This guide explains how to configure the Performance Co-Pilot™ (PCP) software to monitor INFORMIX™-OnLine database performance. The software described here is packaged as the Performance Co-Pilot for Informix 7 product. You need to have both Performance Co-Pilot (version 1.1 or later) and INFORMIX-OnLine (version 7) installed on your system to take advantage of this software.
To use this product, you must be familiar with the operation and administrative procedures for the Performance Co-Pilot base product and the INFORMIX-OnLine Serve. For some of the more technical discussion in this document, you may need to obtain and read the following reference material:
Performance Co-Pilot User's and Administrator's Guide, shipped online as part of the Performance Co-Pilot base product and viewable under IRIS InSight.™
INFORMIX-OnLine Dynamic Server Administrator's Guide, available from Informix Software, Inc.
The Performance Co-Pilot for Informix 7 product consists of extensions to both the collection and monitoring facilities of the PCP. The collection of performance metrics is extended as follows:
Informix System Monitoring Interface (SMI) metrics are integrated into the PCP framework with a new Performance Metrics Domain Agent (PMDA).
Source code for a PMDA to measure response times for an Informix application is shipped as part of the product. This agent software can be customized to collect site-specific performance measures for a “typical” Informix application.
The monitoring functionality of Performance Co-Pilot is extended to include:
An Informix virtual processors visualizer, vpvis(1), which provides a three-dimensional view of processor resources consumed by Informix database servers. The visualizer also provides a template for customized site-specific visualizations of database performance using the Performance Co-Pilot pmview(1) construction kit.
Example configurations for the Performance Co-Pilot strip chart tool pmchart(1) are provided for Informix performance metrics.
Performance Co-Pilot logger pmlogger(1) configuration files designed to log the performance metrics required by the visualization tools are included.
This section presents an overview of the Performance Co-Pilot architecture for collecting and monitoring performance data. Read this section if you intend to collect and/or monitor Informix performance, but are not thoroughly familiar with the Performance Metrics Name Space (PMNS) and the Performance Metrics Collection Subsystem (PMCS), as presented in the Performance Co-Pilot User's and Administrator's Guide.
On any given system, the Performance Co-Pilot is installed in one or both of the following configurations:
| Collector Configuration |
| |
| Monitor Configuration |
|
Similarly, both of the Performance Metrics Domain Agents (PMDAs) shipped as part of the product may be installed in either a Collector or a Monitor configuration. The Collector configuration allows for the collection (as well as monitoring, given a current Monitor license) of performance data. A Monitor configuration enables monitoring of performance of remote systems only.
The Performance Metrics Collection Daemon (PMCD) is responsible for gathering performance data relating to the system on which it runs. It manages a collection of Performance Metrics Domain Agents (PMDAs, or simply agents), each of which is responsible for gathering data within its assigned domain. For example a program named pmdainfmx7 collects performance metrics from the INFORMIX-OnLine Server domain.
If you intend to monitor multiple INFORMIX-OnLine servers, be aware that each performance domain is represented by a unique integer domain number. Consequently, you need to plan the allocation of unique and consistent domain numbers for these servers. This issue is addressed separately for the two agents, shipped as part of the product, in the sections titled “The Informix 7 Agent” and “The Informix Response-Time Agent.”
Refer to the Performance Co-Pilot User's and Administrator's Guide and the reference page for pmcd(1) for more detailed information on PMNS, PMCD, agents, and domains.
Figure 1-1 shows a typical use of the Performance Co-Pilot to monitor an Informix database server running on a remote system. The server system often has no graphics capability, so a workstation is used to monitor its performance. Notice that each system has its own unique copy of the name space.
Figure 1-2 shows a more complicated arrangement. Note once again that each system has its own distinct copy of the name space. The name space on any system needs to contain entries only for the INFORMIX-OnLine Servers that it may monitor. For example the monitoring system's name space contains entries for all of the servers. The Collector systems usually contain name space entries only for their own database servers[1] .
Both of the PMDAs that are part of the product are supplied with three shell scripts:
| Install | Installs the agent for a given INFORMIX-OnLine Server in either Collector or Monitor configuration. | |
| Remove | Undoes the configuration for a given server. | |
| Verify | Verifies the correctness of a configuration and availability of performance data. |
For a Collector configuration, the Install script performs the following chores:
Adds the new metric names into the name space in the directory /usr/pcp/pmns on the current host.
