Всем привет, кто-то может подсказать по pg_test_fsync 5 seconds per test O_DIRECT

supported on this platform for open_datasync and open_sync.

Compare file sync methods using one 8kB write:
(in wal_sync_method preference order, except fdatasync is Linux's default)
open_datasync 2409.192 ops/sec 415 usecs/op
fdatasync 2604.794 ops/sec 384 usecs/op
fsync 126.190 ops/sec 7925 usecs/op
fsync_writethrough n/a
open_sync 106.442 ops/sec 9395 usecs/op

Compare file sync methods using two 8kB writes:
(in wal_sync_method preference order, except fdatasync is Linux's default)
open_datasync 1758.350 ops/sec 569 usecs/op
fdatasync 2465.963 ops/sec 406 usecs/op
fsync 100.991 ops/sec 9902 usecs/op
fsync_writethrough n/a
open_sync 47.574 ops/sec 21020 usecs/op

Compare open_sync with different write sizes:
(This is designed to compare the cost of writing 16kB in different write
open_sync sizes.)
1 * 16kB open_sync write 83.602 ops/sec 11961 usecs/op
2 * 8kB open_sync writes 38.940 ops/sec 25681 usecs/op
4 * 4kB open_sync writes 26.933 ops/sec 37129 usecs/op
8 * 2kB open_sync writes 14.568 ops/sec 68643 usecs/op
16 * 1kB open_sync writes 6.358 ops/sec 157272 usecs/op

Test if fsync on non-write file descriptor is honored:
(If the times are similar, fsync() can sync data written on a different
descriptor.)
write, fsync, close 91.906 ops/sec 10881 usecs/op
write, close, fsync 99.959 ops/sec 10004 usecs/op

Non-sync'ed 8kB writes:
write 320374.843 ops/sec 3 usecs/op

все ли ок или где почитать на что надо смотреть?

А на то, что должно давать "железо", похоже?