Die neue Version von PerfVolume kann nun mit beliebig vielen Volumes auf einem Filer umgehen. Die vormals berüchtigten Fehlermeldungen “Not enough memory to get instances” gehört damit der Vergangenheit an. Sollte die Anzahl der Instanzen größer als das einstellbare Maximum sein, sendet das Plugin noch zur Laufzeit einen zweiten Aufruf, um die noch fehlenden Instanzen abzuholen.

Die Kernaufgabe des Checks PerfVolume ist gleich geblieben: Die laufende Überwachung der Latenz (per volume latency) auf NetApp-Filern.

Zusätzlich sind einige neue Performance-Counter überwachbar (SAN*, CIFS*, FCP*) und der Schalter --explain=counters zeigt alle am Filer verfügbaren Counter an.

Beispielausgabe:

$ ./get_netapp_perfdata.pl -H sim812 -m 7 -o volume --explain=counters

Performance counters for perf-object 'volume' available on this filer:
==============================================================================
Counter-Name                Description (Unit, Properties, Privilege)
==============================================================================
------------------------------------------------------------------------------
asynchronous_frees          Number of asynchronous frees per second to the
                            volume (per_sec, rate, diag)
------------------------------------------------------------------------------
avg_latency                 Average latency in microseconds for the WAFL
                            filesystem to process all the operations on the
                            volume; not including request processing or
                            network communication time (microsec, average,
                            basic)
[...]
------------------------------------------------------------------------------

cifs_read_latency           Average time for the WAFL filesystem to process
                            CIFS read requests to the volume; not including
                            CIFS protocol request processing or network
                            communication time which will also be included in
                            client observed CIFS request latency (microsec,
                            average, advanced)
------------------------------------------------------------------------------
cifs_read_ops               Number of cifs reads per second to the volume
                            (per_sec, rate, advanced)
------------------------------------------------------------------------------
cifs_write_data             Bytes written per second via cifs to the volume
                            (b_per_sec, rate, advanced)
------------------------------------------------------------------------------
cifs_write_latency          Average time for the WAFL filesystem to process
                            CIFS write requests to the volume; not including
                            CIFS protocol request processing or network
                            communication time which will also be included in
                            client observed CIFS request latency (microsec,
                            average, advanced)
------------------------------------------------------------------------------
cifs_write_ops              Number of cifs writes per second to the volume
                            (per_sec, rate, advanced)
------------------------------------------------------------------------------
clone_afs_full_file         Number of full file clones from AFS (none, raw,
                            diag)
------------------------------------------------------------------------------
clone_afs_sub_file          Number of sub file clones from AFS (none, raw,
                            diag)
------------------------------------------------------------------------------
clone_blks_copied           Number of time clone splits resulted in copying
                            blocks (none, raw, diag)
------------------------------------------------------------------------------
clone_eio_blks              Number of time EIO is returned because of absense
                            of backing snapshot (none, raw, diag)
------------------------------------------------------------------------------


[...]

------------------------------------------------------------------------------
nfs_read_latency            Average time for the WAFL filesystem to process
                            NFS protocol read requests to the volume; not
                            including NFS protocol request processing or
                            network communication time which will also be
                            included in client observed NFS request latency
                            (microsec, average, advanced)
------------------------------------------------------------------------------
nfs_read_ops                Number of NFS reads per second to the volume
                            (per_sec, rate, advanced)
------------------------------------------------------------------------------
nfs_write_data              Bytes written per second via NFS to the volume
                            (b_per_sec, rate, diag)
------------------------------------------------------------------------------
nfs_write_latency           Average time for the WAFL filesystem to process
                            NFS protocol write requests to the volume; not
                            including NFS protocol request processing or
                            network communication timewhich will also be
                            included in client observed NFS request latency
                            (microsec, average, advanced)
------------------------------------------------------------------------------
nfs_write_ops               Number of NFS writes per second to the volume
                            (per_sec, rate, advanced)
------------------------------------------------------------------------------
node_name                   System node name (none, string, basic)
------------------------------------------------------------------------------
nonzero_dbys_cnt            Number of times blks_vvol_dbys_df_cnt counter
                            being not zero, at the end persistent delayed
                            free merge scores. (none, raw, diag)
------------------------------------------------------------------------------
other_latency               Average latency in microseconds for the WAFL
                            filesystem to process other operations to the
                            volume; not including request processing or
                            network communication time (microsec, average,
                            basic)
------------------------------------------------------------------------------
other_ops                   Number of other operations per second to the
                            volume (per_sec, rate, basic)
------------------------------------------------------------------------------

[...]
==============================================================================
Counters listed, now configure your nagios ...
Not all of the above counters are supported by this plugin yet!
For enhancing the plugin with additional counters contact
distribution@rfi.net