Caché Monitoring Guide
Monitoring Performance Using ^pButtons
[Back] [Next]
   
Server:docs2
Instance:LATEST
User:UnknownUser
 
-
Go to:
Search:    

This chapter describes the ^pButtons utility, a tool for collecting detailed performance data about a Caché instance and the platform on which it is running. You can send the report it produces to the InterSystems Worldwide Response Center (WRC) to help diagnose system problems. ^pButtons is similar to Diagnostic Reports (see Using the Caché Diagnostic Report in this guide), but focuses on performance data. ^pButtons works with all versions of Caché since release 5.0.

Note:
This utility may be updated between releases. Contact the InterSystems Worldwide Response Center (WRC) for information about downloading pButtons_package.zip from ftp://ftp.intersys.com/pub/performance/.
You can run the profiles in the Caché Terminal (see Running the ^pButtons Utility in this chapter) or schedule runs using the Task Manager in the Management Portal (see Scheduling the ^pButtons Utility with Task Manger in this chapter). In addition, you can add, modify, and delete profiles using the API that is included with the utility.
This chapter includes the following:
Note:
For a detailed practical look at the generation and use of ^pButtons reports to evaluate system performance by an InterSystems senior technology architect, see InterSystems Data Platforms and performance – Part 1 and InterSystems Data Platforms and performance – Part 2 on InterSystems Developer Community.
Running the ^pButtons Utility
The ^pButtons utility lets you select one or more profiles to run. (The profiles available vary depending on your Caché release and any customization that as been performed.) Based on the selected profile(s), it generates a set of log files, which are placed in the output directory. By default, the output directory is the manager's directory of the Caché instance, (install-dir\mgr); alternatively, you can specify the output directory as described in the Change Output Directory section in this chapter.
By default, ^pButtons provides the following profiles:
To run the ^pButtons utility:
  1. Enter the following command, which is case-sensitive and must be run in the %SYS namespace, in the Caché Terminal:
    %SYS> do ^pButtons
    
  2. From the main menu that is displayed, enter the number of the profile you want to run, or press Enter to exit the main menu:
    Current log directory: c:\intersystems\cache20111\mgr\
    Windows Perfmon data will be left in raw format.
    Available profiles:
         1 12hours - 12-hour run sampling every 10 seconds 
         2 24hours - 24-hour run sampling every 10 seconds 
         3 30mins  - 30-minute run sampling every 1 second 
         4 4hours  - 4-hour run sampling every 5 seconds 
         5 8hours  - 8-hour run sampling every 10 seconds 
         6 test    - 5-minute TEST run sampling every 30 seconds
     
    select profile number to run: 
    
