Friday, 17 May 2013

nmon command


Purpose

Displays local system statistics in interactive mode and records system statistics in recording mode.

Syntax

Interactive mode:
nmon [-h ]
nmon [ -s < seconds > ] [ -c < count > ] [ -b ] [ -B ] [ -g < filename > ] [ -k disklist ] [ -C < process1:process2:..:processN > ]
Recording mode:
nmon [ -f | -F filename | -x | -X | -z ] [ -r < runname > ] [ -t | -T | -Y ] [ -s seconds ] [ -c number ] [ -w number ] [ -l dpl ] [ -d ] [-g filename ] [ -k disklist ] [ -C <process1:process2:..:processN > ] [ -G ] [ -K ] [ -o outputpath ] [ -D ] [ -E ] [ -J ] [ -V ] [ -P ] [ -M ] [ -N ] [ -W ] [ -S ] [ -^ ] [ -O ] [ -L ] [ -I percent ] [ -A ] [ -m < dir > ] [ -Z priority ]
Note: In recording mode, specify only one of the -f-F-z-x, or -X flags as the first argument.

Description

The nmon command displays and records local system information. The command can run either in interactive or recording mode. If you specify any of the -F-f-X-x, and -Z flags, the nmon command is in recording mode. Otherwise, the nmon command is in interactive mode.
The nmon command provides the following views in interactive mode:
  • System resources view(using the r key)
  • Process view (using the t and u keys)
  • AIO processes view (using the A key)
  • Processor usage small view (using the c key)
  • Processor usage large view (using the C key)
  • Shared-processor logical partition view (using the p key)
  • NFS panel (using the N key)
  • Network interface view (using the n key)
  • WLM view (using the W key)
  • Disk busy map (using the o key)
  • Disk groups (using the g key)
  • ESS vpath statistics view (using the e key)
  • JFS view (using the j key)
  • Kernel statistics (using the k key)
  • Long term processor averages view (using the l key)
  • Large page analysis (using the L key)
  • Paging space (using the P key)
  • Volume group statistics (using the V key)
  • Disk statistics (using the D key)
  • Disk statistics with graph (using the d key)
  • Memory and paging statistics (using the m key)
  • Adapter I/O statistics (using the a key)
  • Shared Ethernet adapter statistics (using the O key)
  • Verbose checks OK/Warn/Danger view (using the v key)
  • Detailed Page Statistics (using the M key)
  • Fibre Channel adapter statistics (using the ^ key)

In the recording mode, the command generates the nmon files. You can view these files directly by opening them or with post processing tools such as nmon analyzer. The nmon tool disconnects from the shell during the recording, ensuring that the command continues running even if you log out.
If you use the same set of keys every time the nmon command is started, you can place the keys in the NMON shell variable. For example, you can run the following command:

export NMON=mcd

Then, run the nmon command.
To stop the nmon command from the command line, use the kill -USR2 with the nmon process ID.
To print the background process IDs of the nmon recording, run the nmon command with the -p flag.
To limit the processes that the nmon command lists (online and to a file), you can utilize the following options:
  • Set the program names in environment variables from NMONCMD0 to NMONCMD63
  • Use the -C flag with cmd:cmd:cmd parameter. For example, you can enter the following command:
    nmon -C ksh:vi:syncd
To limit the disks that the nmon lists to a maximum of 64 (online only), use the -k flag with the diskname parameter. For example, you can enter the following command:
nmon -k hdisk2,hdisk0,hdisk3
The nmon tool disconnects from the shell during the recording, ensuring that the command continues running even if you log out. This function is not true in the case of recordings triggered using the on-demand recording facility.
Recording or monitoring journaled file system (JFS) statistics in nmon can prevent unloading a file system because the file system is in use while collecting statistics.
Inside workload partitions (WPAR), the nmon command shows global values for processors and memory statistics. The rest of the values are WPAR specific. The following statistics cannot be retrieved inside aWPAR, and the nmon screen does not support them inside aWPAR:
  • Disks, disk I/O graphs, disk busy map, disk groups
  • Disk adapters
  • Paging space
  • Volume group
  • ESS/vpaths
  • Fibre Channel adapters
  • VIOS Shared Ethernet adapters

Flags in Interactive Mode

You can use the following flags in the interactive mode.
ItemDescription
-s < seconds >Time interval between refreshing the screen. The default value is 2 seconds.
-c < count >Number of times the screen must be refreshed.
-g < filename >A file that contains user-defined disk groups that can be specified using the filenameparameter. Each line in the file begins with a group name. The list of hard disks follows the group name and is separated by spaces. The file can contain a maximum of 64 disk groups. A hard disk can belong to various disk groups.
-bDisplays the view in black and white mode.
-BDoes not include boxes in the view. By default, the command displays boxes.
-hDisplays help information.
-k < disklist >Reports only the disks in the disk list.

