Note that there are at least 2 separate sub-tasks here, with potentially different priorities:

• deploying the objstorage API server on uffizi (the master object storage)
• deploying the objstorage API server on banco (the slave object storage)

I'm not sure if the archiver @qcampos is developing uses objstorage for both read and write. If so, both tasks are at the same priority; if not depoying on banco (for the write end of the archival) is more urgent.

In T405#5935, @zack wrote:

I'm not sure if the archiver @qcampos is developing uses objstorage for both read and write. If so, both tasks are at the same priority; if not depoying on banco (for the write end of the archival) is more urgent.

The archiver uses objstorage only for writing, and storage for reading.
However, it could be replaced by only an objstorage + database for reading as well, and may possibly be an improvement as the storage is an unnecessary intermediate layer.

Ready to be deployed on banco.