> Here is what I think should happen.
>
> (1) Subscribe to the first copy. This lets people work.
> Subscribe to additional copies if there are requests, and
> these subscriptions may be for subsets as needed. I can
> imagine that some analyses do not care about a new version,
> e.g. it fixes muon alignment but I am looking strictly at
> electrons. Or I am looking at SM t-tbar and don't care about
> all the SUSY stuff. We should end up with less than 3 full copies.
I talked to Peter about this. Peter will figure out exactly what will happen. In the mean time, we can subscribe to everything. In that case, the first FDR-1 AOD will arrive for analysis. Others can continue to come and we don't have to be too concern whether they fill up our storage or not. And we can always delete CCRCs.
>
> (2) It's essential to take part in the sustained transfer to
> flush out problems. What is the minimum disk space we can
> devote to this? As soon as a dataset is done, it can be
> deleted, right?
Min 24TB, likely 30TB without impact to production. If there is no objection, I will tell Alexei that we want all AODs.
--
Wei Yang | [log in to unmask] | 650-926-3338(O)
|