Flags in Recording Mode

ItemDescription
-AIncludes the Asynchronous I/O section in the view.
-cSpecifies the number snapshots that must be taken by the command. The default value is 10000000.
-dIncludes the Disk Service Time section in the view.
-DSkips the Disk Configuration section.
-ESkips the ESS Configuration section.
-fSpecifies that the output is in spreadsheet format. By default, the command takes 288 snapshots of system data with an interval of 300 seconds between each snapshot. The name of the output file is in the format ofhostname_YYMMDD_HHMM.nmon.
-FSpecifies that the output is in spreadsheet format and the name of the output file is filename. The filenameparameter specifies the name of the output file.
-gSpecifies the file that contains the user-defined disk groups, using the filename parameter. Each line in the file begins with a group name. The list of disks follows the group name and is separated with spaces. The file can contain a maximum of 64 disk groups. A disk can belong to various disk groups.
-GUses Greenwich mean time (GMT) instead of local time. This method is helpful when you compare nmon files from many LPARs of 1 system for processor view but the LPARs are in different time zones.
-ISpecifies the percentage of process threshold at which the command ignores the TOP processes statistics. The default percentage is zero. The command does not save the TOP processes statistics if the process is using less processor than the specified percentage.
-JSkips the JFS section.
-kSpecifies a list of disks to be recorded.
-KIncludes the RAW Kernel section and the LPAR section in the recording file. The -K flag dumps the raw numbers of the corresponding data structure. The memory dump is readable and can be used when the command is recording the data.
-lSpecifies the number of disks to be listed on each line. By default, 150 disks are listed per line. For EMC disks, specify a value of 64.
-LIncludes the large page analysis section.
-mChanges the directory before the command saves the data to a file.
-MIncludes the MEMPAGES section in the recording file. The MEMPAGES section displays detailed memory statistics per page size.
-NIncludes the NFS section in the recording file. To collect the NFSv4 statistics, specify -NN.
-oSpecifies the file name or directory to which the recorded file is to be stored.
-OIncludes the Shared Ethernet adapter (SEA) VIOS sections in the recording file.
-PIncludes the Paging Space section in the recording file.
-rSpecifies the value for the runname field written to the spreadsheet file. By default, the value is the hostname.
-sSpecifies the interval in seconds between 2 consecutive recording snapshots.
-SIncludes WLM sections with subclasses in the recording file.
-tIncludes the top processes in the output. You cannot specify the -t-T, or -Y flags with each other.
-TIncludes the top processes in the output and saves the command-line arguments into the UARG section. You cannot specify the -t-T, or -Y flags with each other.
-VIncludes disk volume group section.
-wSpecifies the size of timestamp (Tnnnn) to be recorded. The timestamp is recorded in the .csv file. The value of the number parameter ranges from 4 through 16. For NMON analyzer, use the values 4 or 8.
-WIncludes the WLM sections into the recording file.
-xSpecifies the sensible spreadsheet recording for duration of 1 day for capacity planning. By default, the recording is done every 900 seconds for 96 times. This flag is equivalent to -ft -s 900 -c 96.
-XSpecifies the sensible spreadsheet recording for duration of 1 hour for capacity planning. By default, the recording is done every 30 seconds for 120 times. This flag is equivalent to -ft -s 30 -c 120.
-YIncludes the top process in the recording with all of the commands of the same name added and recorded. You cannot specify the -t-T, or -Y flags together.
-zSpecifies the sensible spreadsheet recording for duration of 1day for capacity planning. By default, the recording is done every 900 seconds for 96 times. This flag is equivalent to -f -s 900 -c 96.
-ZSpecifies the priority of the nmon command that is running. A value of -20 means important. A value of 20 means not important. Only root user can specify negative value.
-^Includes the Fibre Channel (FC) sections.

Parameters

ItemDescription
disklistSpecifies a list of disks.
dirSpecifies a directory.
dplSpecifies the number of disks to list on each line.
filenameSpecifies a file that contains the disk group you select.
numberSpecifies the number of refreshes.
countSpecified the number of times to record.
percentSpecifies the percentage of processor usage.
prioritySpecifies the priority of processes to be run.
runnameSpecifies the value for the runname field in the spreadsheet file to be run.
secondsSpecifies the interval, in seconds, of refreshing the snapshot.
outputpathSpecifies the path for the output file.

Subcommands

ItemDescription
spaceRefreshes the screen immediately.
.Displays only busy disks and processes.
~Switches to the topas screen.
^Displays the Fibre Channel adapter statistics
+Doubles the screen refresh time.
-Decreases the screen refresh time by half.
0Resets the peak values of statistics (displayed on the screen) to zero. Applicable only for panels that display peak values.
aDisplays the I/O statistics of the adapters.
ASummarizes the Async I/O (AIO server) processes.
bDisplays the view in black and white mode.
cDisplays processor statistics with bar graphs.
CDisplays processor statistics. It is useful for comparison when the number of processors ranges from 15 to 128.
dDisplays the I/O information of disks. To display specific disks only, specify the -k flag.
DDisplays the I/O statistics of disks. To get additional statistics of the disks, press the D key more than once.
eDisplays the I/O statistics of the ESS virtual path logical disks.
gDisplays the I/O statistics of the Disk Group. You must specify the -g flag with this key.
hDisplays the online help information.
jDisplays the JFS statistics.
kDisplays the internal statistics of the kernel.
lDisplays the processor statistics in long format. More than 75 snapshots are displayed with bar graphs.
mDisplays the memory and paging statistics.
MDisplays multiple page size statistics in pages. If you press the M key twice, the statistics are displayed in megabytes.
nDisplays the network statistics.
NDisplays the statistics of the NFS Network file system. If you press the N key twice, you see the NFSv4 statistics.
oDisplays the map of Disk I/O.
ODisplays only the Shared Ethernet adapter VIOS.
pDisplays the statistics of the partitions.
PDisplays the statistics of the paging space.
qQuits. You can also use the x, or Ctrl+C key sequence.
rDisplays the resource type, system name, cache details, AIX® version, and the LPAR information.
SDisplays the WLM with subclasses.
tDisplays the statistics of top processes. You can press the following keys with this subcommand:
  • 1: Displays basic details.
  • 2: Displays accumulated process information.
  • 3: Sorts the view by processor.
  • 4: Sorts the view by size.
  • 5: Sorts the view by I/O information.
uDisplays the top processes with the command arguments. To refresh the arguments for new processes, press the u key twice.
UDisplays the top processes with the command arguments, and the workload class or workload partitioninformation.
vHighlights status of pre-defined system resources and categorizes them as either danger, warnings, or normal.
VDisplays the statistics of the Disk Volume Group.
wDisplays the wait processes when used with the top processes.
WDisplays the statistics of the Workload Manager (WLM).
[Triggers a custom on-demand recording. The recording initiated exits along with the interactive nmon if not stopped earlier.
]Stops a custom recording triggered by ] .