Creates an ndbm(3B) database of help text for the new metrics.
Installs the PMDA executable into the directory /usr/pcp/lib.
Installs the relevant pmchart(1) and pmlogger(1) configuration files into the directories /usr/pcp/config/pmchart and /usr/pcp/pmlogger, respectively.
Updates the PMCD configuration file /etc/pmcd.conf.
Notifies PMCD of the new configuration.
At your discretion, when the Install script has been run, you can run the Verify script to check that the new metrics are now available.
For a Monitor configuration, only the following subset of the above actions is required:
Adds the new metric names into the name space in the directory /usr/pcp/pmns on the current host.
Installs any relevant pmchart(1) and pmlogger(1) configuration files into the directories /usr/pcp/config/pmchart and /usr/pcp/pmlogger, respectively.
At your discretion, when the Install script has been run, you can run the Verify script to check that the new metrics are now available.
This is an initial version of the Performance Metrics Domain Agent (PMDA) for collecting performance data from an INFORMIX-OnLine Server (version 7) See Figure 1-3. This PMDA integrates performance metrics obtained using the Informix System Monitoring Interface (SMI) into the Performance Co-Pilot framework. From here on, this agent is called the infmx7 PMDA.
In outline, the installation procedure for the infmx7 PMDA proceeds as follows:
Software Installation – Use the software installation tool inst(1M) to copy the software from the distribution media. The software is placed in the directory /usr/pcp/pmdas/infmx7.
Collector Configuration – Use the shell script Install in the directory /usr/pcp/pmdas/infmx7 to configure and initiate execution of the agent on the hosts where metrics are to be collected.
Monitor Configuration – Use the shell script Install in the directory /usr/pcp/pmdas/infmx7 to configure the name space for the agent on the hosts where metrics are to be monitored only.
The agent configuration may subsequently be undone using the shell script Remove and the availability of performance data verified using the shell script Verify. Both of these scripts can be found in the directory /usr/pcp/pmdas/infmx7.
 | Note: This PMDA has been tested on the following releases of the INFORMIX-OnLine Server product. While the agent may remain partially functional for other releases of Informix, metric values may be missing or incorrectly reported. Therefore, make sure that this agent is installed to monitor this release only: |
INFORMIX 7.1 UD1 for IRIX 5.3.
A valid installation of the Performance Co-Pilot product, version 1.1 (PCP 1.1) or later, is a prerequisite. Further, a Collector configuration of PCP 1.1 is required for a Collector configuration of this PMDA.
Before you attempt to configure the PMDA for a particular OnLine server, you need to establish the server name (INFORMIXSERVER), and for a Collector configuration, the Informix home directory (INFORMIXDIR). Further, if you intend to monitor multiple Informix OnLine servers you need to plan the allocation of performance metrics domain numbers for these servers. The following sections cover these issues.
The Install script requires that you supply the server name of the Informix OnLine instance to which the PMDA connects to obtain performance data. This name is typically supplied as the value of the environment variable INFORMIXSERVER when executing database applications or administrative programs. If you do not know what name you should use, contact the database administrator at your site.
The server name is also used to construct the performance metric names for the metrics supported by this PMDA. For example, a site with the servers production and development has the performance metric names:
informix.production.profile.dskreads informix.development.profile.dskreads |
denoting the number of disk reads by the production and development servers, respectively.
When you install the Informix PMDA in a Collector configuration (such as at the host where performance data is being collected), you need to supply the name of the Informix home directory. This name is typically given as the value of the environment variable INFORMIXDIR when executing database applications or administrative programs. If you do not know what name to use, contact the database administrator at your site.
In order to distinguish metrics from multiple OnLine servers, each server needs to be assigned a unique performance metrics domain number. For a detailed discussion of domain numbers, refer to the Performance Co-Pilot User's and Administrator's Guide.
The range of domain numbers 48-55 is reserved for the SMI metrics from Informix database servers. For example, a site with servers called production and development would normally assign the numbers 48 and 49 for the two servers. For an existing configuration, you can find the domain number by using the pminfo(1) command:
% pminfo -m informix.production.profile.dskreads informix.production.profile.dskreads PMID: 48.0.0 % pminfo -m informix.development.profile.dskreads informix.development.profile.dskreads PMID: 49.0.0 |
On a collection host, the domain numbers currently in use also appear in the Performance Metrics Collection Daemon (PMCD) configuration file /etc/pmcd.conf.