  3. After you enter the profile you want to run, the utility displays information about the data it is collecting:
    select profile number to run: 1
    Collection of this sample data will be available in 1920 seconds.
    The runid for this data is 20111007_1041_30mins.
The generated log files are located in the output directory. The files are identified by the runid, which is uniquely named as follows: YYYYMMDD_HHMM_profile_name.log, where YYYYMMDD_HHMM is the year, month, day, hour, and minute the utility started to collect data; and profile_name is the name of the profile you selected.
After the utility finishes collecting data (that is, at the end of the period of time specified in the profile), you can generate a readable performance report; for information, see the Generating the ^pButtons Performance Reports section in this chapter.
Abort ^pButtons
If you want to stop a runniong profile, you can abort the collection of data and optionally delete all .log files for the profile with the $$Stop^pButtons(runid) command. For example, to abort the collection of data for a report identified by the runid20111220_1327_12hours and delete all .log files written so far, enter the following command in the Caché Terminal in the %SYS namespace:
do Stop^pButtons("20111220_1327_12hours")
To stop the job without deleting log files, enter:
do Stop^pButtons("20111220_1327_12hours",0)
For more information about this command, see $$Stop^pButtons("runid") in the Run ^pButtons Programmatically subsection.
Note:
You must have permission to stop jobs and delete files.
Run ^pButtons Programmatically
You can run the ^pButtons utility programmatically using entry points for the start, collect, preview, and stop functions as described in the following table:
Note:
You can run multiple profiles concurrently.
Command Description
$$run^pButtons("profile")
Starts the specified profile. If successful, returns the runid; if unsuccessful, returns 0.
$$literun^pButtons("profile")
Same as $$run^pButtons("profile"), except that it does not include operating-system data.
This command is intended for servers that are running multiple instances of Caché, where the operating-system data would be duplicated.
$$Collect^pButtons("runid")
Produces a readable HTML performance report file for the specified runid. If successful, returns 1 and the report filename; if unsuccessful, returns 0 followed by a carat and the reason for the failure.
$$Preview^pButtons("runid")
Produces a readable HTML interim (incomplete) performance report file for the specified runid. If successful, returns 1. If unsuccessful, returns 0 followed by a carat and the reason for the failure.
$$Stop^pButtons("runid",[0])
Stops (aborts) ^pButtons from collecting data for a specified runid and by default deletes the associated .log files produced by the utility. To stop data collection without deleting the .log files, include the 0 parameter following the runid. If successful, returns: 1:2:3:4_1:2:3:4; if unsuccessful, returns 0 followed by a carat and the reason for the failure.
The “successful” return status is made up of two parts separated by an underscore: OS-specific and Caché-specific; within each part, colon-separated values specify:
  1. Number of jobs successfully stopped
  2. Number of jobs that failed to stop
  3. Number of files successfully deleted
  4. Number of files not deleted
$$waittime^pButtons("runid") Reports the time until the final HTML file for the specified runid will be complete. If the runid is finished, returns ready now, otherwise returns a string of the form XX hours YY minutes ZZ seconds.
In the following example the runid, which is created by the ^pButtons utility, is obtained programmatically, then tested to determine if a full or interim report has been generated. It determines programmatically that a full report has not been created because the profile has not finished (“0^not ready” is returned), but an interim report has been created (“1” is returned). Based on this information, you know that an HTML file has been generated.
%SYS>set runid=$$run^pButtons("30mins")
 
%SYS>set sc=$$Collect^pButtons(runid)
pButtons run 20111004_1238_30mins is not yet ready for collection.
 
%SYS>write sc
0^not ready
 
%SYS>set sc=$$Preview^pButtons(runid)
 
