Hackathon issues are discussed and transformed into working items
This issue summarizes feedback collected during the hackathon meeting, it is a list of points that need discussion, prioritization, and transformation into working items.
- clen-up oid management in taskflow pipeline (currently we have this strange fall-back to b2share, we should be clear if we want to keep that or make catalog mandatory and have fall-back to clean http, etc) #73
- replication (there is an idea of supporting replication use case, rather than deploying the data to the target location on workflow execution, a periodic rule make sure that the required data are present at required locations can be added) #66
- tests, the new repository for dags do not include tests (needs to be understood how the test are added and yet excluded from deployement) #65 (closed)
- webDAV dir sync (deploy data from a webDAV directory to target ssh location) #63
- UFTP (to enable transfers to/from JSC UFTP support might be beneficial especially with SSO) #76 (closed)
- SSO (at least a read-only view of DLS and W on Catalog?) #75 (closed)
- add logic to stage-in to avoid overwriting (we have something similar in image transfer)
- expose errors through alien4cloud (during the hands-on it was clear that suggestion to the users, e.g., wrong vault_id, missing oid, etc, could be very beneficial, we have it partly in DLS but this is not visible to alien4cloud)
- http timeout handling (hands-on case where the DLS read only 300 bytes from image service and ended upon time out without errors) #74 (closed)
- metadaserch in DC (not sure if this is required, as you can still just get all the records on the client and filter them?) #71 (closed)
Edited by Jedrzej Rybicki