Tru64 UNIX Collect

»

HP Tru64 UNIX

Tru64 UNIX

» Tru64 UNIX V5.1B-6
» Tru64 UNIX V5.1B-5
» Documentation
» Information library
» Software web index
» Software products library
» Patch database
» Services
» Developer & Solution Partner Program
» Send us your comments
» Support Statements

Evolving business value

» Tru64 UNIX to HP-UX 11i transition benefits calculator
» Alpha RetainTrust Program
» Transition

Related links

» Alpha systems
» HP-UX 11i
» Integrity servers
» Linux
» HP storage
» HP solutions
HP-UX 11i: measurably better TCO!

» Overview

» Features

Product details

» Resources

Options

Collect operates with a large number of options. The following option descriptions will help you understand the full capability of Collect.

Option Description
–(dash)
 
Directs the output from the Collect utility to stdout, which is usually the screen or window from which the Collect command was entered. This is the default behavior if no data collection file is specified using the -f option.
 
–a
 
Simultaneously displays Collect data on the screen (stdout) while also recording the data in a file when the -f option is specified.
 
–C [start_time],
[end_time]
 
Extracts a series of samples from a file according to the specified start time and end time for the series. The format of the time string is:

[+]Year:Month:Day:Hour:Minute:Second

For example: +2010:11:18:23:30:00. Every time string field except the Second field is optional.

The optional plus (+) sign at the beginning of the time string indicates that time is relative to the beginning of the data collection period. If + is not specified, the -C option indicates absolute time. When time string fields are absent and absolute format is specified, the values from the beginning of collection are used.

If start_time is omitted, the start of the collection period is used. If end_time is omitted, the end of the collection period is used.
 

–d
 
Prints debug information to stdout.
 
–D device1,
[device2, ...deviceN]
 
Specifies which disks are included for data collection, using the device special filename of the disks, such as dsk3 for SCSI disk number 3. You can obtain a list of disk devices from the device directories under /dev, such as /dev/disk or /dev/tape. The hwmgr command can also be used to find devices. See the hwmgr(8) reference page for information on the command options.

You can also use regular expressions to specify a group of disks. For example dsk* selects all disks. For information on regular expressions, refer to the grep(1) reference page or the Programming Support Tools guide.
 

–e pmdtlncfyh
 
Excludes the specified subsystems from the data collection and playback. Do not enter a space between letters when specifying options. For example, the following command specifies that only the CPU and file system data are excluded:

# collect -e cf

The option letters map to the following subsystems:

p –proc
Specifies that the Process Statistics (RSS and VSZ in megabytes) are excluded from data collection. When included, this data appears similar to the following in the output from the collect command:
 

PID User %CPU RSS VSZ UsrTim SysTim IBk OBk Maj Min Command
0 root 2.1 12M 342M 0.00 0.01 0 0 0 0 kern idle
.
.
.
  m –mem
Specifies that the MEMORY STATISTICS are excluded from data collection. When included, this data appears similar to the following in the output from the collect command:

# MEMORY STATISTICS=
 

(<-------- MegaBytes -------->) (<-------- Pages/sec -------->)
Free Swap Act InAc Wire UBC PI PO Zer Re COW SW HIT PP ALL
135 1 44 22 40 36 0 0 7 0 0 0 0 0 0

d –disk
Specifies that the DISK Statistics are excluded from data collection. When included, this data appears similar to the following in the output from the collect command:
 
DISK Statistics
DSK NAME B/T/L R/S RKB/S W/S WKB/S AVS AVW ACTQ WTQ %BSY
0 dsk0 0/0/0 53 431 0 5 8.49 0.00 0.46 0.00 43.09
 
t –tape
Specifies that tape device data is excluded from data collection.

l –LSM volume
Specifies that LSM volume data is excluded from data collection. When included, this data appears similar to the following in the output from the collect command (Collect displays one volume at a time):
 

# LSM Volume Statistics
#VOL NAME R/S RKB/S RAVS W/S WKB/S WAVS
1 rootvol 0 0 0.00 0 12 45.62
 
n –net
Specifies that the Network Statistics are excluded from data collection. When included, this data appears as follows in the output from the collect command:
 
# Network Statistics
#Cnt Name Inpck InErr Outpck OutErr Coll IKB OKB %BW
2 tu0 89 0 2 0 0 10 0 0
 
c –CPU
Specifies that the CPU SUMMARY and CPU STATISTICS are excluded from data collection. When included, this data appears as follows in the output from the collect command:
 
CPU SUMMARY
USER SYS IDLE WAIT INTR SYSC CS RUNQ AVG5 AVG30 AVG60 FORK VFORK
13 16 71 0 149 492 725 0 0.13 0.05 0.01 0.30 0.00
SINGLE CPU STATISTICS
CPU USER SYS IDLE WAIT
0 13 16 71 0
 
f –file system
Specifies that the FileSystem Statistics are excluded from data collection. When included, this data appears as follows in the output from the collect command:
 