%SYS>write sc
1
%SYS>
Generating ^pButtons Performance Reports
The ^pButtons utility automatically generates a full (complete) readable HTML performance report from the log files produced by the ^pButtons utility. You can also use the Preview^pButtons entry point to produces an interim (incomplete) report using the data that is being collected by the profile you selected when you ran the ^pButtons utility.
The generated report files are located in the output directory which, by default, is the manager's directory of the Caché instance (install-dir\mgr). The files are uniquely identified by names, which are in the following format: hostname_instance_runid.html, where hostname is the hostname of the system on which the instance of Caché is running; instance is the name of the instance for which performance data has been collected; and runid is the unique identifier generated when the ^pButtons utility was run. If the report is an interim report, _Pn is appended to the file name, where P identifies it as a preliminary report and n is the number of the preliminary report.
Scheduling the ^pButtons Utility with Task Manager
This section provides examples using the Task Manager (see [Home] > [Task Manager]) in the Management Portal to schedule ^pButtons and Collect^pButtons to run.
Note:
The examples describe only the fields that are required. You can edit other fields as desired.
Example 1: Weekly 24-Hour Run
In this example, a task is created to schedule the ^pButtons utility to run a profile named 24hours (which collects performance data for 24 hours) every Thursday at 09:00, and Collect^pButtons to run every Friday at 11:00. To schedule these tasks use the procedure described in Schedule Task Manager in the “Managing Caché” chapter of Caché System Administration Guide.
In this example, first create the task to run ^pButtons:
  1. From the [Home] > [Task Manager] page of the Management Portal, choose the Schedule New Task option to start the Task Scheduler Wizard. Then enter the following information in the specified fields:
  2. Click Next. Then enter the following information in the specified fields:
  3. Click Finish.
Then create a task to run Collect^pButtons:
  1. From the [Home] > [Task Manager] page of the Management Portal, choose options The Schedule New Task option to start the Task Scheduler Wizard. Then enter the following information in the specified fields:
  2. Click Next. Then enter the following information in the specified fields:
  3. Click Finish.
Example 2: Daily 2–Minute Run
In this example, a task is created to schedule the ^pButtons utility to run a profile named 2mins (which collects performance data for two minutes) every day at 12:00, and Collect^pButtons to run every day at 12:30. To schedule these tasks use the procedure described in Schedule Task Manager in the “Managing Caché” chapter of Caché System Administration Guide.
In this example, first create the task to run ^pButtons:
  1. From the [Home] > [Task Manager] page of the Management Portal, choose options The Schedule New Task option to start the Task Scheduler Wizard. Then enter the following information in the specified fields:
  2. Click Next. Then enter the following information in the specified fields:
  3. Click Finish.
Then create a task to run Collect^pButtons:
  1. From the [Home] > [Task Manager] page of the Management Portal, choose options The Schedule New Task option to start the Task Scheduler Wizard. Then enter the following information in the specified fields:
  2. Click Next. Then enter the following information in the specified fields:
  3. Click Finish.
Customizing the ^pButtons Utility
This section describes the tasks you can accomplish with the API:
Change Output Directory
The default output directory for both the log files and the resulting HTML report file is the manager's directory of the Caché instance (install-dir\mgr) for which you are running the ^pButtons utility. You can change the default directory using the commands described in the following table.
Note:
These commands do not affect currently running profiles, whether or not the HTML report files have been produced; that is, no files associated with currently running profiles are moved to the new output directory.
Command Description
do setlogdir^pButtons("directory")
Sets the pathname of output directory to directory; if the does not exist, it is created.
If you do not specify an absolute pathname (for example, C:\pButtonReports), the directory is assumed to be relative to the manager's directory.
do clrlogdir^pButtons()
Resets the output directory pathname to the default manager's directory.
Get Version Information
You can find the current version of the ^pButtons utility using the following commands:
Manipulate Profiles
You can use the APIs described in the following sections to manipulate the profile definitions.
Create New Profiles
You can create a new profile with the following API command:
set rc=$$addprofile^pButtons("profilename","description",interval,count)
