Collect configuration file
The Collect configuration file (Collect.cfg ) is used to define devices, disks, and network interfaces that the collectors do not see by default, or to override various default data collection options. A sample file, Collect.cfg.sample, is located in the following directory:
- On Windows: %BEST1_COLLECT_HOME%\bgs\setup
- On UNIX: $BEST1_HOME/bgs/setup
You can open the file with a text editor, such as Notepad on Windows or Vi on UNIX.
The file is also used by the UNIX system collector on UNIX and Microsoft Windows platforms. The system configuration options are not currently used by the Windows collectors. Therefore, delete the contents between the BEGIN_SYSTEM_COLLECT and END_SYSTEM_COLLECT if you will be using this file for Windows platforms.
To use the Collect configuration file
- Copy the sample file to the following location:
- On Windows: %BEST1_COLLECT_HOME%\local\setup
- On UNIX: $BEST1_HOME/local/setup
- Rename the file from Collect.cfg.sample to Collect.cfg.
- Edit the file to conform with your system's configuration and save.
To filter out specific processes
You can filter out specific processes by using the BEGIN_IGNORE_PROCESS option in the
Collect.cfg file, where you can define one or more command patterns that you do not want collected. The command patterns are only compared against the Command metric.
To filter out specific processes, perform the following task:
- Open collect.cfg.
- Add the following lines to the file:
IGNORE = "ProcessName1"
... ... and so on.
- Save collect.cfg.
- Restart the Agent.
After you restart the agent, you will see that
ProcessName2are not present in the process statistics UDR that was collected.
Options for the Collect.cfg file
The following table lists and describes various flags (options) you can set in different sections of the collect.cfg file.
Use to override the default data collection configuration for disks. Specify TRUE to suppress the collection of statistics (such as name, model, and size) for disks. The default is FALSE.
Use to override the default data collection configuration for controllers. Specify TRUE to suppress the collection of statistics (such as for name, type, and parent) for disk controllers. The default is FALSE.
Use to specify the name of the disk.
Use to specify the disk model.
Use to specify the maximum storage size of the disk.
Use to override the data collection for disks on a Tru64 UNIX 5.0 computer. Specify TRUE to suppress the collection of disk statistics on this platform. The default is FALSE.
Use to override the default data collection configuration for disks on Oracle Solaris computers. Specify TRUE to suppress the collection of the disk model name on this platform. Disk size will be reported as 0. The default is FALSE.
Use to specify the physical name of the hardware circuit that translates disk access instructions from the operating system to the physical disk.
Use to specify the type of disk controller. Common disk controller types are IDE and SCSI.
Use to specify the name of the parent device, if applicable. For example, there could be one parent device (Standard PCI IDE Controller or similar) and two child devices (Primary IDE Controller and Secondary IDE Controller).
Use to override the default data collection for interfaces. Specify TRUE to suppress the collection of the network interface statistics (interface name). The default is FALSE.
Use to specify the physical name of the interface card.
Use to specify the type of interface card.
Use to specify the name of the parent device, if applicable. For example, there could be one parent device and two child devices.
Use to override the default data collection configuration for logical volume statistics on is collected on Tru64 UNIX computers running Linux Security Modules (LSM). Specify TRUE to suppress data collection for logical volumes on Tru64 UNIX computers running Linux Security Modules (LSM). The default is FALSE.
On Solaris and Linux systems, the MINIMUM_TIME option defines the frequency of collecting the optional resident set size (RSS) memory metrics (which can be lower than the frequency of collecting process statistics in general). The default is 60 seconds. In earlier versions, this option collected data according to how long the process had been active, and then sampled every 10 seconds. To enable, set ENABLE to TRUE.
Use to enable the collection of Solaris microstate accounting information. This capability is disabled by default. To enable, set ENABLE to TRUE. SAMPLES is the number of samples the collector will attempt to enable microstate accounting for all processes. The default is 3.
On AIX 5.2 64-bit systems the Perform Collector uses a lot of system time obtaining Process Statistics. This is related to how the Collector obtains the number of open file descriptors. To work around this issue, set the ENABLE flag to FALSE to suppress the collection of open descriptors in the Process Statistics metric group. By default, Perform will continue to collect the number of open file descriptors (ENABLE=TRUE). This option can be used for all platforms.
There is a high processor utilization on Solaris 2.8 64-bit systems when there are a large number of processes running. If you are experiencing significant performance issues, you can suppress the collection of the Scheduling Policy Info metric group by setting the BEGIN_SCHEDULER_POLICY option ENABLE flag to FALSE. By default, this flag is set to TRUE. You will see a decrease in CPU utilization when the ENABLE flag is set to FALSE, resulting in a dramatic difference in performance. This option can be used for all platforms.
Use to override the default data collection configuration for File Systems Statistics. Specify TRUE to enable the data collection of remote file systems. The default is ENABLE=FALSE.
Use to override the default data collection configuration time-out value. The default is 10.
Use to override the default data collection configuration for specific file types (for example, autofs and nfs3 files). Uncomment the section and specify the filesystem type in the TYPE field.
Use to override the data collection of systems with unknown model types. A system model is unknown if it is not defined in the hardware table.
IGNORE_ SUPPLEMENTAL_ GROUPS
Use to override the default data collection configuration for group IDs. Specify FALSE to enable the reporting of multiple group IDs for all users and only report one group per user. The default is TRUE.
On Solaris 8 64-bit computers, you might see invalid CPU statistics. This issue is caused by specific Solaris patches (#108528-28 and #108528-29). To work around this issue, set the COLLECTION_MECHANISM flag to kstat. This option causes the Collector to use the kstat API to collect CPU-related statistics, rather than collecting the statistics directly from the kernel. This option can be used for all version of Solaris.
Specifies the name or the IP address of the HMC that manages the AIX partition.
Specifies the user name that the Perform Collector uses to connect to the HMC. The default is hmcviewer.
Specifies the name of the partition for which data is to be collected.
Specifies the directory where OpenSSH is installed. By default, Gateway Server searches for OpenSSH in the /usr/bin directory. Use this option if you have OpenSSH installed in a different directory in your environment (for example, in /usr/local/bin, or /opt/openssh/bin ).
SYSTEM_COLLECT ARM_COLLECT UMX_COLLECT NT_SYSTEM_ COLLECT
Directs the Collector to generate a log file containing an additional level of logging that provides a record of when metric groups are being collected, how many records were collected, the amount of CPU used to collect the records, and the response time for collection. The default setting for all collectors is TRUE.