated pool account. Wrong.
> After changing it to the DPNS GID of the VO, it worked.
>
> A related point: I understand from Jean-Philippe that you may have
> written a tool to migrate a VO's files from one pool to another. If
> you have this (and even better if it is not written in Perl) can you
> send it to me?
>
> Thanks,
>
> JT
>
> Antonio Delgado Peris wrote:
>> Hi,
>>
>> Is it possible that your VO-dedicated pools are of, e.g., permanent
>> type while the client trying to write in it is asking for volatile
>> type (as rfcp or globus-url-copy do) and so the open pool is the only
>> one matching?
>>
>> Antonio.
>>
>>
>>
>> Jeff Templon escribió:
>>> Hi *,
>>>
>>> I am trying to figure out how DPM does its pool selection. In the
>>> DPM manual, there is a statement that if a VO-specific pool exists,
>>> as well as a general pool, the VO-specific pool will be used first.
>>> This is not what is happening here. I see the following in the dpm
>>> log file in /var/log/dpm/log:
>>>
>>> dpm_selectfs: selected pool: Volatile
>>>
>>> where Volatile is our "GID 0" pool. The Atlas-D pool is not getting
>>> selected. How do I figure out why a certain pool is / is not
>>> getting selected? Can I increase a logging level somewhere?
>>>
>>> It would also be nice if someone would write an actual DPM manual.
>>> I tried looking at the documentation, but the Wiki structure is not
>>> really amenable to understanding the DPM.
>>>
>>> JT
>>>
>>> ps: one more thing: the exact passage in the manual says:
>>>
>>>> If you define :
>>>> one pool dedicated to group1 / VO1
>>>> one pool open to all groups / VOs
>>>> then, the dedicated pool will be used until it is full.
>>>
>>> We have one pool dedicated to group1
>>> another pool to group2
>>> another pool to group3
>>> one pool open to all groups
>>>
>>> which is not *exactly* the same as what is described in the manual.
>>> Is this just accidentally being too precise with words, or does it
>>> really only work with a single vo-specific pool?
|