where profilename, which is required, must be unique and cannot contain spaces or white-space characters; description, which is required, should be meaningful because it displayed in the menu when you run the ^pButtons utility; interval, which is required, is the frequency with which to run each sample, in seconds (in the range of 1 second to 300 seconds); and count, which is required, is the number of times to run the profile.
Note:
An interval of 1 second is only allowed if the profile duration is an hour or less.
For example, to create a profile named 2minrun that runs a sampling every 10 seconds until it runs 12 samplings (for a total of 120 seconds, or two minutes), enter the following:
set rc=$$addprofile^pButtons("2minrun","A 2-minute run sampling every 10 seconds",10,12)
The next time you run the ^pButtons utility, the list of profiles includes the following profile name and description:
2minrun     A 2-minute run sampling every 10 seconds
Generate Profile
Alternatively, you can quickly generate new profiles (with a meaningful name and description) with the following API command:
set rc=$$genprofile^pButtons("duration"[,interval])
where duration (interval * count), which is required, must be in the format "hh:mm", "hh:", or mm; and interval, which is optional, is the frequency with which to run each sample.
Note:
The maximum duration is 24 hours (86400 seconds); if you specify a longer duration, ^pButtons reduces it to 24 hours. The duration must be double-quoted only if it contains a colon (:); the colon denotes hours.
The minimum interval, if specified, is 2 seconds, unless the duration (that is, interval * count) is less than one hour, in which case the minimum interval is 1 second). If you specify an invalid interval, ^pButtons increases it to the required minimum. If the interval is not specified, it defaults to 10 seconds.
For example, to generate a profile named 12hours (with a generated profile name and description) that runs samples every 5 minutes (300 seconds) over 12 hours, enter the following:
set rc=$$genprofile^pButtons("12:",300)
In addition, to generate a profile named 90mins that runs samples every 10 seconds for 90 minutes, enter the following:
set rc=$$genprofile^pButtons(90)
The next time you run the ^pButtons utility, the list of profiles includes the following profile names and descriptions:
12hours     A 12 hour run sampling every 300 seconds
90mins      A 90 minute run sampling every 10 seconds
Edit Profiles
You can edit an existing profile (except for the predefined “test” profile) with the following API command:
set rc=$$editprofile^pButtons("profilename","description",[interval],[count])
where profilename, which is required, must be unique and cannot contain spaces or white-space characters; description, which is required, should be meaningful because it displayed in the menu when you run the ^pButtons utility; interval, which is optional, is the time to run each sample, in seconds (in the range of 2 seconds to 300 seconds); and count, which is optional, is the number of times to run the profile.
Note:
The arguments are positional; if, for example, to edit the count argument (and keep the value specified in the interval argument), you must include the comma separator, as follows: set rc=$$editprofile^pButtons("2minrun","A 5-minute run sampling every 30 seconds",,50).
If the duration exceeds 24 hours (86400 seconds), it is automatically reduced to 24 hours.
For example, to modify the 2minrun profile to run a sampling every 30 seconds until it runs 10 samplings (for a total of 300 seconds, or five minutes), enter the following:
set rc=$$editprofile^pButtons("2minrun","A 5-minute run sampling every 30 seconds",30,10)
The next time you run the ^pButtons utility, the list of profiles includes the following profile name and description:
2minrun     A 5-minute run sampling every 30 seconds
Copy Profiles
You can copy an existing profile to a file with a different name with the following API command:
set rc=$$copyprofile^pButtons("sourceprofilename","targetprofilename")
where sourceprofilename, which is required, is the name of an existing profile, and targetprofilename, which is required, must be unique and cannot contain spaces or white-space characters.
For example, to make a copy of the 2minrun profile, enter the following:
set rc=$$copyprofile^pButtons("2minrun","5minrun")
The next time you run the ^pButtons utility, the list of profiles includes the following profile names and description:s
2minrun     A 2-minute run sampling every 30 seconds
5minrun     A 2-minute run sampling every 30 seconds
You can now edit the new profile as described in Edit Profiles in this section of the guide.
Delete Profiles
You can delete existing profiles (except for the predefined “test” profile) with the following API command:
set rc=$$delprofile^pButtons("profilename")
where profilename, which is required and must be double-quoted, is the name of the profile you want to delete.
For example, to delete the 2minrun profile, enter the following:
set rc=$$delprofile^pButtons("2minrun")
The next time you run the ^pButtons utility, the list of profiles does not include 2minrun profile.
Performance Reports Created by ^pButtons Utility
The ^pButtons utility generates platform-specific reports as described in this chapter. The report is divided into sections, as illustrated in the following listing:
Configuration

