lustre2

README

Lustre Input Plugin

The Lustre® file system is an open-source, parallel file system that supports many requirements of leadership class HPC simulation environments.

This plugin monitors the Lustre file system using its entries in the proc filesystem.

Configuration

# Read metrics from local Lustre service on OST, MDS
[[inputs.lustre2]]
  ## An array of /proc globs to search for Lustre stats
  ## If not specified, the default will work on Lustre 2.5.x
  ##
  # ost_procfiles = [
  #   "/proc/fs/lustre/obdfilter/*/stats",
  #   "/proc/fs/lustre/osd-ldiskfs/*/stats",
  #   "/proc/fs/lustre/obdfilter/*/job_stats",
  #   "/proc/fs/lustre/obdfilter/*/exports/*/stats",
  # ]
  # mds_procfiles = [
  #   "/proc/fs/lustre/mdt/*/md_stats",
  #   "/proc/fs/lustre/mdt/*/job_stats",
  #   "/proc/fs/lustre/mdt/*/exports/*/stats",
  # ]

Metrics

From /proc/fs/lustre/obdfilter/*/stats and /proc/fs/lustre/osd-ldiskfs/*/stats:

• lustre2
  • tags:
    • name
  • fields:
    • write_bytes
    • write_calls
    • read_bytes
    • read_calls
    • cache_hit
    • cache_miss
    • cache_access

From /proc/fs/lustre/obdfilter/*/exports/*/stats:

• lustre2
  • tags:
    • name
    • client
  • fields:
    • write_bytes
    • write_calls
    • read_bytes
    • read_calls

From /proc/fs/lustre/obdfilter/*/job_stats:

• lustre2
  • tags:
    • name
    • jobid
  • fields:
    • jobstats_ost_getattr
    • jobstats_ost_setattr
    • jobstats_ost_sync
    • jobstats_punch
    • jobstats_destroy
    • jobstats_create
    • jobstats_ost_statfs
    • jobstats_get_info
    • jobstats_set_info
    • jobstats_quotactl
    • jobstats_read_bytes
    • jobstats_read_calls
    • jobstats_read_max_size
    • jobstats_read_min_size
    • jobstats_write_bytes
    • jobstats_write_calls
    • jobstats_write_max_size
    • jobstats_write_min_size

From /proc/fs/lustre/mdt/*/md_stats:

• lustre2
  • tags:
    • name
  • fields:
    • open
    • close
    • mknod
    • link
    • unlink
    • mkdir
    • rmdir
    • rename
    • getattr
    • setattr
    • getxattr
    • setxattr
    • statfs
    • sync
    • samedir_rename
    • crossdir_rename

From /proc/fs/lustre/mdt/*/exports/*/stats:

• lustre2
  • tags:
    • name
    • client
  • fields:
    • open
    • close
    • mknod
    • link
    • unlink
    • mkdir
    • rmdir
    • rename
    • getattr
    • setattr
    • getxattr
    • setxattr
    • statfs
    • sync
    • samedir_rename
    • crossdir_rename

From /proc/fs/lustre/mdt/*/job_stats:

• lustre2
  • tags:
    • name
    • jobid
  • fields:
    • jobstats_close
    • jobstats_crossdir_rename
    • jobstats_getattr
    • jobstats_getxattr
    • jobstats_link
    • jobstats_mkdir
    • jobstats_mknod
    • jobstats_open
    • jobstats_rename
    • jobstats_rmdir
    • jobstats_samedir_rename
    • jobstats_setattr
    • jobstats_setxattr
    • jobstats_statfs
    • jobstats_sync
    • jobstats_unlink

Troubleshooting

Check for the default or custom procfiles in the proc filesystem, and reference the Lustre Monitoring and Statistics Guide. This plugin does not report all information from these files, only a limited set of items corresponding to the above metric fields.

Example Output

lustre2,host=oss2,jobid=42990218,name=wrk-OST0041 jobstats_ost_setattr=0i,jobstats_ost_sync=0i,jobstats_punch=0i,jobstats_read_bytes=4096i,jobstats_read_calls=1i,jobstats_read_max_size=4096i,jobstats_read_min_size=4096i,jobstats_write_bytes=310206488i,jobstats_write_calls=7423i,jobstats_write_max_size=53048i,jobstats_write_min_size=8820i 1556525847000000000
lustre2,host=mds1,jobid=42992017,name=wrk-MDT0000 jobstats_close=31798i,jobstats_crossdir_rename=0i,jobstats_getattr=34146i,jobstats_getxattr=15i,jobstats_link=0i,jobstats_mkdir=658i,jobstats_mknod=0i,jobstats_open=31797i,jobstats_rename=0i,jobstats_rmdir=0i,jobstats_samedir_rename=0i,jobstats_setattr=1788i,jobstats_setxattr=0i,jobstats_statfs=0i,jobstats_sync=0i,jobstats_unlink=0i 1556525828000000000

Documentation

Overview

Package lustre2 (doesn't aim for Windows) Lustre 2.x Telegraf plugin Lustre (http://lustre.org/) is an open-source, parallel file system for HPC environments. It stores statistics about its activity in /proc

Index

• type Lustre2
  • func (l *Lustre2) Gather(acc telegraf.Accumulator) error
  • func (l *Lustre2) GetLustreProcStats(fileglob string, wantedFields []*mapping) error
  • func (*Lustre2) SampleConfig() string

Types

type Lustre2

type Lustre2 struct {
	OstProcfiles []string `toml:"ost_procfiles"`
	MdsProcfiles []string `toml:"mds_procfiles"`
	// contains filtered or unexported fields
}

Lustre proc files can change between versions, so we want to future-proof by letting people choose what to look at.

func (*Lustre2) Gather

func (l *Lustre2) Gather(acc telegraf.Accumulator) error

Gather reads stats from all lustre targets

func (*Lustre2) GetLustreProcStats

func (l *Lustre2) GetLustreProcStats(fileglob string, wantedFields []*mapping) error

func (*Lustre2) SampleConfig

func (*Lustre2) SampleConfig() string