Output Details

This section provides explanations to the metrics that are displayed on nmon screen.
System resources view



This view provides general information about the system resources. To display this view, press the r key. It contains information about the following resources:
  • The number of processors in the system.
  • The number of online processors that are active in the system.
  • The frequency of the processors.
  • The version of AIX and its technical level.
  • The type of the running kernel.
  • The logical partition.
  • The power savings mode of the logical partition.
  • The model of the hardware.
  • The processor architecture of the system.
  • The type of the platform bus.
  • The cache information of processors.
  • The number of active events.
  • The old serial number. This number is the system ID of the partition before the dynamic configuration event.
  • The current serial number. This number is the current system ID or the system ID of the partition after the dynamic configuration event.
  • The local time of the last dynamic reconfiguration event. This information is labeled with the "When" keyword.
AIO Processes View
The AIO processes view provides information about the asynchronous I/O (AIO) processes. To display this view, press the A key. The following columns are displayed on the screen:
ItemDescription
Total AIO ProcessesThe total number of AIO processes.
Actually in useThe number of AIO processes that uses more than 0.1% of the processor.
CPU UsedThe percentage of the processor that is used by all of the kernel processes.
All time peakThe maximum number of kernel processes that are running since the system starts.
Recent peakThe recent maximum number of kernel processes that use more than 0.1% of the processor.
PeakThe maximum percentage of the processor that is used by all of the kernel processes.
Process View
The Process View provides details of the processes in the system. To display this view, press the t key or the v key. It contains the following columns are displayed on the screen:
ItemDescription
pidThe ID of the process.
ppidThe ID of the parent process.
UserThe user ID of the process.
Proc GroupThe ID of the process group.
NiceThe initial priority of a process. This value is set by the nice command.
PriorityThe base schedule priority of a process.
StatusThe status of a program.
Proc_FlagThe flag of a process.
ThrdsThe number of threads.
FilesThe maximum file index that is in use.
ForegroundForeground process or background process.
CommandThe name of the command.
Time StartThe time when the command started.
CPU-TotalThe total time that the process takes since it starts.
Child TotalThe total time that the child process takes since it starts.
Delta-TotalThe total time taken by the process in the interval.
%CPU UsedThe percentage of the processor that is used in the last interval.
Size KBThe size of the pages in kilobytes.
Res SizeThe sum of real-memory data (resident set) and real-memory (resident set) text size of the process.
Res SetThe sum of real-memory data (resident set) and real-memory (resident set) text size of the process.
Res TextThe real-memory text size of the process.
Res DataThe real-memory data size of the process.
Char I/OThe number of I/O characters per second from the last interval.
RAM UseThe percentage of the RAM that is used.
Paging I/OThe I/O page faults per second in the last interval.
Paging OtherThe non-I/O page faults per second in the last interval.
Paging RepagesThe number of repage faults per second in the last interval.
ClassThe Workload Manager class name of the process.
Processor Usage Small View
The Processor Usage Small View provides a brief summary of the user, system, idle, and wait time of logical processors, the corresponding entitlement, and the virtual processor used. You can generate the Processor Usage Small View using the c key.
Processor Usage Large View
The Processor Usage Large View displays the use of logical processor in a graph. To display this view, you can press the C key.
The following labels are used to identify time spent in different modes:
  • s: Labels the percentage of time spent in system mode
  • u: Labels the percentage of time spent in user mode
Shared-Processor Logical Partition View
The Shared-Processor Logical Partition View includes flags that indicate the following information of a partition:
  • Whether the partition is an LPAR or not
  • Whether the partition can be an LPAR or not
  • Whether the partition is shared or dedicated
  • Whether the SMT is turned on or off
  • Whether the shared-partition is capped or uncapped
To display this view, you can press the p key.
Processors:
The following metrics of the processor status are displayed in this view:
ItemDescription
Max Phys in SysMaximum number of physical processors in the system
Phys CPU in systemNumber of physical processors in the system
Virtual OnlineNumber of online virtual processors
Logical onlineNumber of online logical processors
Physical poolNumber of shared physical processors in the shared pool ID that this partition is assigned to
SMT threads/CPUNumber of SMT threads per processor
Capacity:
The following information displays the processor capacity:
ItemDescription
Cap. Processor MinMinimum number of processing units that are defined for this LPAR
Cap. Processor MaxMaximum number of processing units that are defined for this LPAR
Cap. IncrementGranularity at which changes to the entitled capacity can be made
Cap. UnallocatedSum of the number of processor units that are unallocated from shared LPARs in an LPARgroup
Cap. EntitledEntitled capacity
MinReqVirtualCPUMinimum required virtual processors for this LPAR
ID Memory:
The following metrics of the ID memory are displayed:
ItemDescription
LPAR ID Group:PoolID of an LPAR group and its pool ID
Memory (MB/GB) Min:MaxMinimum and maximum memory that is defined for this LPAR in megabytes or gigabytes
Memory(MB/GB) OnlineOnline real memory in megabytes or gigabytes
Memory Region LMBSize in bytes of one logical memory block (LMB)
