logo
Manual Pages
Table of Contents

NAME

na_sysstat - report filer performance statistics

SYNOPSIS

sysstat [ interval ] sysstat [ -c count ] [ -s ] [ -u | -x | -m | -f | -i | -b ] [ interval ]

DESCRIPTION

sysstat reports aggregated filer performance statistics such as the current CPU utilization, the amount of network I/O, the amount of disk I/O, and the amount of tape I/O. When invoked with no arguments sysstat prints a new line of statistics every 15 seconds. Use control-C or set the interval count (-c count ) to stop sysstat.

OPTIONS

-c count
Terminate the output after count number of iterations. The count is a positive, nonzero integer, values larger than LONG_MAX will be truncated to LONG_MAX.
-s
Display a summary of the output columns upon termination, descriptive columns such as `CP ty' will not have summaries printed. Note that, with the exception of `Cache hit', the `Avg' summary for percentage values is an average of percentages, not a true mean of the underlying data. The `Avg' is only intended as a gross indicator of performance. For more detailed information use tools such as na_nfsstat, na_netstat, or statit.
-f
For the default format display FCP statistics.
-i
For the default format display iSCSI statistics.
-b
Display the SAN extended statistics instead of the default display.
-u
Display the extended utilization statistics instead of the default display.
-x
Displays the extended output format instead of the default display. This includes all available output fields. Be aware that this produces output that is longer than 80 columns and is generally intended for "offline" types of analysis and not for "realtime" viewing.
-m
Displays multi-processor CPU utilization statistics. In addition to the percentage of the time that one or more CPUs were busy (ANY), the average (AVG) is displayed, as well as, the individual utilization of each processor.
interval
A positive, non-zero integer that represents the reporting interval in seconds. If not provided, the default is 15 seconds.

DISPLAYS

The default output format is as follows:
   CPU    NFS   CIFS   HTTP      Net kB/s    Disk kB/s     Tape kB/s    Cache
                                 in   out    read write    read write     age
  ###%  #####  #####  #####   ##### #####   ##### #####   ##### #####     >##
The FCP default output format is as follows:
   CPU    NFS   CIFS    FCP      Net kB/s    Disk kB/s      FCP  kB/s   Cache
                                 in   out    read write      in   out     age
  ###%  #####  #####  #####   ##### #####   ##### #####   ##### #####     >##
The iSCSI default output format is as follows:
   CPU    NFS   CIFS  iSCSI      Net kB/s    Disk kB/s    iSCSI  kB/s   Cache
                                 in   out    read write      in   out     age
  ###%  #####  #####  #####   ##### #####   ##### #####   ##### #####     >##
The SAN extended statistics output format is as follows:
   CPU   FCP iSCSI Partner  Total    FCP   kB/s   iSCSI   kB/s Partner  kB/s   Disk  kB/s   CP   CP Disk
                                      in    out      in    out      in   out   read write  time  ty util
  ###% ##### #####   ##### ######  #####  #####   #####  #####   ##### #####  ##### #####  ###%   A ###%
The utilization output format is as follows:
   CPU   Total    Net kB/s   Disk kB/s   Tape kB/s Cache Cache  CP  CP Disk
         ops/s    in   out  read write  read write   age   hit time ty util
  ###% ####### ##### ##### ##### ##### ##### #####   >##  ###% ###%  A ###%
The extended display output format is as follows:
   CPU   NFS  CIFS  HTTP   Total    Net kB/s  Disk kB/s    Tape kB/s Cache Cache  CP  CP Disk FCP   iSCSI   FCP  kB/s
                                    in   out  read write  read write   age   hit time ty util                in   out
  ###% ##### ##### ##### ####### ##### ##### ##### ##### ##### #####   >##  ###% ###%  A ###% ##### ##### ##### #####
The summary output format is as follows (for -u)
  --
  Summary Statistics (#### samples ## secs/sample)
   CPU   Total    Net kB/s   Disk kB/s   Tape kB/s Cache Cache  CP  CP Disk
         ops/s    in   out  read write  read write   age   hit util ty util

