dmwm / PHEDEX

CMS data-placement suite
8 stars 18 forks source link

Add more "type" to requestlist API #983

Open amaltaro opened 9 years ago

amaltaro commented 9 years ago

Hi Tony,

is it possible to add a type=move to this requestlist API? I've been just mislead by its "xfer" output, e.g.: https://cmsweb-testbed.cern.ch/phedex/datasvc/xml/prod/requestlist?dataset=/QCD_Pt-15to3000_Tune4C_Flat_14TeV_pythia8/GEM2019Upg14DR-Phase1NoAgedJan23_PU50BX25_DES19_62_V8-v1/GEN-SIM-RECO

One does not know whether that was a move or normal replica transfer. Unless you go to the request number itself. Thanks.

TonyWildish commented 9 years ago

Hi Alan,

that shouldn't be too hard, will take care of it.

Cheers, Tony

amaltaro commented 9 years ago

Hi Tony, do you have any estimate on when this fix/feature can be developed?

TonyWildish commented 9 years ago

Hi Alan,

re-assigning to Chih-Hao, hopefully he will have time to tackle it.

T

huangch-fnal commented 9 years ago

"move" is not a request "type". Request type has only two possible values: "xfer" and "delete". "move" is an attribute of xfer requests. How do we want to do this? [1] add is_move=y (default any) as input parameter, or [2] display is_move value in the output so that the processing party can select based on this value?

TonyWildish commented 9 years ago

how about... [3], which is the sum of [1] and [2] :-) is that reasonable, and not too messy to do?

T

nikmagini commented 9 years ago

Hi Tony, Chih-Hao,

we already have another xfer-only parameter in the RequestList APIs: 'group' Since for 'group' you implemented [1], I vote [1] also for 'move' for consistency. (in addition - putting 'move' or 'group' in the output doesn't make sense for deletion requests, so IMO it's better to put them only in the argument)

Also note that the argument should be called 'move' instead of 'is_move' for consistency with other APIs (e.g. subscribe, subscriptions)

Finally, it's worth to implement the 'move' filter also for the 'transferrequests' API; for this API my vote would be [3].

TonyWildish commented 9 years ago

this all sounds good to me, thanks Nicolo'