# FileSystem Statistics
# FS Filesystem Capacity Free
0 root_domain#root 128 30
1 /proc 0 0
2 usr_domain#usr 700 147
3 usr_domain#var 700 147
 
y –tty
Specifies that the TTY Statistics are excluded from data collection. When included, this data appears as follows in the output from the collect command:
 
# TTY Statistics
# In Out Can Raw
  3 1489 0 3
 
h –header
Specifies that the RECORD N headers are excluded from data collection. When included, this data appears as follows in the output from the collect command:

# RECORD 1 (943046665:0)
(Fri Nov 30 16:24:25 2000) ##

See also the -s option, which you use to specify subsystems that must be included in data collection.
 

–f
 
Records data in the specified file. The argument is a path name to a file such as /usr/users/collectdata/nov13. By default, the Collect utility creates a compressed file and appends a .cgz extension to the file name that you specify. For example, the file nov13 will be created as nov13.cgz by Collect. See the -o option if you want to create an uncompressed file.

As the file created is in binary format, you must use the Collect utility's -p option to replay the contents of the file.

See also the -a option, which enables you to simultaneously direct the output from the Collect utility to stdout (usually the terminal from which the Collect utility is invoked). You can also specify other data collection options with the -f option, such as -s or -n, to control what information is recorded in the file.
 

–F
 
Display or record full process information process lines, which are longer than 80 columns. The process priorities are shown and the RSS and VSZ values are in kilobytes rather than formatted to fit into four columns. The following is example output, except that here the column widths have been manually adjusted to show the example output:
 
## RECORD 1 (943048211:40) (Fri Nov 1 16:50:11 1999) ##
 
Process Statistics (RSS & VSZ in KBytes)
PID PPID User %CPU RSS VSZ UsrTim SysTim Pri IBk OBk Maj Min Command
0 root 2.4 1984 3744 0.00 0.02 0 0 0 0 0 0 kern idle
1 0 root 0.0 96 480 0.00 0.00 44 0 0 0 0 init
.
.
.
  Compare the preceding output to the example output for Process Statistics shown in the entry for the -e option.
 
–h
 
Display a usage summary (help) for the collect command line options.
 
–H h d w m time
[,how_long]
 
Runs the Collect utility in historical mode. The how_long argument defines the length of time that the logs are preserved. The how_long argument is optional, and if you do not specify it the log preservation default is one week. The following values for time can be specified for each argument:

h &nd hourly at [0-59] minutes
An hourly rollover at the specified minute.
A value of -Hh3 will roll over the Collect log every hour at three minutes past the hour. For example: 0.03, 1:03, 2:03, and 3:03.

d –daily at [00:00-23:59] hours and minutes
A daily roll-over at the specified hour and minute in 24-hour time format.

For example, a value of -Hd14:2 will roll over the Collect log every day at system time 14:02 (2:02 PM).

w –weekly at [0-6@00:00-23:59] days, hours, and minutes
A weekly roll-over at the specified day, hour and minute in seven-day 24-hour time format. A value of Zero in the day field represents Sunday.

For example, a value of -Hw1@10:25 will roll over the Collect log every Monday at 10:25 AM.

m –monthly at [1-31@00:00-23:59] date, hours, and minutes
A monthly roll-over at the specified date, hour and minute in 31-day, 24–hour time format.

For example, a value of -Hm3@21:15 will roll over the Collect log every third day of the month at 21:15 (9:15 PM).

As for the value of time, you can specify day and week values for how_long. For example, -Hd14:12,2d5w will roll over the log every day at 14:12 (2:15 PM) and keep the log for 2 days and 5 weeks.
 

–i I:[PI]
 
Specifies a time value for the interval (I) and, optionally, the process interval (PI) in seconds. This enables you to control the rate at which data is collected from subsystems. Floating-point values are permitted.

When you use this option, the initialization message echoed by the Collect utility is updated to confirm the value of I, as follows:

# collect -i 2:8
Initializing (2.0 seconds) (float OK)
#collect -i 5:12
PROC_INTERVAL must be
evenly divisible by INTERVAL

Note that in the second command, an error message is displayed as the value of PI must always be evenly divisible by the value of I.
 

–l
 
Seek to last valid record and print it. This is primarily used by the graphical interface to get the ending time of the collection period.
 
–L vol1//vol01
 
Collects data from one disk group listed in /dev/vol.
 
–M
 
Monitors free disk space. Collect suspends writing to disk when free disk space rises above a declared threshold, and resumes when free space rises above the threshold. The format is as follows:

-M [suspend_value,resume_value]

In the following example, Collect suspends disk writes when free disk space falls below 250 MB, and resumes writing when free disk space rises above 300 MB:

# collect -M 250,300
 

–nNum
 
Select only top Num processes, where Num is an integer. This option is useful with the -S sorting option.
 
