Details
-
Enabler
-
Could have
-
None
-
Data Processing
-
-
-
5
-
5
-
2
-
0.4
-
Team_NALEDI
-
Sprint 4
-
-
-
-
16.6
-
Stories Completed, Outcomes Reviewed, NFRS met, Demonstrated, Satisfies Acceptance Criteria, Accepted by FO
-
-
SPO-1591
Description
We will test a delivery prototype based on Rucio
Who?
AA Operators: AIV Engineers, Commissioning Scientists
Also: SDP developers, SKA Software Integrators
What?
A test SDP to SRC delivery of a modest number of AA0.5-size data sets using Rucio. This would aim to prototype a Rucio-based solution to understand how data could be stored, accessed and transferred along with suitable metadata for AA0.5.
Why?
Whilst data delivery requirements are relatively modest for AA0.5, there is still a need to enable management, interrogation and potential transfer off-site of the data products and relevant metadata from commissioning tests. This could be of the scale of hundreds of files. Building on recent work to test Rucio as a potential data-delivery solution this presents an opportunity to de-risk a potential long-term solution and improve early user-access.
-------
Notes:
This should build on work from previous PIs see e.g. SP-1701 and SP-2220 which established Rucio instances and tested authentication and data replication.
Old description before the ticket was pivoted (for reference)
- The SDP deployment is modified such that it allocates a configurable amount of storage, which can be mounted into all workflows.
- It should be possible to configure it such that its lifetime is not limited to the SDP deployment (i.e. persistent)
- A directory structure convention is established and documented that keeps different kinds of data (see PB dependency type?) from different EBs/PBs separate
- Receive and batch workflows are modified such that they mount the storage and they read and write files as specified
- Stretch: It is demonstrated how we can synchronise the contents of this storage off-site (either as a job/service or part of the workflows)
- Stretch: (Automated?) service / job to clean up old files
- Stretch: Provide ska-sdp commands for listing and downloading (/synchronising/cleaning up?) objects in storage
- Stretch: Tango interface to monitor storage fill status / initiate clean-up
Attachments
Issue Links
- clones
-
SP-1211 Explore use of Rucio for SDP Delivery
- Discarded
-
SP-1338 Define and create (very minimal) PoC for system-wide data management
- Discarded
- relates to
-
SP-2446 Support replication of a prototype SDP->SRC data transfer with Rucio
- Done
-
ADR-55 Definition of metadata for data management at AA0.5
- decided