Time (in seconds):
ItemDescription
Time Dispatch WheelInterval during which each virtual processor receives its entitlement
MaxDispatch LatencyMaximum latency in seconds between the dispatch of the LPAR on the physical processors
Time Pool IdleTime in seconds that the shared processor pool is idle
Time Total DispatchTotal time in seconds that the LPAR dispatches
Minimum and Maximum Values of Processors
The following minimum and maximum values of processors are displayed:
ItemDescription
Virtual CPU ( Min - Max )Minimum number and maximum number of virtual processors in the LPARdefinition
Logical CPU ( Min - Max )Minimum number and maximum number of logical processors
Weight
The following information about the weight of the processor is displayed:
ItemDescription
Weight VariableVariable weight of the processor capacity
Weight UnallocatedUnallocated variable weight available for this partition
NFS Panel
The NFS Panel provides information about the Network File System (NFS). To display this view, press the N key. The following metrics are included in the view:
ItemDescription
RootNFS V2 server and client root requests
WrcacheNFS server and client write cache requests
NullNFS server and client write cache requests
GetattrNFS server and client get attributes requests
SetattrNFS server and client set attributes requests
LookupNFS server and client filename lookup requests
ReadlinkNFS server and client read link requests
ReadNFS server and client read requests
WriteNFS server and client write requests
CreateNFS server and client file creation requests
MkdirNFS server and client directory creation requests
SymlinkNFS server and client symbolic link creation requests
RemoveNFS server and client file removal requests
RmdirNFS server and client directory removal requests
RenameNFS server and client file renaming requests
LinkNFS server and client link creation requests
ReaddirNFS server and client read-directory requests
FsstatNFS server and client file-status requests
AccessNFS V3 server and client access requests
MknodNFS V3 server and client mknod creation requests
readdir+NFS V3 server and client read-directory plus requests
FsinfoNFS V3 server and client file information requests
PathconfNFS V3 server and client path configuration requests
CommitNFS server and client commit requests
Bad callsNFS server and client failed calls
CallsNFS server and client requests
The following NFS V4 client/server statistics are printed when you press the N key twice.
ItemDescription
AccessNFS V4 server and client access requests
acl_readNFS V4 client reading access control list (ACL)
acl_stat_lNFS V4 client retrieving long ACL information
acl_writeNFS V4 client write access control list (ACL)
ClntconfirmNFS V4 client confirm operations
CloseNFS V4 client closing files
CommitNFS V4 server and client committed
CompoundNFS V4 server compound calls
CreateNFS V4 server and client creating a non-regular object
DelegpurgeNFS V4 server purge delegations awaiting recovery
DelegreturnNFS V4 server and client returning delegation
FinfoNFS V4 client obtaining file information
getattrNFS V4 server and client retrieving attributes
getfhNFS V4 server retrieving file handles
LinkNFS V4 server and client linking operations
LockNFS V4 server and client locking operations
lockt/testNFS V4 server testing the specified lock or NFS V4 client lock test
locku/unlockNFS V4 server or NFS V4 client unlock operations
lookupNFS V4 server and client looking up filenames
lookuppNFS V4 server looking up parent directories
mkdirNFS V4 client creating directories
mknodNFS V4 client creating special files
NullNFS V4 server null calls or NFS V4 client null calls
nverifyNFS V4 server verifying difference in attributes
openattrNFS V4 server opening named attribute directories
openconfirmNFS V4 server and client confirming the open for usage
opendowngradeNFS V4 server and client downgrading the access for a specified file
OpenNFS V4 server and client open operations
operationsNFS V4 server and client operations
pcl_readNFS V4 client extracting numeric data from printer control language (PCL) files
pcl_readstat_lNFS V4 client pcl_stat long operations
pcl_statNFS V4 client pcl_stat operations
pcl_writeNFS V4 client pcl_write operations
putfhNFS V4 server setting current file handles
putpubfhNFS V4 server setting public file handles
putrootfhNFS V4 server setting root file handles
readdirNFS V4 server and client reading directories
readlinkNFS V4 server and client reading symbolic links
ReadNFS V4 server and client reading data from files
releaseNFS V4 server and client release_lock operations
removeNFS V4 server and client removing file system object
renameNFS V4 server and client renaming object names
renewNFS V4 server and client renewing leases
replicateNFS V4 client replicate operations
restorefhNFS V4 server restoring file handles
rmdirNFS V4 client removing directories
savefhNFS V4 server saving file handles
secinfoNFS V4 server and client obtaining security information
setattrNFS V4 server and client setting object attributes
setclientNFS V4 server and client setclient operations
statfsNFS V4 client file statistics requests
symlinkNFS V4 client symbolic link operations
verifyNFS V4 client verifying same attributes
writeNFS V4 server and client writing to files
Network Interface View
The Network Interface View shows the statistics errors for the network. You can view this information by pressing the n key.
If the screen is updated 3 times with no network errors, the Network Interface View does not contain the network error statistics.
The following metrics are displayed in this view:
ItemDescription
I/F NameInterface name
Recv-KB/sData received in kilobytes per second in the interval
Trans-KB/sData transmitted in kilobytes per second in the interval
PackinNumber of packets received in the interval
PackoutNumber of packets sent in the interval
InsizeAverage size of packet received in the interval
OutsizeAverage size of packet sent in last interval
Peak->RecvPeak value of received data in kilobytes per second
