Use the startstats command to modify the interval at which per-node statistics for
volumes, managed disks (MDisks), and nodes are collected.
Syntax
startstats -interval time_in_minutes
Parameters
- -intervaltime_in_minutes
- (Required) Specifies the time in minutes. This is the time interval between
the gathering of statistics, from 1 to 60 minutes in increments of
1 minute.
Description
Running the startstats command
will reset the statistics timer to zero (0), and give it a new interval
at which to sample. Statistics are collected at the end of each sampling
period as specified by the -interval parameter.
These statistics are written to a file, with a new file created at
the end of each sampling period. Separate files are created for MDisks,
volumes and node statistics.
The files generated are written
to the /dumps/iostats directory.
A maximum
of 16 files are stored in the directory at any one time for each statistics
file type, for example:
Nm_stats_nodepanelname_date_time
Nv_stats_nodepanelname_date_time
Nn_stats_nodepanelname_date_time
Statistics files are created for all time intervals. Before
the 17th file for each type is created, the oldest file of that type
is deleted.
These files can be listed by using
the lsdumps command.
The
following naming convention is used for these files:
stats_type_stats_nodepanelname_date_time
Where stats_type is Nm for
MDisks, Nv for volumes, and Nn for
node statistics. nodepanelname is the current configuration
node panel name, date is in the format of yymmdd,
and time is in the format of hhmmss.
The
following is an example of an MDisk statistics file name: Nm_stats_000229_031123_072426
The
following is an example of a volume statistics file name: Nv_stats_000229_031123_072426
The
following is an example of a node statistics file name: Nn_stats_000229_031123_072426
Statistics
are collected for each MDisk and recorded in theNm_stats_nodepanelname_date_time file,
including the following statistical information:
- The number of SCSI read and write commands that are processed
during the sample period
- The number of blocks of data that are read and written during
the sample period
- Per MDisk, cumulative read and write external response
times in milliseconds
- Per MDisk, cumulative read and write queued response
times
Statistics are collected for each volume and recorded
in the Nv_stats_
nodepanelname_date_timefile,
including the following statistical information:
- The total number of processed SCSI read and write commands
- The total amount of read and written data
- Cumulative read and write response time in milliseconds
- Statistical information about the read/write cache usage
- Global Mirror statistics including latency
Statistics are collected for the node from which the
statistics file originated and recorded in the Nn_stats_nodepanelname_date_time file,
including the following statistical information:
- Usage figure for the node from which the statistic file
was obtained
- The amount of data transferred to and received from each port
on the node to other devices on the SAN
- Statistical information about communication to other nodes on
the fabric
An invocation example
startstats -interval 25
The
resulting output:
No feedback