Heavy disk I/O
We have a mini cluster from 2 couchbase 2.0 nodes (replica).
We see quite a lot of disk I/O (10 - 20 Megs / sec writing) from couchbase, our disk write queue seems to be really small compared to other users (up to 100-200).
I am not sure how can I investigate this, I/O is similar on both nodes, it causes about 15% of IOWAIT on cpu.
We use virtual nodes ( they have: 11170547 items in buckej - we use only one bucket, and doing about 400 ops /second)
Can you guys advice what / where to check?