Use the shell script Install in the directory /usr/pcp/pmdas/infmx7 to configure and initiate execution of the agent on the hosts where metrics are to be collected. The script performs the following functions:
Adds the new metric names into the Performance Metrics Name Space (PMNS) in the directory /usr/pcp/pmns.
Creates an ndbm(3B) database of help text for the new metrics in the directory /usr/pcp/pmns.
Installs the executable /usr/pcp/lib/pmdainfmx7.
Installs the relevant pmchart(1) and pmlogger(1) configuration files into the directories /usr/pcp/config/pmchart and /usr/pcp/pmlogger, respectively.
Updates the Performance Metrics Collection Daemon (PMCD) configuration file /etc/pmcd.conf.
Notifies PMCD of the new configuration.
At your discretion, when the Install script has been run, you can run the Verify script to check that the new metrics are now available.
Before running Install, you need to have on hand the server name (INFORMIXSERVER), performance metrics domain number for that server, and the name of the Informix home directory (INFORMIXDIR).
Here is a listing of a Collector configuration run for the server cooked with the Informix home directory set to /usr/informix.
% su # cd /usr/pcp/pmdas/infmx7 # ./Install Are Informix metrics to be collected on this machine (y/n)? [y] Installing in collector configuration Informix server name? [] cooked Performance metrics domain number? [48] Informix home directory? [] /usr/informix Run PMDA as Unix user? [root] NOTE: User 'root' must have access to the Informix SMI tables. Updating the Performance Metrics Name Space (PMNS) ... Updating help files ... Installing executables ... Updating the PMCD control file ... Waiting for PMCD to terminate... Performance Co-Pilot PMCD started (logfile /var/tmp/pmcd.log) Performance Co-Pilot Primary Logger started Verify that informix.cooked metrics have appeared (y/n)? [y] Invoking ./Verify Checking access to host localhost ... ok Informix servers in the name space are: cooked Verifying Informix server cooked on host localhost Waiting for connect to Informix server ... Waiting for connect to Informix server ... Metrics from the sysprofile SMI table ... ok Metrics from the sysshmvals SMI table ... ok Metrics from the syschunks SMI table ... ok Metrics from the syschkio SMI table ... ok Metrics from the syslogs SMI table ... ok Metrics from the syslrus SMI table ... ok Metrics from the syspools SMI table ... ok Metrics from the syssessions SMI table ... ok Metrics from the syssesprof SMI table ... ok Metrics from the sysptprof SMI table ... ok Metrics from the systhreads SMI table ... ok Metrics from the sysvplst SMI table ... ok Metrics from the sysvpprof SMI table ... ok Metrics from the sysseswts SMI table ... ok Metrics from the systhreadwaits SMI table ... ok Verify done. Install done |
Alternately, you may supply all the information required by the Install script as command line arguments:
Install options |
Where the options available are:
| -C | Collector configuration. | |
| -d number | Specifies a PCP domain number. | |
| -H homedir | Specifies an Informix home directory. | |
| -s server | Specifies an Informix server name. | |
| -u username | Run the PMDA as the specified user. |
The configuration task on hosts (typically workstations) used to monitor the performance of Informix servers on remote systems is also accomplished using the Install script in the directory /usr/pcp/pmdas/infmx7. In this case the script performs the following functions:
Adds the new metric names into the Performance Metrics Name Space (PMNS) in the directory /usr/pcp/pmns.
Installs the relevant pmchart(1) and pmlogger(1) configuration files into the directories /usr/pcp/config/pmchart and /usr/pcp/pmlogger, respectively.
At your discretion, when the Install script has been run, you can run the Verify script to check that the new metrics are now available.
Before running Install, you will need to have on hand the server name (INFORMIXSERVER) and performance metrics domain number for that server.