–o [tmfnzlq]
 
Options that enable you to control the data collection procedure:

–t (time)
Shows absolute system and user time (T in data recorded for the process subsystem, the way the ps command does. The default is to show a one-second normalized delta since the last sample, thus making graphs of these time values more useful.

–m (memory)
Shows 8192 byte pages instead of megabytes (MB) for absolute memory values.

–f (force)
Do not prompt before overwriting an existing output file.

–n (nice)
Do not allow the Collect utility to set high scheduling priority for itself using nice.

–z (zipped)
Do not write a compressed (zipped) output file.

–l (lock)
Prevents the Collect utility from locking its pages into memory.

–q (queue)
Causes the Collect utility to use instantaneously measured queue lengths, instead of calculated averages.
 

–p collect_datafile
[–f output_datafile]
 
When the existing collect_datafile is specified alone, the Collect utility will play back the contents of the file to stdout (usually the terminal window from which the collect command was entered). You use options such as -e to filter the data read from the collect_datafile. As the file contents will be large, you may want to pipe the output to the more command or use the grep command to search for specific data items.

To convert data files created using previous versions of Collect, use the -f option to specify an output_datafile.
 

–P pid1 ... | [[ Ppid1... ]|
[Ccommand1 ... ]|
[Uuser/UID1 ... ]]
 
The following process identifiers can be specified:

pid [pid,...pid]
Collect data only for processes in list. Specify % to include the process for the collect command.

Ppid [pid,...pid]
Collect data only for processes whose parent PID (PPID) is specified, or that are members of a process group (PGID) with the same ID.

Ccommand [command,...command]
Collect data only for processes whose process names contain above string. This can be a partial string, but must match exactly. Regular expressions are not allowed.

Uuser/UID[user/UID,...user/UID]
Collect data only for processes owned by the specified users. User identifiers (UIDs) can be used in place of the user name. See the /etc/passwd file for a list of user account names and associated UIDs.
 

–R
 
Specify the duration of data collection. Either of the following formats can be specified:
 
 
Num Unit [ Num Unit [ ...] ]
Where Num is an integer and Unit is one of:
    w – weeks, such as 4w for four weeks.
    d – days, such as 2d for two days.
    h – hours, such as 12h for twelve hours.
    m – minutes, such as 30m for thirty minutes.
    s – seconds, such as 45s for 45 seconds.
    Any valid combination of times can be entered, such as 4w2d6h45m20s.
[+]Year:Mon:Day:Hour:Min:Sec
    The same time format described for the -C option, except that a plus (+) sign indicates the value is relative to the current time. Without a +, the value is an absolute time at which the data collection period should end.
 
–s [pmdtlncfyh]
 
Include the specified subsystems in data collection and play-back, which can be: Proc, Mem, Disk, Tape, Lsm, Net, Cpu, Filesys, mQueue, ttY, and Header. The option letters (p m d...) map to these subsystems and are described under the entry for the -e option.

Do not enter a space between letters when specifying options. For example, the following command specifies that only the CPU and file system data are included:

# collect -s cf

Note that if you specify a subsystem that is not available on the local system, only a RECORD N header will be displayed. The following example shows what happens when t (tape) is specified, but no tape device exists on the system:

# collect -s t
.
.
#### RECORD 4 (943046239:0)
(Fri Nov 19 16:17:19 1999) ####
#### RECORD 5 (943046249:0)
(Fri Nov 19 16:17:29 1999) ####
 

–S
 
Sorts processes according to their %CPU usage (percentage of processing time used).
 
–t
 
Prefixes a "tag" to all data lines to facilitate manipulation of data by scripts.
 
–T
 
Specifies only that total disk and tape throughput be recorded or displayed as the Sum MB/sec. All other subsystems are deselected.
 
–v
 
Enables verbose mode, listing the devices attached to the system, as in the following example:

cariad > collect -v
No objects found of type
hardware/tape

  found 4 Disks, 0 Tapes
  found CPU 0 at slot [0]
  found CPU 1 at slot [1]
  max_procs = 16384
  SAMPLE: 0
  Initializing (10.0 seconds)

 

–V
 
Displays Collect executable and data file versions. Also, if used with the -p (play-back file) option, also displays the version of the data file.
 
–W
 
Declares how often Collect should write to disk, using the following format:

-p number_unit

Where number is an integer representing the number of write instances and unit is one of the time options (wdhm), such as in the following examples:

-W 1h

Collect writes data to disk once per hour.

-W 1h30m

Collect writes data to disk every 90 minutes.
 

More Collect product details

» Overview
» Data fields
» Examples
» Product availability
» Frequently asked questions