Based on pg_test_fsync results, should we choose open_datasync or fdatasync as wal_sync_method? Can we rely on pg_test_fsync for choosing the best wal_sync_method or is there any other way?
Probably the default of fdatasync. The place where pg_test_fsync would tell me not to use fdatasync is if it were so fast that it was not credible that it was honestly syncing the data. I don't think pg_test_fsync does a good job of exercising the realistic differences between fdatasync and open_datasync. So unless it shows that one of them is lying about the durability, it doesn't offer much help.