Here is a listing of a Monitor configuration run for the server cooked that has the default domain number 48:
% su # cd /usr/pcp/pmdas/infmx7 # ./Install Are metrics to be collected on this machine (y/n)? [y] n Installing in monitor configuration Informix server name? [] cooked Performance metrics domain number? [48] Updating the Performance Metrics Name Space (PMNS) ... Verify that informix.cooked metrics have appeared (y/n)? [y] Invoking ./Verify Host from which metrics to be collected? [localhost] moomba Verifying Informix server cooked on host moomba Waiting for connect to Informix server ... Waiting for connect to Informix server ... Waiting for connect to Informix server ... Metrics from the sysprofile SMI table ... ok Metrics from the sysshmvals SMI table ... ok Metrics from the syschunks SMI table ... ok Metrics from the syschkio SMI table ... ok Metrics from the syslogs SMI table ... ok Metrics from the syslrus SMI table ... ok Metrics from the syspools SMI table ... ok Metrics from the syssessions SMI table ... ok Metrics from the syssesprof SMI table ... ok Metrics from the sysptprof SMI table ... ok Metrics from the systhreads SMI table ... ok Metrics from the sysvplst SMI table ... ok Metrics from the sysvpprof SMI table ... ok Metrics from the sysseswts SMI table ... ok Metrics from the systhreadwaits SMI table ... ok Verify done. Install done |
Alternately, you may supply all the information required by the Install script as command line arguments:
Install options |
Where the options available are:
| -d number | Specifies a PCP domain number. | |
| -M | Monitor configuration. | |
| -s server | Specifies Informix server name. |
The Verify script in the directory /usr/pcp/pmdas/infmx7 may be run at any time to check the availability of Informix performance metrics from either the local or a remote host.
Here is a listing of a verify run for the server cooked from the remote host moomba:
% cd /usr/pcp/pmdas/infmx7 % ./Verify Host where metrics are to be collected? [localhost] moomba Checking access to host moomba ... ok Informix servers in the name space are: cooked raw Which server do you wish to verify? cooked Verifying Informix server cooked on host moomba Waiting for connect to Informix server ... Waiting for connect to Informix server ... Waiting for connect to Informix server ... Metrics from the sysprofile SMI table ... ok Metrics from the sysshmvals SMI table ... ok Metrics from the syschunks SMI table ... ok Metrics from the syschkio SMI table ... ok Metrics from the syslogs SMI table ... ok Metrics from the syslrus SMI table ... ok Metrics from the syspools SMI table ... ok Metrics from the syssessions SMI table ... ok Metrics from the syssesprof SMI table ... ok Metrics from the sysptprof SMI table ... ok Metrics from the systhreads SMI table ... ok Metrics from the sysvplst SMI table ... ok Metrics from the sysvpprof SMI table ... ok Metrics from the sysseswts SMI table ... ok Metrics from the systhreadwaits SMI table ... ok Verify done. |
Alternately, you may supply the information required by the Verify script as command line arguments:
Verify options |
Where the options available are:
| -h hostname | Metrics are to be collected from the host specified by hostname. | |
| -s server | Specifies an Informix server name. |
To remove the performance metrics for a particular server, use the Remove script in the directory /usr/pcp/pmdas/infmx7. This applies to both Collector and Monitor configurations.
Here is a listing of a remove run for the server cooked:
% cd /usr/pcp/pmdas/infmx7 % ./Remove Informix server name? [cooked] Culling the Performance Metrics Name Space ... Updating the PMCD control file (/etc/pmcd.conf) ... Notifying PMCD ... Removing help files ... - /usr/pcp/help/informix.cooked.dir - /usr/pcp/help/informix.cooked.pag Check that informix.cooked metrics have gone away ... OK Remove done |
Alternately, you may supply the information required by the Remove script as command line arguments:
Remove -s server |
Where server specifies an Informix server name.
This section presents useful hints and problem workarounds available at the time of writing. You should also check the product release notes for important information.
In order to minimize overhead, the infmx7 PMDA does not connect to the Informix server on startup. It is only when the first request for either metric values or an instance domain is received that the connection is established. Connection to the server can take a number of seconds to set up, during which time the error message
Try again. Information not currently available
is returned. You can force a connection attempt by requesting a metric value, using the pminfo(1) command. For example, use the command:
pminfo -f informix.server.profile.dskreads
Normally the PMDA runs with IRIX user-id root. It is required that root have read access to the Informix SMI tables. Otherwise, specify the appropriate user name when you run Install.
Values for metrics with either of the prefixes
informix.server.session.wait informix.server.thread.wait
are available only if the INFORMIX-OnLine Server configuration file sets the WSTATS parameter value to 1.
If there is a problem in the operation of the PMDA, check the PMDA log file: /usr/tmp/infmx7.server.log. The host name of the server replaces server in the given filename.
