[seekdb][observer] Size automatic log disk from available space

[hnwyllmm]

Task Description

When log_disk_size is left at its automatic value, seekdb currently derives the log disk size from the memory budget. On high-memory hosts with a relatively small amount of remaining disk space, this can request a log disk larger than the redo filesystem can provide and make the first systemd startup fail with OB_LOG_OUTOF_DISK_SPACE.

Solution Description

For automatic log_disk_size, calculate the size from the redo filesystem's currently available capacity (f_bavail * f_bsize) and the existing configured/default percentage. Keep explicit byte-size configuration semantics unchanged. Reject an automatically calculated value below the existing server minimum instead of forcing a fixed 2 GiB default. Also return the selected automatic percentage correctly.

Passed Regressions

• make -C build_release ob_server passed.
• git diff --check passed.

Upgrade Compatibility

The change only affects automatic log_disk_size calculation when the configured byte size is zero. Explicit log_disk_size behavior remains unchanged.

Other Information

This MR does not add a unit test. It does not change datafile sizing and does not introduce joint capacity planning for data and log paths on the same filesystem.

Release Note

[hnwyllmm]

The mapping Dima issue is "[[SeekDB][systemd] Initial startup fails in high-memory, low-disk environments due to overly large automatic log_disk_size calculation".