Home / server / Questions / 13032
Accepted
deinspanjer
Asked: 2009-05-28 07:36:45 +0800 CST

What are some good ways to monitor IO throughput and load from the client of an NFS share?

  • 772

I have an ETL process that opens files from an NFS mount to process them. The number of files to be opened simultaneously is configurable.

I don't have shell access to the NFS server.

What are some good shell utilities I can use on the client to determine how effectively I'm using the NFS share? I want to maximize my throughput (balanced against flooding my ETL with too much data) and obviously, I don't want to put too great a strain on the NFS server.

How much can I do from the client vs. having to log on to the NFS server and run IO monitoring utilities there?

performance linux nfs monitoring

2 Answers

  1. Best Answer

    I would start with nfsstat and iostat.

    For example, you might want to watch:

    watch -n1 'nfsstat -c'
iostat -n 1 1
    • 2

  2. You might also find pipe viewer useful, depending on what your ETL process is doing with your data.

    As per the following tutorial, it's basic usage is of the form:

    pv access.log | gzip > access.log.gz

    Where it would let you know how much data is completed vs how much is left to go.

    You can also chain commands through it as so:

    pv -cN source access.log | gzip | pv -cN gzip > access.log.gz
    • 1