So Confusing,Who Can explain the terms of monitoring for me?

Resident item ratio - The amount of active key-value pairs that are in memory.

Replica resident item ratio - The amount of replica key-value pairs that are in memory.

Disk write queue size - The amount of items that are waiting to be persisted to disk.

Disk fetches per sec - The amount of items that are being read from disk and brought into memory.

Items persisted per sec - The amount of items that are persisted to disk per second

RAM ejections per sec - The amount of items that are being kick out of memory to make room for items that have been fetched from disk

OOM errors per sec - The amount of keys-value pairs per second that could not be put into memory because there wasn't enough room. If this happens it means that you need to add more memory to your cluster because the key meta data has filled ram and values cannot be added to ram. This could happen if the amount of key-value pairs in your system is very large and your cluster memory is small in comparision.

Temp OOM errors per sec - This is similar to OOM errors, but not necessarily bad. It means that we are ejecting key-value pairs from memory and we can't handle your request right now. Try again later and it should work. In this case your client should exponentially back-off and the operation should eventually succeed. These should be rare.

Connections count - The amount of incoming connection to a Membase server

CAS hits per sec - The number of cas operations that have succeeded and the key-value pair was in memory.

CAS badval per sec - The number of cas operations that have failed because of a bad cas value

CAS misses per sec - The number of cas operations that have succeeded but the key-value pair was not in memory and had to be fetched from disk

VBucket errors per sec - The number of times a key-value pair was not on the server it was sent to and had to be redirected to the correct server in the cluster.

Thank you for your reply!