Use of FDB disk snapshot facility


I am trying to follow the instructions here Disk snapshot backup and Restore — FoundationDB 7.2 with a test setup and here is what I find.

  • whitelist_binpath doesn’t work if there are quotes around the binary path. you may want to correct the documentation. Or let me know where I can send a PR.

  • In a single-node setup, where coord is same node as storage for example, FDB seems to invoke the ‘snapshot’ command multiple times (once for each role). At that point, how would the command know how to avoid duplicate snapshots of the same path/drive? In a production setup, I can understand tlog/storage/coord will be separate processes (and indirectly separate data directory). Is this a strict requirement for snapshot system to work?