Peak->TransPeak value of sent data in kilobytes per second
Total RecvTotal received data in megabytes per second
Total SentTotal sent data in megabytes per second
MTUMaximum size of transport unit in bytes
IerrorNumber of input errors
OerrorNumber of output errors
CollisionNumber of collision
Mbits/sAdapter bit rate in megabits per second
DescriptionDescription of the interface
WLM View
The WLM View displays the information about workload management. You can display this view using the W key. To turn on the subclasses section, press the S key from WLM View. To turn off the subclasses section, press the S key again.
The following metrics are displayed in this view:
ItemDescription
CPUPercentage of processor use of the class.
MEMPercentage of physical memory use of the class.
BIOPercentage of disk I/O bandwidth use for the class.
Process (Procs)Number of processes in the class.
Tier (T)Tier number. The value ranges from zero through nine.
Inheritance (I)Values of the inheritance attribute. A value of zero means no. A value of one means yes.
LocationValues of location. A value of one means avoiding transfer of segments to shared classes. Otherwise, a value of zero is displayed.
Disk Busy Map
The Disk Busy Map shows the use statistics of disks. To display this map, press the o key. A maximum of 100 disks is shown per screen. Only the disks with the names ranging from hdisk0 through hdisk100 are displayed. The following table shows the symbols for the ranges of names.
SymbolsNames
_Less than 5
.Less than 10
-Less than 20
+Less than 30
oLess than 40
0Less than 50
OLess than 60
8Less than 70
XLess than 80
#Less than 90
@Less than 100 and equal to 100
Disk Groups
Multiple disks can be monitored by placing them in groups. To display this view, press the g key.
You need to create a group configuration file containing the lines as shown in the following example:
<Group_name1> <disk_name1> <disk_name2> ....
<Group_name2> <disk_nameA> <disk_nameB> ... 
In the example, <Group_name1> is the name of the first disk in the group; <disk_name1> and <disk_name2> are the first and second disks in the group.
To see the Disk Group I/O, run the nmon command with the -g flag and a group file, and then press the g key. The following metrics are shown in this view:
ItemDescription
NameDisk Group name. You can specify a maximum of 64 groups. A disk can be in multiple groups.
DisksNumber of disks in the group.
Read/Write-KB/sData transfer rate of read and written data in kilobytes per second in the interval.
TotalMB/sSum of read and written data in megabytes per second in the interval.
Xfers/sNumber of read and written data transfers per second in the interval.
BlockSizeKBBlock size in kilobytes read or written per transfer operation.
ESS Vpath Statistics View
This view provides the ESS Vpath Statistics. To display this view, press the e key. The following metrics are included in this view:
ItemDescription
NameName of the virtual path.
SizeSize of the ESS path.
AvgBusyAverage busy use of the disk.
Write-KB/sTransfer rate of written data in kilobytes per second in the interval.
Read-KB/sTransfer rate of read data in kilobytes per second in the interval.
Xfers/sNumber of read and write transfers per second.
Total vpathsNumber of virtual paths.
JFS View
This view provides the Journaled File System (JFS) statistics. To display this view, press the j key. The following statistics are recorded in this view:
ItemDescription
FileSystemName of the file system.
Size (MB)Size in megabytes for the file system.
Free (MB)Available free space in megabytes in the file system.
%UsedPercent of file system used.
%InodesPercent of file system used by i-nodes.
Mount pointLocal mount point.
Kernel Statistics
This view contains the statistics of the kernel. To display this view, press the k key. The following statistics are displayed in this view:
ItemDescription
runqueueAverage number of threads that are ready to run but are waiting for an available processor.
pswitchNumber of processor switches per second in the interval.
forkNumber of forks per second in the interval.
execNumber of execs per second in the interval.
msgNumber of interprocess communication (IPC) messages sent and received per second in the interval.
semNumber of semaphore operation system calls per second in the interval.
hw intrpNumber of device interrupts per second in the interval.
sw intrpNumber of off-level handlers called per second in the interval.
SwapinNumber of processes in swap queue per second in the interval.
SyscallNumber of system calls per second in the interval.
readNumber of read calls per second in the interval.
writeNumber of write calls per second in the interval.
readchNumber of characters transferred through read system call per second in the interval.
WritechNumber of characters transferred through write system call per second in the interval.
R + W (MB/s)Number of read and write characters in megabytes per second in the interval.
UptimeTime duration for which the system is up.
igetNumber of inode lookups per second in the interval.
dirblkNumber of 512-byte block reads by the directory search routine to locate an entry for a file per second in the interval.
nameiNumber of vnode lookup from a path name per second in the interval.
kschedNumber of kernel processes created per second in the interval.
koverfNumber of kernel process creation attempts where the user forked to the maximum limit or the configuration limit of processes reached per second in the interval.
kexitNumber of kernel processes that become zombies per second in the interval.
Long Term Processor Averages View
This view provides information about the instantaneous system. To display this view, press the l key. You can use the following labels to identify the time spent in different modes:
  • s: Labels the percentage of the time spent in system mode.
  • u: Labels the percentage of the time spent in user mode.
  • w: Labels the percentage of the time spent in wait mode.
The following metrics are displayed on this view:
ItemDescription
EntitledCPUEntitled capacity of the partition.
UsedCPUNumber of physical processors used by the partition.
Large Page Analysis
This view provides analysis of the large page. To display this view, press the L key. The following information is displayed:
ItemDescription
CountNumber of large pages and their total size.
FreePercentage of free large pages and their size.
In UsePercentage of large pages in use and their size.
SizeSize of a large page.
High water markLarge page high watermark.
Paging Space
This view prints the paging-space statistics. To display this view, press the p key. The following metrics are displayed in the view:
ItemDescription
PagingSpaceNumber of paging space.
Volume-GroupNumber of volume groups.
TypeType of logical volumes. The types can be NFS or LV.
LPsSize of logical partitions.
MBSize in megabytes.
UsedPercentage of use for volume groups.
IOpendingNumber of pending I/O in the paging space.
Active/InactiveActive or inactive paging space.
Auto/NotAutoIndicates whether the paging space is auto loaded or not.
Volume Group Statistics
This view provides statistics for the volume group. To display this view, press the V key. The following information is displayed in the view:
ItemDescription
NameVolume group name.
DisksNumber of disks in the group.
AvgBusyAverage busy of the disks in the volume group.
Read/Write-KB/sData transfer rate of read and written data in kilobytes per second in the interval.
TotalMB/sSum of read and written data in megabytes per second in the interval.
Xfers/sNumber of read and written transfers per second in the interval.
BlockSizeKBBlock size read or written per transfer in kilobytes per second in the interval.
Disk Statistics
This view provides statistics for disks. To display this view, press the D key. You can press the D key for the following times to view various metrics:
  • Once: Shows disk numbers
  • Twice: Shows disk descriptions
  • Three times: Shows service times
  • Four times: Shows disk statistics with graphs similar to the graph shown on pressing the d key
Disk Numbers (Pressing the D key once)
The following metrics are shown in this view:
ItemDescription
NameName of the disks.
BusyAverage busy of the disks.
Read-KB/sData transfer rate of read data in kilobytes per second in the interval.
Write-KB/sData transfer rate of written data in kilobytes per second in the interval.
Transfers/secNumber of read and written transfer per second in the interval.
SizeKBBlock size read or written per transfer in kilobytes per second in the interval.
PeakPeak percentage of average busy.
Peak KB/sPeak read and written data in kilobytes per second.
qDepthNumber of requests sent to disk and are not completed.
Totals Size (GB)Total size of disks in gigabytes.
Totals Free (GB)Total free space left in disks in gigabytes.
Totals Read (MB/s)Total data transfer rate of read data from all disks in megabytes per second.
Totals Write (MB/s)Total data transfer rate of written data to all disks in megabytes per second.
Disk Descriptions (Pressing the D key twice)
The following metrics are shown in this view:
ItemDescription
NameDisk names.
Size (GB)Size of disks in gigabytes.
Free (GB)Free space left in disk in gigabytes.
Disk PathsNumber of paths defined to the disk.
Disk AdapterName of disk adapters.
Volume GroupVolume group that the disk belongs to.
Disk DescriptionDescription of the disk.
Totals Size (GB)Total size of disks in gigabytes.
Totals Free (GB)Total free space left in disks in gigabytes.
Totals Read (MB/s)Total data transfer rate of read data from all disks in megabytes per second.
Totals Write (MB/s)Total data transfer rate of written data to all disks in megabytes per second.
Service Times (Pressing the D key three times)
The following metrics are displayed in the view:
ItemDescription
DiskName of the disk.
Service (in msecs)Average service time per request in milliseconds.
Wait (in msecs)Average waiting time per request in milliseconds.
ServQ sizeAverage number of requests in service queue.
WaitQ sizeAverage number of requests waiting to be accomplished.
ServQ FullNumber of times the disk is not accepting any coming requests.
Totals Size (GB)Total size of disks in gigabytes.
Totals Free (GB)Total free space left in disks in gigabytes.
Totals Read (MB/s)Total data transfer rate of read data from all disks in megabytes per second.
Totals Write (MB/s)Total data transfer rate of written data to all disks in megabytes per second.
Disk Statistics With Graphs (Pressing the D key four times)
This view displays disk statistics with graphs. To display this view, press the d key. The following metrics are displayed in this view:
ItemDescription
NameName of the disk.
BusyAverage percentage of busy for the disk.
Read-KB/sData transfer rate of read data in kilobytes per second.
Write-KB/sData transfer rate of written data in kilobytes per second.
Memory and Paging Statistics
The view provides information about the memory and paging statistics. To display this view, press the m key. The following metrics are included in this view:
ItemDescription
%UsedPercentage of used space in physical memory and paging space.
%FreePercentage of free space in physical memory and paging space.
MB UsedPhysical memory and paging space that are used in megabytes.
MB FreePhysical memory and paging space that are free in megabytes.
Pages/sec to Paging SpaceNumber of I/O pages transferred to or from the paging space per second.
Pages/sec to file systemNumber of I/O pages transferred to or from the file system per second.
Page ScansNumber of page scans by clock.
Page FaultsNumber of page faults.
Page CyclesNumber of page replacement cycles.
Page StealsNumber of page steals.
NumpermNumber of frames used for files (in 4-KB pages).
ProcessPercentage of real memory used by process segments.
SystemPercentage of real memory used by system segments.
FreePercentage of real memory that is free.
TotalPercentage of total real memory used.
Min/MaxpermThe minperm and maxperm values for page steals.