This Performance Metrics Domain Agent (PMDA) illustrates how to measure application-level response time for an Informix application, and incorporate that response time into the Performance Metrics Collection Systems (PMCS) of the Performance Co-Pilot (PCP). From here on, this is called the infmxping PMDA or agent.
The two metrics provided by the infmxping agent are:
| infmxping.SERVER.response_time |
This metric is the most recently observed response time for the probe transaction. | |
| infmxping.SERVER.time_stamp |
|
In outline, the installation procedure for the infmxping PMDA proceeds as follows:
Software Installation – Use the software installation tool inst(1M) to copy the software from the distribution media. The software is placed in the directory /usr/pcp/pmdas/infmxping.
Collector Configuration – Use the shell script Install in the directory /usr/pcp/pmdas/infmxping to configure and initiate execution of the agent on the hosts where metrics are to be collected.
Monitor Configuration – Use the shell script Install in the directory /usr/pcp/pmdas/infmxping to configure the name space for the agent on the hosts where metrics are to be monitored only.
The agent configuration may subsequently be undone using the shell script Remove and the availability of performance data verified using the shell script Verify. Both of these can be found in the directory /usr/pcp/pmdas/infmxping.
| Note: This PMDA has been tested on only the following releases of the INFORMIX-OnLine Server product. A valid installation of the Performance Co-Pilot product, version 1.1 (PCP 1.1) or later, is a prerequisite. Further, a Collector configuration of PCP 1.1 is required for a Collector configuration of this PMDA. This release is supported: |
INFORMIX 7.1 UD1 for IRIX 5.3.
Before you attempt to configure the PMDA for a particular OnLine server, you need to establish the server name (INFORMIXSERVER), and for a Collector configuration, the Informix home directory (INFORMIXDIR). Further, if you intend to monitor multiple Informix OnLine servers you need to plan the allocation of performance metrics domain numbers for these servers. The following sections cover these issues.
The Install script requires that you supply the server name of the Informix OnLine instance to which the PMDA connects to the probe transaction. This name is typically supplied as the value of the environment variable INFORMIXSERVER when executing database applications or administrative programs. If you do not know what name you should use, contact the database administrator at your site.
The server name is also used to construct the performance metric names for the metrics supported by this PMDA. For example, a site with the servers production and development has the performance metric names
infmxping.production.response_time infmxping.development.response_time |
denoting the response-times to the probe transaction by the production and development servers, respectively.
When you install the infmxping PMDA in a Collector configuration (such as at the host where performance data is being collected), you need to supply the name of the Informix home directory. This name is typically given as the value of the environment variable INFORMIXDIR when executing database applications or administrative programs. If you do not know what name to use, contact the database administrator at your site.
In order to collect ping metrics from multiple Informix OnLine servers, a separate instance of the PMDA is launched for each server. Each PMDA needs to be assigned a unique performance metrics domain number in order to distinguish ping metrics for that particular PMDA from all other metrics being collected and monitored. For a detailed discussion of domain numbers, refer to the Performance Co-Pilot User's and Administrator's Guide.
If you intend to collect and monitor just one server, the default domain number (252) is adequate. For any additional domain numbers, it is recommended that you choose unused numbers starting from 128 and higher. For example, a site with the servers production and development could assign the numbers 252 and 128 for the two servers. For an existing configuration, you can find the domain number by using the pminfo(1) command:
% pminfo -m infmxping.production.time_stamp infmxping.production.time_stamp PMID: 252.0.0 % pminfo -m infmxping.development.time_stamp infmxping.development.time_stamp PMID: 128.0.0 |
On a collection host, the domain numbers currently in use also appear in the performance metrics collection daemon (PMCD) configuration file /etc/pmcd.conf.
Use the shell script Install in the directory /usr/pcp/pmdas/infmxping to configure and initiate execution of the agent on the hosts where metrics are to be collected. The script performs the following functions:
Adds the new metric names into the Performance Metrics Name Space (PMNS) in the directory /usr/pcp/pmns.
Creates an ndbm(3B) database of help text for the new metrics.
Installs the executable /usr/pcp/lib/pmdainfmxping.
Updates the Performance Metrics Collection Daemon (PMCD) configuration file /etc/pmcd.conf.
Notifies PMCD of the new configuration.
