[Go-essp-tech] QC and replication

[Jeffrey F. Painter]

I think you would like to have the DRS date-based version numbers for 
the NCC data.

For the data which PCMDI hosts as a node, these DRS version numbers can 
presently be extracted from the THREDDS catalog.
That is not necessarily true of other nodes' catalogs, whether or not 
you can get their data through the PCMDI gateway.
Unfortunately I don't know how to find such information from NCC.

- Jeff Painter

On 7/27/11 8:36 AM, Estanislao Gonzalez wrote:
> Hi all,
>
> We are currently getting NCC data for QCL2 here at DKRZ. The data was
> published in a NCC datanode linked to PCMDI's Gateway in a non-DRS
> conform structure.
>
> I´m getting the data "as is", but I'll need to version it in order to
> validate it for QC (QC needs a version for this).
>
> I'm not sure how to proceed now, and would like to get some feedback
> regarding:
> 1) I can't separate output1 from output2 without any knowledge of the
> model run, should I trust the separation already in place? (It was
> assigned somehow to those groups, but I have no information on how, the
> directory structure doesn't reflecct this)
> 2) I´ll have to version it and move it to a proper DRS structure. This
> will break the relation to the original files. Is it still possible to
> mark it as a replica? (the dataset id should be the same one.. I think).
> Is it desiderable to have replicas whose directory structure doesn't
> match the original file?
> 3) The original files are not vesioned, so any change will overwrite
> the previous file. That means that our "replicas" will have no further
> "original" to be attached to. Is this also a problem?
>
> As usual I'm against such workarounds, but this is what we have now and
> I have no possibility to force/motivate people to follow the standard
> procedure which was already agreed on.
>
> I can't seem to find a way to solve this inconsistencies without adding
> more confusion to the curent problem.
>
> Any ideas?
>
> Thanks,
> Estani
>