Min/MaxfreeThe minfree and maxfree pages free list.
Min/MaxpgaheadMinimum and maximum number of page ahead pages.
Total VirtualTotal virtual memory.
Accessed VirtualActive virtual memory.
NumclientNumber of client frames.
MaxclientMaximum number of client frames.
UserReal memory used by non-system segments.
PinnedReal memory that is pinned.
The AMS statistics are displayed in the topas_nmon memory panel. To display this view, press the m key. The following metrics are included in this view:
ItemDescription
PoolAMS pool ID of the pool that the logical partition (LPAR) belongs to.
WeightWeight of the variable memory.
pMemPhysical memory currently backing up the logical memory partition (in MB).
hpiNumber of hypervisor page-ins.
hpitTime spent in hypervisor page-ins (in seconds).
Storage Pool Statistics for Next Gen VIOS
This view provides the logical organization of one or more physical volumes in a Next Gen VIOS environment that blocks the storage. The block storage capacity of the storage pool is the summation of the capacity for all physical volumes in the pool.
start of changeThe topas_nmon recording includes the following storage pool metrics:end of change
List of cluster configuration for all nodes:
ItemDescription
start of changeCluster Configurationend of changeRecords the cluster configuration by using the topas_nmon recording for both AIX and VIOS.
CLUSTER NAMEUnique name that is used to identify a specific Next Gen VIOS cluster.
TypeType of the cluster.
SHIDUnique identifier to specify size of the cluster.
UUIDUnique cluster identifier.
Note: start of change The topas_nmon records cluster configuration for both AIX and VIOS while the remaining mapping information is specific to VIOS.end of change
List of available storage pools:
ItemDescription
PoolStorage pool name.
Size (mb)Total size in MB.
Free (mb)Free space in MB.
LUsNumber of logical units.
TypeType of pool.
PoolIDPool identifier.
Logical unit information:
ItemDescription
Lu(Disk) NameLogical unit name.
Size (MB)Total size allocated for the logical unit.
Lu UdidLogical unit identifier.
Adapter I/O Statistics View
This view provides the adapter I/O statistics. To display this view, press the a key. The following metrics are displayed in this view:
ItemDescription
AdapterName of the adapter.
Busy%Bandwidth use of the adapter.
Read-KB/sData transfer rate of read data in kilobytes per second.
Write-KB/sData transfer rate of written data in kilobytes per second.
TransfersNumber of read and write transfers.
DisksNumber of disks.
Adapter-TypeType of the adapter.
Shared Ethernet adapter
This view provides shared Ethernet adapter statistics in a Virtual I/O Server (VIOS). To display this view, press the O key. The following metrics are displayed in this view:
ItemDescription
NumberSerial number.
NameName of the shared Ethernet adapter.
Recv-KB/sData transfer rate of received data in kilobytes per second.
Trans-KB/sData transfer rate of sent data in kilobytes per second.
PackinNumber of packets received per second in the interval.
PackoutNumber of packets sent per second in the interval.
InsizeAverage size per second for received packet in the interval.
OutsizeAverage size per second for outgoing packet in the interval.
Verbose Checks OK/Warn/Danger
This view prints the statistics for processor, memory, and disks. It also prints the status message, such as OK, Warn, or Danger, based on the system metrics exceeding pre-defined threshold values. To display this view, press the v key.
Detailed Page Statistics
This view provides page statistics. To display this view, press the M key.
If you press the M key once, the view contains the statistics in pages. If you press the M key twice, the page statistics are shown in megabytes.
The following metrics are shown in this view:
ItemDescription
NumframesNumber of real memory frames of this page size.
NumfrbNumber of pages on free list.
NumclientNumber of client frames.
NumcompressNumber of frames in compressed segments.
NumpermNumber of frames in non-working segments.
NumvpagesNumber of accessed virtual pages.
MinfreeMinimum free list.
MaxfreeMaximum free list.
NumpoutNumber of page-outs.
NumremoteNumber of remote page-outs.
NumwseguseNumber of pages in use for working segments.
NumpseguseNumber of pages in use for persistent segments.
NumclseguseNumber of pages in use for client segments.
NumwsegpinNumber of pages pinned for working segments.
NumpsegpinNumber of pages pinned for persistent segments.
NumclsegpinNumber of pages pinned for client segments.
numpgsp_pgsNumber of allocated page space.
numrallocNumber of remote allocations.
pfrsvdblksNumber of system reserved blocks.
PfavailNumber of pages available for pinning.
PfpinavailApplication level number pages available for pinning.
system_pgsNumber of pages on segment control blocks (SCB) that are marked withV_SYSTEM.
nonsys_pgsNumber of pages on SCBs not marked with V_SYSTEM.
NumpermioNumber of pageouts in non-working storage.
PgexctNumber of page faults.
PgrclmNumber of page reclaims.
PageinsNumber of paged-in pages.
PageoutsNumber of paged-out pages.
PgspginsNumber of paged-in pages from page space.
PgspgoutsNumber of paged-out pages from page space.
NumsiosNumber of I/O started.
NumiodoneNumber of I/O completed.
ZerofillsNumber of zero-filled pages.
ExfillsNumber of exec-filled pages.
ScansNumber of page scans by clock.
CyclesNumber of clock hand cycles.
pgstealsNumber of page steals.
Fibre Channel Adapter Statistics
This view contains information about the Fibre Channel adapter. You can see this view by pressing the caret (-^) key. The following metrics are included in this view:
ItemDescription
NumberSerial number.
NameName of the Fibre Channel adapter.
Receive-KB/sData transfer rate of received data in kilobytes per second.
Transmit-KB/sData transfer rate of sent data in kilobytes per second.
Requests InNumber of requests received per second in the interval.
Requests OutNumber of requests sent per second in the interval.
OutsizeAverage outgoing packet size per second in the interval.
start of change