At your discretion, when the Install script has been run, you can run the Verify script to check that the new metrics are now available. You are prompted at the end of the Install run to choose whether to run the Verify script or not.
Before running Install you need to have on hand the server name (INFORMIXSERVER), performance metrics domain number for ping metrics from that server, and the name of the Informix home directory (INFORMIXDIR).
Here is a listing of a Collector configuration run for the server cooked with Informix home directory set to /usr/informix.
% su # cd /usr/pcp/pmdas/infmxping # ./Install Are metrics to be collected on this machine (y/n)? [y] Installing in collector configuration Interval (seconds) between probe transactions? [60] Informix server name? [] cooked Performance metrics domain number? [252] Informix home directory? [] /usr/informix Run PMDA as Unix user? [root] NOTE: User `root' requires create database access to Informix. Updating the Performance Metrics Name Space (PMNS) ... Updating help files ... Installing executables ... Updating the PMCD control file ... Waiting for PMCD to terminate... Performance Co-Pilot PMCD started (logfile /var/tmp/pmcd.log) Performance Co-Pilot Primary Logger started Verify that infmxping.cooked metrics have appeared (y/n)? [y] Invoking ./Verify Checking access to host localhost ... ok Informix servers in the name space are: cooked Verifying infmxping metrics server cooked on host localhost Checking metric infmxping.cooked.time_stamp ... ok Checking metric infmxping.cooked.response_time ... ok Verify done. Install done |
Alternately, you may supply all the information required by the Install script as command line arguments:
Install options |
Where the options available are:
| -C | Collector configuration. | |
| -d number | Specifies a PCP domain number. | |
| -H homedir | Specifies an Informix home directory. | |
| -s server | Specifies an interval separating probe transactions. | |
| -t seconds | Specifies an Informix server name. | |
| -u username | Specifies an alternate user name under which to run. |
The configuration task on hosts (typically workstations) used to monitor the performance of Informix servers on remote hosts is also accomplished using the Install script in the directory /usr/pcp/pmdas/infmxping. In this case the script adds the new metric names into the Performance Metrics Name Space (PMNS) in the directory /usr/pcp/pmns.
At your discretion, when the Install script has been run, you can run the Verify script to check that the new metrics are now available.
Before running Install you need to have on hand the server name (INFORMIXSERVER) and the performance metrics domain number for ping metrics from that server.
Here is a listing of a Monitor configuration run for the server cooked, which has the default domain number 252:
% su # cd /usr/pcp/pmdas/infmxping # ./Install Are ping metrics to be collected on this machine (y/n)? [y] n Installing in monitor configuration Informix server name? [] cooked Perrformance metrics domain number? [252] Updating the Performance Metrics Name Space (PMNS) ... Verify that informix.cooked metrics have appeared (y/n)? [y] Invoking ./Verify Host where metrics are to be collected? [localhost] moomba Verifying infmxping metrics for server cooked on host moomba Checking metric infmxping.cooked.time_stamp ... ok Checking metric infmxping.cooked.response_time ... ok Verify done. Install done |
Alternately, you may supply all the information required by the Install script as command line arguments:
Install options |
Where the options available are:
| -d number | Specifies a PCP domain number. | |
| -M | Monitor configuration. | |
| -s server | Specifies an Informix server name. |
The Verify script in the directory /usr/pcp/pmdas/infmxping may be run at any time to check the availability of Informix ping performance metrics from either the local or a remote host.
Here is a listing of a verify run for the server cooked from the remote host moomba:
% cd /usr/pcp/pmdas/infmxping % ./Verify Host where metrics are to be collected? [localhost] moomba Checking access to host moomba ... ok Informix servers in the name space are: cooked raw Which server do you wish to verify? cooked Verifying infmxping metrics for server cooked on host moomba Checking metric infmxping.cooked.time_stamp ... ok Checking metric infmxping.cooked.response_time ... ok Verify done. |
Alternately, you may supply the information required by the Verify script as command line arguments:
Verify options |
Where the options available are:
| -h hostname | Metrics are to be collected from the host specified by hostname. | |
| -s server | Specifies an Informix server name. |
To remove the ping performance metrics for a particular server, use the Remove script in the directory /usr/pcp/pmdas/infmxping. This applies to both Collector and Monitor configurations.
