Interesting situation. I agree that after 30s to a minute that most things should have been flushed. As a side note, it would be interesting to benchmark async vs wapbl. I have never really looked, but it has always seemed that it would be nice to have: statistics visibility into the number of dirty buffers/etc. in various caches a way to force flushes and clear caches (individually) Specifically, I think it would be great if 'systat vmstat' had a count of dirty buffers. Perhaps this is doable now and I just don't know how. Another question is if the disk had write caching enabled, but I would also expect it to flush the write cache quickly. It would be nice to have visibility into that cache, but I don't know if the ata interface supports it.
Description: PGP signature