Mapping information

The mapping information about shared storage pools that are captured in nmon recording are as follows:
  1. Physical location to client ID
  2. Client ID to virtual target device
  3. Virtual target device to backing device
  4. Cluster to disks
Note: The recording on mapping information is specific to Virtual I/O Server (VIOS) .
end of change

Environment Variables

Environment variables NMON_STARTNMON_ENDNMON_SNAP, and NMON_ONE_IN are used for collecting external data while recording in nmon format.
ItemDescription
NMONCMD0, NMONCMD1, ..., NMONCMD63You can monitor only the processes that are set in these variables when these environment variables are set. Alternatively, you can use the -C flag to restrict the commands in the process listing of the nmon command. For example, you can run the nmon -C db2:websm:nmon:topas command.
NMONContains the set of key strokes corresponding to the initial set of panels to be displayed when the nmon command is started.
TIMESTAMPYou can specify the TIMESTAMP variable to the following values:
TIMESTAMP = 0
The recorded lines contain the nmon Tnnnn timestamps at the beginning of the line and work with the nmon data file.
TIMESTAMP = 1
The lines contains timestamps that have the hours, minute, seconds, day, month, and year. This value can be used if you do not want to merge the data with the nmon file for analysis.
NMON_STARTExternal command to be started when the nmon recording begins.
NMON_ENDExternal command to be started when the nmon recording ends.
NMON_SNAPExternal command to be started periodically to record metrics.
NMON_ONE_INYou can specify the NMON_ONE_IN variable to the following values:
NMON_ONE_IN=1
Runs the snap command every time the recording is done.
NMON_ONE_IN=n
Runs the snap command after the number of recordings specified by the n parameter is done.

Examples

  1. To generate the nmon recording in the current directory for two hours, capturing data every 30 seconds, enter the following command:
    nmon -f -s 30 -c 240
  2. To display the memory and processor statistics immediately after the nmon command is started, do the following steps:
    1. Enter the following command:
      export NMON=mc
    2. Run the nmon command.
  3. To run the nmon command for 20 seconds with the screen refreshing at 10 seconds, enter the following command:
    nmon -c 10 -s 2
  4. To run nmon in black and white mode, enter the following command:
    nmon -b
  5. To view the process information, do the following steps:
    1. Run the nmon command.
    2. Press the t key.
  6. To view the list of views that nmon command provides, press the key h.
  7. The following sample explains the steps to collect external data. In the sample, the mystart file, the mysnap file, and the myend file are executable and are in the path that the $PATH defines.
    1. Set the environment variables as indicated in the following example:
      $export TIMESTAMP=0
      $export NMON_START="mystart"
      $export NMON_SNAP="mysnap"
      $export NMON_END="myend"
      $export NMON_ONE_IN=1
      In the previous example, the value of one is the default value for the NMON_ONE_IN environment variable. It generates one set of external recorded data for every snapshot of nmon recording.
    2. Modify the content of the mystart file as the following:
      ps -ef >start_ps.xt
      echo "PROCCOUNT,Process Count, Procs" >ps.csv
    3. Modify the content of the mysnap file as the following:
      echo PROCCOUNT,$1,`ps -ef | wc -l` >>ps.csv
    4. Modify the content of the myend file as the following:
      echo PROCCOUNT,$1,`ps -ef | wc -l` >>ps.csv
    5. Run the nmon command as follows:
      nmon -f -s 2 -c 10
      The recording finishes in 20 seconds.
    The output of the ps.csv file is similar to the following sample:
    PROCCOUNT,Process Count, Procs
    PROCCOUNT,T0001, 43
    PROCCOUNT,T0002, 43
    PROCCOUNT,T0003, 43
    PROCCOUNT,T0004, 43
    PROCCOUNT,T0005, 43
    PROCCOUNT,T0006, 43
    PROCCOUNT,T0007, 43
    PROCCOUNT,T0008, 43
    PROCCOUNT,T0009, 44
    PROCCOUNT,T0010, 44
    PROCCOUNT,T0010, 44
    To concatenate the generated nmon file with the ps.csv file that is generated by external recording, enter the following command:
    cat  filename.nmon ps.csv > c.csv
    To get the graph, open the c.csv file in nmon analyzer.
  8. start of changeTo view the hdisk details, enter the nmon command with -k flag :
    nmon -k hdisk1,hdisk2
    The previous command shows the disk details for hdisk1 and hdisk2. For hdiskpower devices, enter the following command:
    nmon -k hdiskpower or
    nmon -k power
    Note: The nmon -k hdisk matches all the hdisk devices on the LPAR and does not the match the hdiskpower devices.
    All hdiskpower devices display as power in interactive and recording modes. For example, nmon -k hdiskpower1 matches the device hdiskpower1 and nmon -k hdiskpower matches all hdiskpower devices on the LPAR.
    Note: The output of the lsconf and lspv commands in the nmon recording file is not affected by the changes to the nmon-kcommand

0 blogger-disqus:

Post a comment