telegraf/plugins/inputs/nfsclient

Telegraf plugin: NFSCLIENT

Plugin arguments:

• fullstat bool: Collect per-operation type metrics

Description

The NFSCLIENT plugin collects data from /proc/self/mountstats, by default it will only include a quite limited set of IO metrics. If fullstat is set, it will collect a lot of per-operation statistics.

Measurements & Fields

• nfsstat_read

  • read_bytes (integer, bytes) - The number of bytes exchanged doing READ operations.
  • read_ops (integer, count) - The number of RPC READ operations executed.
  • read_retrans (integer, count) - The number of times an RPC READ operation had to be retried.
  • read_exe (integer, miliseconds) - The number of miliseconds it took to process the RPC READ operations.
  • read_rtt (integer, miliseconds) - The round-trip time for RPC READ operations.

• nfsstat_write

  • write_bytes (integer, bytes) - The number of bytes exchanged doing WRITE operations.
  • write_ops (integer, count) - The number of RPC WRITE operations executed.
  • write_retrans (integer, count) - The number of times an RPC WRITE operation had to be retried.
  • write_exe (integer, miliseconds) - The number of miliseconds it took to process the RPC WRITE operations.
  • write_rtt (integer, miliseconds) - The rount-trip time for RPC WRITE operations.

In addition enabling fullstat will make many more metrics available, but description of those is beyond the scope here. See references for more details.

Tags

• All measurements have the following tags:
  • mountpoint The local mountpoint, for instance: "/var/www"
  • serverexport The full server export, for instance: "nfsserver.example.org:/export"

References

nfsiostat net/sunrpc/stats.c - Linux source code What is in /proc/self/mountstats for NFS mounts: an introduction