  Min
  ###% ####### ##### ##### ##### ##### ##### ##### #####  ###% ###%  * ###%

  Avg
  ###% ####### ##### ##### ##### ##### ##### ##### #####  ###% ###%  * ###%

  Max
  ###% ####### ##### ##### ##### ##### ##### ##### #####  ###% ###%  * ###%
The output column descriptions are:
CPU
The percentage of the time that one or more CPUs were busy doing useful work, during the previous interval seconds;
NFS
The number of NFS operations per second during that time;
CIFS
The number of CIFS operations per second during that time;
HTTP
The number of HTTP operations per second during that time;
FCP
The number of FCP operations per second during that time;
iSCSI
The number of iSCSI operations per second during that time;
Partner
The number of SCSI Partner operations per second during that time;
Net kB/s The number of kilobytes per second of network traffic into and out of the server; Disk kB/s The kilobytes per second of disk traffic being read and written; Tape kB/s The number of kilobytes per second of tape traffic being read and written; FCP kB/s The number of kilobytes per second of fcp traffic into and out of the server; iSCSI kB/s
The number of kilobytes per second of iSCSI traffic into and out of the server; Partner kB/s
The number of kilobytes per second of SCSI Partner traffic into and out of the server; Cache age The age of the data most recently evicted from the buffer pool. This data is usually, but not necessarily, the least recently used. Thus, it is possible for the statistic reported by sysstat to sometimes change erratically as buffers containing data of varying age are reclaimed. Total ops/s
The total number of operations per second (NFS + CIFS + HTTP) Cache hit The WAFL cache hit rate percentage. This value is the percent of instances in which WAFL attempted to load a disk-block that the data was found already cached in memory. A dash in this column indicates that WAFL did not attempt to load any blocks during the measurement interval. CP util The Consistency Point (CP) utilization, the % of time spent in a CP
CP ty
Consistency Point (CP) type, the cause of the CP that was started in the interval. Multiple CPs list no cause, just the number of CPs during the measurement interval. The CP types are as follows:
-
No CP started during sampling interval
number
Number of CPs started during sampling interval, if greater than one
B
Back to back CPs (CP generated CP)
b
Deferred back to back CPs (CP generated CP)
F
CP caused by full NVLog
H
A type H CP is a CP from high watermark in modified buffers. If a CP is not in progress, and the number of buffers holding data that has been modified but not yet written to disk exceeds a threshold, then a CP from high watermark is triggered.
L
A type L CP is a CP from low watermark in available buffers. If a CP is not in progress, and the number of buffers available goes below a threshold, then a CP form low watermark is triggered.
S
CP caused by snapshot operation
T
CP caused by timer
U
CP caused by flush
Z
CP caused by internal sync
V
CP caused by low virtual buffers
M
CP caused by low mbufs
D
CP caused by low datavecs
:
continuation of CP from previous interval
#
continuation of CP from previous interval, and the NVLog for the next CP is now full, so that the next CP will be of type B.
The type character is followed by a second character which indicates the phase of the CP at the end of the sampling interval. If the CP completed during the sampling interval, this second character will be blank. The phases are as follows:
0
Initializing
n
Processing normal files
s
Processing special files
q
Processing quota files
f
Flushing modified data to disk
v
Flushing modified superblock to disk
Disk util The disk utilization (percentage) of the busiest disk since a true aggregate value would probably not show the user that there is some type disk based bottleneck. Do not confuse this with disk space used, this is an access based value.

EXAMPLES

sysstat
Display the default output every 15 seconds, requires control-C to terminate. sysstat 1
Display the default output every second, requires control-C to terminate. sysstat -s 1
Display the default output every second, upon control-C termination print out the summary statistics. sysstat -c 10
Display the default output every 15 seconds, stopping after the 10th iteration. sysstat -c 10 -s -u 2 sysstat -u -c 10 -s 2
Display the utilization output format, every 2 seconds, stopping after the 10th iteration, upon completion print out the summary statistics. sysstat -x -s 5
Display the extended (full) output, every 5 seconds, upon control-C termination print out the summary statistics.

CLUSTER CONSIDERATIONS

In takeover mode, the sysstat command displays the combined statistics from both the failed filer and the live filer. The statistics diplayed by the sysstat command are cumulative; a giveback operation does not zero out the statistics. That is, after giving back its partner's resources, the live filer does not subtract the statistics about operations it performed on behalf of the failed filer in takeover mode.

SEE ALSO

na_netstat(1), na_nfsstat(1), na_partner(1).
Table of Contents