CACHE20161 on machine testsystem

Customer: InterSystems Development
License : 123456

Caché Version String: Cache for Windows (x86-32) 2016.1 (Build 656) Thu Mar 17 2016 17:51:22 EDT
--------------------------------------------------------------
Profile

Profile run "test" started at 10:07 on Jun 01 2016.
Run over 10 intervals of 30 seconds.
--------------------------------------------------------------
license

Product=Enterprise
License Type=Concurrent User
Server=Multi
Platform=Heterogeneous
Licensed Users=1000
Licensed CPUs=16
.
.
.
--------------------------------------------------------------
End of Caché Performance Data Report
The tables in this section describe the sections of each platform-specific report. The sections are listed alphabetically in each table to help you find a specific section more easily. Data that is collected only once is flagged with an asterisk (*). The rest of the data is collected throughout the profile run.
For descriptions of the platform-specific data, see the following tables:
Note:
In all of the following tables, data marked with * is collected once per run.
Caché Performance Data Report for Microsoft Windows Platforms
Section Description
%SS
Four samples taken over the course of the run using the ALL^%SS command.
Configuration *
Caché instance name and hostname from the server, the full Caché version string, the licensed customer name, and the license order number.
cpf file *
A copy of the currently active configuration file.
cstat -c
Four samples taken at even intervals over the course of the run using the command .\bin\cstat -s -p-1 -c-1 -e1 -m8 -n2 -N127. Following is a brief description of each argument:
  • -p-1: samples the process table to include process and global state information.
  • -c-1: samples the Counters section of shared memory to display journal, lock, disk, and resource usage statistics.
  • -e1: the SYSLOG error table.
  • -m8: the file table, which includes all CACHE.DAT and CACHE.EXT files and their attributes.
  • -n2: the network structures table, including local-to-remote database mappings.
  • -N127: ECP statistics for both client and server connections.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide.
cstat -D
Eight samples taken at even intervals over the course of the run using the command ccontrol stat cache --f1 -D10,100. Following is a brief description of each argument:
  • -fl: basic flags.
  • -D10,100: sampling of block collisions every 100 milliseconds over a total sample period of 10 seconds.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide. For information about monitoring block collisions using the ^BLKCOL utility, see the Monitoring Block Collisions Using ^BLKCOL chapter of this guide.
license *
Caché license usage information using Decode^%LICENSE and counts^%LICENSE.
mgstat
Caché-specific data taken over the course of the run using the ^mgstat utility. See the Monitoring Performance Using ^mgstat section of the Caché Monitoring Guide.
perfmon
Output from the Microsoft Windows perfmon utility.
The default presentation of Microsoft Windows perfmon data is raw format. The format can be switched to processed, which removes the repeated server name and splits the datetime column into separate columns, to improve readability.
The following functions allow the querying and updating of the flag that determines whether the perfmon data is manipulated or not:
set rc=$$setperfmonpostproc^pButtons(<onoroff>)
where onoroff can be 1 (on) or 0 (off), or the non-case-sensitive words "on" or "off".
A return code of 1 indicates successful update of the flag, 0 indicates a failed update, and -1 indicates a non-Windows platform.
To determine the current format (raw or processed):
set status=$$getperfmonpostproc^pButtons()
A return code of 1 indicates processed format, 0 indicates raw format.
In addition, the current status of the flag is reported prior to the profile menu display in the interactive run of ^pButtons.
By default, perfmon monitors the counter definitions specified in the default pbctrs.txt file. To monitor previously defined perfmon counters, import the definition into ^pButtons using:
write $$importctrs^pButtons(WindowsCtrName [,pButtonsCtrName [,pButtonsFileName]])
A return code of 0 indicates success and a negative number followed by a reason string indicates failure. Duplicate pButtons counter names are not allowed. If necessary, ^pButtons generates both the internal counter name and file name.
To change the default pButtons counter definition to an exiting definition, use:
write $$setctrdefault(pButtonsCtrName)
Return code of 1 indicates success and 0 followed by a reason string indicates failure. If an invalid counter is specified, the builtin defaualt is set.
To reset the default pButtons counter definition, use:
do clrctrdefault^pButtons()
To associate a specific pButtons counter definition with an existing profile, use:
write $$addctrtoprofile(ProfileName,pButtonsCtrName)
Return code of 1 indicates success and 0 followed by a reason string indicates failure. If either the profile or the counter definition do not exist, the command is not run.
Profile *
Information about the ^pButtons profile that created this log.
tasklist *
Output from the tasklisk -V command; a list of all processes running on the system at the start of the profile run.
Windows info *
Output from the systeminfo command, including the Windows version (excluding hotfix information) and hardware information; for example, processor count, memory installed, and memory used.