Here is a listing of a Remove run for the server cooked:
% cd /usr/pcp/pmdas/infmxping % ./Remove Informix server name? [cooked] Culling the Performance Metrics Name Space ... Updating the PMCD control file (/etc/pmcd.conf) ... Notifying PMCD ... Removing help files ... - /usr/pcp/help/informix.cooked.dir - /usr/pcp/help/informix.cooked.pag Check that infmxping.cooked metrics have gone away ... ok Remove done |
Alternately, you may supply the information required by the Remove script as command line arguments:
Remove -s server |
Where server specifies an Informix server name.
This section presents useful hints and problem workarounds available at the time of writing. You should also check the product release notes for important information.
If there is a problem in the operation of the PMDA, check the PMDA log file: /usr/tmp/infmxping.server.log (where server is replaced by the name of your server host) for error messages.
Normally this PMDA runs with IRIX user-id root. It is required that root have DBA privileges for the Informix server being probed. Otherwise specify the appropriate user name when you run Install.
You may wish to customize this PMDA for your site. The source files (written in C and ESQL/C) are located in the directory /usr/pcp/pmdas/infmxping. The file /usr/pcp/pmdas/infmxping/ping.ec contains the SQL probe query.
The virtual processor visualizer (vpvis(1)) displays a three-dimensional bar chart of Informix virtual processor (VP) and system central processing unit (CPU) utilization. The display is periodically updated with new values retrieved from a target host or performance metrics archive.
The height of the VP bars is proportional to the Informix virtual processor utilization in the modes sys (in the IRIX kernel) and user (in user mode). The height of each CPU bar is proportional to the system central processor utilization in each of the modes idle (with no I/O pending), wait (idle, but waiting for I/O), intr (interrupt processing), sys (in the IRIX kernel), and user. See Figure 1-5 for an example.
The vpvis visualizer uses the performance metrics visualizer pmview(1), so many of the options and all of the user interface follow the description in the pmview(1) reference page. The pmview(1) tool in turn displays the scene within an IRIS Inventor™ examiner viewer. For generic control of the viewer, see the reference page for ivview(1). You must have IRIS Inventor installed on your monitoring system to successfully use the Performance Co-Pilot. Figure 1-5 shows a vpvis display for a 4 CPU CHALLENGE® server running an 11 VP INFORMIX-OnLine Server.
To see more information about the bars, enter Inventor's Selection Mode (press <Escape>, or click the arrow button from the icon panel in the upper right-hand corner of the window). Use the left mouse button to move the cursor over a bar of interest, and the corresponding performance metric's host name, metric name, instance, value, and units appear in the text box above the scene. Click the left mouse button on a bar, and it is highlighted with a wire frame outline; the text box locks on that bar, showing changes in time for the value of the corresponding performance metric, independent of the cursor position. Unlock the text box by moving the cursor into the base plane of the scene and clicking the left mouse button; the contents of the text box revert to tracking the bar under the cursor.
If the configuration of virtual processors changes on the host being monitored, restart vpvis to observe any additional VPs.
A video recorder (VCR) button icon next to the text box displays the current state of the viewer. Clicking on the button brings up the VCR Controls panel. Refer to pmview(1) for details of the supported VCR Controls.
The vpvis command line synopsis looks like this:
vpvis options |
The options supported by vpvis are:
| -a archive | Use an archive log of performance metrics, rather than a live feed from pmcd(1) on some host. Archive is the base name of an archive, previously created by pmlogger(1). | |
| -h host | Fetch performance metrics from pmcd(1) on host, rather than from the default local host. | |
| -n name space | Load an alternative Performance Metrics Name Space from the file name space. | |
| -s server | Monitor the specified Informix server. | |
| -S numsec | Skip the first numsec seconds of data from the archive log, and start the display from that point in time. -S requires -a and an archive to be specified. | |
| -t delta | The refresh or update interval is set to delta seconds. The default is 1 second. | |
| -Z timezone | Use the specified time zone. |
The standard X11 options and the -v and -z flags are also supported by vpvis; they are described in the reference page for pmview(1).
This section presents useful hints and any problem workarounds available at the time of writing. You should also check the product release notes for important information.
If the configuration of virtual processors changes on the host being monitored, you need to restart vpvis to observe any additional VPs.
[1] Although the server systems don't seem to be monitoring themselves, administration of the Performance Co-Pilot requires that a system's name space contain a name space entry for each Informix server it makes available through PMCD.