Caché Performance Data Report for OpenVMS Platforms
Section Description
%SS
Four samples taken over the course of the run using the ALL^%SS command.
Configuration *
Caché instance name and hostname from the server, the full Caché version string, the licensed customer name, and the license order number.
cpf file *
A copy of the currently active configuration file.
cstat -c
Four samples taken at even intervals over the course of the run using the command ccontrol stat cache -p-1 -c-1 -e1 -m8 -n2 -"N"127. Following is a brief description of each argument:
  • -p-1: samples the process table to include process and global state information.
  • -c-1: samples the Counters section of shared memory to display journal, lock, disk, and resource usage statistics.
  • -e1: the SYSLOG error table.
  • -m8: the file table, which includes all CACHE.DAT and CACHE.EXT files and their attributes.
  • -n2: the network structures table, including local-to-remote database mappings.
  • -"N"127: ECP statistics for both client and server connections.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide.
cstat -D
Eight samples taken at even intervals over the course of the run using the command ccontrol stat cache --f1 -D10,100. Following is a brief description of each argument:
  • -fl: basic flags.
  • -D10,100: sampling of block collisions every 100 milliseconds over a total sample period of 10 seconds.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide. For information about monitoring block collisions using the ^BLKCOL utility, see the Monitoring Block Collisions Using ^BLKCOL chapter of this guide.
f$getsyi(X) * Information on node name, hardware name, version, cluster member, cluster notes, active CPU count, and memory size.
license *
Caché license usage information using Decode^%LICENSE and counts^%LICENSE.
mc sysgen show /all * Output from the mc sysgen show /all command.
mgstat
Caché-specific data taken over the course of the run using the ^mgstat utility. See the Monitoring Performance Using ^mgstat section of the Caché Monitoring Guide.
monitor Uses the VMS MONITOR facility to monitor “all” classes at the same interval specified by the running profile and ending at the specified time, using this command: monitor all /interval=INTERVAL /ending=X.
Profile *
Information about the ^pButtons profile that created this log.
Caché Performance Data Report for Apple Mac OS X Platforms
Section Description
%SS
Four samples taken over the course of the run using the ALL^%SS command.
Configuration *
Caché instance name and hostname from the server, the full Caché version string, the licensed customer name, and the license order number.
cpf file *
A copy of the currently active configuration file.
cstat -c
Four samples taken at even intervals over the course of the run using the command ccontrol stat cache -p-1 -c-1 -e1 -m8 -n2 -N127. Following is a brief description of each argument:
  • -p-1: samples the process table to include process and global state information.
  • -c-1: samples the Counters section of shared memory to display journal, lock, disk, and resource usage statistics.
  • -e1: the SYSLOG error table.
  • -m8: the file table, which includes all CACHE.DAT and CACHE.EXT files and their attributes.
  • -n2: the network structures table, including local-to-remote database mappings.
  • -N127: ECP statistics for both client and server connections.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide.
cstat -D
Eight samples taken at even intervals over the course of the run using the command ccontrol stat cache --f1 -D10,100. Following is a brief description of each argument:
  • -fl: basic flags.
  • -D10,100: sampling of block collisions every 100 milliseconds over a total sample period of 10 seconds.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide. For information about monitoring block collisions using the ^BLKCOL utility, see the Monitoring Block Collisions Using ^BLKCOL chapter of this guide.
ipcs *
Interprocess communication configuration information, including shared memory, semaphores, and message queues; output from ipcs -a command.
license *
Caché license usage information using Decode^%LICENSE and counts^%LICENSE.
MacOS Info *
OS version and hardware information. Output from the sw_vers, uname -a, mount, and netstat commands.
mgstat
Caché-specific data taken over the course of the run using the ^mgstat utility. See the Monitoring Performance Using ^mgstat section of the Caché Monitoring Guide.
Profile *
Information about the ^pButtons profile that created this log.
ps: Four samples taken at even intervals over the course of the run using the command ps -eflv.
sar -d Disk (block) device throughput and latency statistics.
sar -g Page out rates.
sar -n DEV Network device throughput.
sar -n EDEV Network device error rates.
sar -p Page in and page fault rates.
sar -u CPU usage statistics.
sysctl -a * Kernel and system parameter settings.
vm_stat * memory page information.
Caché Performance Data Report for HP HP-UX Platforms
Section Description
%SS
Four samples taken over the course of the run using the ALL^%SS command.
Configuration *
Caché instance name and hostname from the server, the full Caché version string, the licensed customer name, and the license order number.
cpf file *
A copy of the currently active configuration file.
cstat -c
Four samples taken at even intervals over the course of the run using the command ccontrol stat cache -p-1 -c-1 -e1 -m8 -n2 -N127. Following is a brief description of each argument:
  • -p-1: samples the process table to include process and global state information.
  • -c-1: samples the Counters section of shared memory to display journal, lock, disk, and resource usage statistics.
  • -e1: the SYSLOG error table.
  • -m8: the file table, which includes all CACHE.DAT and CACHE.EXT files and their attributes.
  • -n2: the network structures table, including local-to-remote database mappings.
  • -N127: ECP statistics for both client and server connections.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide.
cstat -D
Eight samples taken at even intervals over the course of the run using the command ccontrol stat cache --f1 -D10,100. Following is a brief description of each argument:
  • -fl: basic flags.
  • -D10,100: sampling of block collisions every 100 milliseconds over a total sample period of 10 seconds.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide. For information about monitoring block collisions using the ^BLKCOL utility, see the Monitoring Block Collisions Using ^BLKCOL chapter of this guide.
df -k *
Information about mounted file systems, including mount points, logical volumes, and free space; output from df -k command.
HP-UX info *
OS version and hardware information. Output from theuname -a, file /vmunix, vmstat -f, /usr/sbin/sysdef, and /usr/sbin/kctune -g commands.
ipcs *
Interprocess communication configuration information, including shared memory, semaphores, and message queues; output from ipcs -a command.
license *
Caché license usage information using Decode^%LICENSE and counts^%LICENSE.
mgstat
Caché-specific data taken over the course of the run using the ^mgstat utility. See the Monitoring Performance Using ^mgstat section of the Caché Monitoring Guide.
mount *
Information on all file systems and their mount options.
ps: Four samples taken at even intervals over the course of the run using the command ps -efl.
Profile *
Information about the ^pButtons profile that created this log.
sar -b
buffer cache statistics.
sar -d
Disk (block) device throughput and latency statistics.
sar -Mu
CPU statistics (-u) and per-CPU statistics (-M).
sar -w
Buffer cache statistics.
ServiceGuard * Contents of the/etc/cmcluster.conf if it exists.
vmstat
CPU and memory usage statistics.
Caché Performance Data Report for IBM AIX® Platforms
Section Description
%SS
Four samples taken over the course of the run using the ALL^%SS command.
AIX info *
Output from the oslevel. uname -a, prtconf, and lspv commands
Configuration *
Caché instance name and hostname from the server, the full Caché version string, the licensed customer name, and the license order number.
cpf file *
A copy of the currently active configuration file.
cpu type *
Information on processors installed and whether or not SMT is enabled; output from lsattr -El proc0.
cstat -c
Four samples taken at even intervals over the course of the run using the command ccontrol stat cache -p-1 -c-1 -e1 -m8 -n2 -N127. Following is a brief description of each argument:
  • -p-1: samples the process table to include process and global state information.
  • -c-1: samples the Counters section of shared memory to display journal, lock, disk, and resource usage statistics.
  • -e1: the SYSLOG error table.
  • -m8: the file table, which includes all CACHE.DAT and CACHE.EXT files and their attributes.
  • -n2: the network structures table, including local-to-remote database mappings.
  • -N127: ECP statistics for both client and server connections.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide.
cstat -D
Eight samples taken at even intervals over the course of the run using the command ccontrol stat cache --f1 -D10,100. Following is a brief description of each argument:
  • -fl: basic flags.
  • -D10,100: sampling of block collisions every 100 milliseconds over a total sample period of 10 seconds.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide. For information about monitoring block collisions using the ^BLKCOL utility, see the Monitoring Block Collisions Using ^BLKCOL chapter of this guide.
df -k *
Information about mounted file systems, including mount points, logical volumes, and free space; output from df -k command.
filesystems *
Current /etc/filesystems file.
ioo -a * Current values of I/O tunable parameters; output from ioo -a command.
Included only if the user initiating the ^pButtons profile run has root access.
iostat -DIT
Long listing of extended disk/device statistics with sample time for IBM AIX® 5.3 and newer; output from iostat -DIT command.
Information varies for releases before IBM AIX® 5.3.
ipcs *
Interprocess communication configuration information, including shared memory, semaphores, and message queues; output from ipcs -a command.
license *
Caché license usage information using Decode^%LICENSE and counts^%LICENSE.
mount *
Information on all file systems and their mount options.
mgstat
Caché-specific data taken over the course of the run using the ^mgstat utility. See the Monitoring Performance Using ^mgstat section of the Caché Monitoring Guide.
Profile *
Information about the ^pButtons profile that created this log.
ps: Four samples taken at even intervals over the course of the run using the command ps aux.
sar -d
Included only if the user initiating the^pButtons profile run has root access and /usr/sbin/sar exists.
sar -r
Included only if the user initiating the^pButtons profile run has root access and /usr/sbin/sar exists.
sar -u
CPU statistics that includes micropartitioning information if used.
Included only if the user initiating the^pButtons profile run has root access and /usr/sbin/sar exists.
vmo –a
Current values of virtual memory tunable parameters; output from vmo -a command.
Included only if the user initiating the^pButtons profile run has root access.
vmstat -s *
Absolute counts of virtual memory statistics, including total page ins and page outs.
vmstat -t
Virtual memory and CPU (paging, queuing, and CPU) statistics with timestamps.
vmstat -v *
Samples virtual memory statistics, including free pages, pbuf usage, and fsbuf usage.
Caché Performance Data Report for Oracle Solaris Platforms
Section Description
%SS
Four samples taken over the course of the run using the ALL^%SS command.
Configuration *
Caché instance name and hostname from the server, the full Caché version string, the licensed customer name, and the license order number.
cpf file *
A copy of the currently active configuration file.
cstat -c
Four samples taken at even intervals over the course of the run using the command ccontrol stat cache -p-1 -c-1 -e1 -m8 -n2 -N127. Following is a brief description of each argument:
  • -p-1: samples the process table to include process and global state information.
  • -c-1: samples the Counters section of shared memory to display journal, lock, disk, and resource usage statistics.
  • -e1: the SYSLOG error table.
  • -m8: the file table, which includes all CACHE.DAT and CACHE.EXT files and their attributes.
  • -n2: the network structures table, including local-to-remote database mappings.
  • -N127: ECP statistics for both client and server connections.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide.
cstat -D
Eight samples taken at even intervals over the course of the run using the command ccontrol stat cache --f1 -D10,100. Following is a brief description of each argument:
  • -fl: basic flags.
  • -D10,100: sampling of block collisions every 100 milliseconds over a total sample period of 10 seconds.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide. For information about monitoring block collisions using the ^BLKCOL utility, see the Monitoring Block Collisions Using ^BLKCOL chapter of this guide.
df -k *
Information about mounted file systems, including mount points, logical volumes, and free space; output from df -k command.
license *
Caché license usage information using Decode^%LICENSE and counts^%LICENSE.
mgstat
Caché-specific data taken over the course of the run using the ^mgstat utility. See the Monitoring Performance Using ^mgstat section of the Caché Monitoring Guide.
Profile *
Information about the ^pButtons profile that created this log.
ps: Four samples taken at even intervals over the course of the run using the command ps -efl.
sar -b
Buffer cache usage and efficiency statistics.
sar -d
Disk (block) device throughput and latency statistics.
sar -r
Free memory report.
sar -u
CPU usage statistics include iowait percentage.
vmstat
CPU, queue and memory usage statistics.
Caché Performance Data Report for Red Hat Linux/SuSE Linux Platforms
Section Description
%SS
Four samples taken over the course of the run using the ALL^%SS command.
Configuration *
Caché instance name and hostname from the server, the full Caché version string, the licensed customer name, and the license order number.
cpf file *
A copy of the currently active configuration file.
cstat -c
Four samples taken at even intervals over the course of the run using the command ccontrol stat cache -p-1 -c-1 -e1 -m8 -n2 -N127. Following is a brief description of each argument:
  • -p-1: samples the process table to include process and global state information.
  • -c-1: samples the Counters section of shared memory to display journal, lock, disk, and resource usage statistics.
  • -e1: the SYSLOG error table.
  • -m8: the file table, which includes all CACHE.DAT and CACHE.EXT files and their attributes.
  • -n2: the network structures table, including local-to-remote database mappings.
  • -N127: ECP statistics for both client and server connections.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide.
cstat -D
Eight samples taken at even intervals over the course of the run using the command ccontrol stat cache --f1 -D10,100. Following is a brief description of each argument:
  • -fl: basic flags.
  • -D10,100: sampling of block collisions every 100 milliseconds over a total sample period of 10 seconds.
For more information about the cstat utility, see the Monitoring Caché Using the cstat Utility appendix of this guide. For information about monitoring block collisions using the ^BLKCOL utility, see the Monitoring Block Collisions Using ^BLKCOL chapter of this guide.
df -k *
Information about mounted file systems, including mount points, logical volumes, and free space; output from df -k command.
free -m
Memory usage statistics in MB (-m).
iostat
CPU and disk throughput.
license *
Caché license usage information using Decode^%LICENSE and counts^%LICENSE.
mgstat
Caché-specific data taken over the course of the run using the ^mgstat utility. See the Monitoring Performance Using ^mgstat section of the Caché Monitoring Guide.
Profile *
Information about the ^pButtons profile that created this log.
ps: Four samples taken at even intervals over the course of the run using the command ps -efly.
sar -d
Disk (block) device throughput and latency statistics.
sar -u
CPU usage statistics include iowait percentage.
vmstat -n
CPU, queuing, paging statistics. Only one header is printed (-n) .