| Previous | 14.1 Specifics of measurement server types | Next |
|---|
The VMware data collector can read performance indicators from VMware virtual environments and can help the system diagnostic with its collector pages. The data collector can directly connect to an ESXi host or to a vCenter. Currently version 4.0 and later is supported by the collector.
Equipment parameters:
· Host (VMWARE EQ 1 HOST): name / IP address of the ESXi host or the vCenter
· Port (VMWARE EQ 2 PORT): port of the HTTPS VMware service, usually 443
· User name (VMWARE EQ 4 USER): user name used to log in
· Password (VMWARE EQ 5 PWD): either a clear-text password or a value in the format FILE:file_name where file_name points to a file on the PVSR server which contains the password in a clear-text format
· Type (VMWARE EQ 3 TYPE): mandatory parameter, this determine the type of the VMware object: Datacenter, Datastore, ESXi host or Virtual machine
· Item name (VMWARE EQ 6 OBJNAME): name of the monitored VMware object
· Note (VMWARE EQ 7 NOTE): the users can write comments for the virtual machines using the VMware management application. PVSR takes the value of this comment and shows as the value of this parameter. The parameter is viewable to non-administrators as well
· VM folders (VMWARE EQ 8 VMFOLDER): the system automatically fills in the value for those VMs which are placed inside a folder in the vSphere client
· VM datacenter info (X VMWARE EQ 9 DATACENTER): the system automatically fills in the name of the datacenter for every item
· Operating system (Z 01 GUEST_OS): the operating system running on the VM. Non-modifiable automatically discovered parameter for VMs
· VM version (Z 02 VM_VERSION): the VMware guest version. Non-modifiable automatically discovered parameter for VMs
· CPU (Z 03 CPU): CPU information. Non-modifiable automatically discovered parameter for VMs
· Memory (Z 04 MEMORY): memory information. Non-modifiable automatically discovered parameter for VMs
· Memory overhead (Z 05 MEMORY_OVERHEAD): the memory overhead of the VM. Non-modifiable automatically discovered parameter for VMs
· VMware Tools (Z 06 VMWARE_TOOLS): the VMware Tools status on the VM. Non-modifiable automatically discovered parameter for VMs
· IP address (Z 07 IPADDR): the IP address of the VM. Non-modifiable automatically discovered parameter for VMs
· Host name (Z 08 HOSTNAME): the host name of the VM. Non-modifiable automatically discovered parameter for VMs
· Host cluster (Z 09 HOSTCLUSTER)
It is important that PVSR needs an additional Perl library for the HTTPS protocol called Crypt-SSLeay-0.58 which must be installed after the PVSR installation.
PVSR supports the discovery of the virtual environment. All the possible equipments and measurements can be discovered with the use of the simple New equipment or the advanced “New equipments from template” option on the Site and equipment configuration page and the Default VMware equipment template has to be selected. The application logs in with the given credentials, searches for elements to be monitored and displays them. After the OK button is clicked PVSR creates the necessary equipments and measurements. The operation depends on whether we choose to connect to a vCenter or to an ESXi host:
· vCenter: The datastores are created below a site called Datastores. PVSR also creates a site using the name of the vCenter and for each ESXi host it creates an equipment and also a site below it and places the virtual machines under these ESXi sites
· ESXi: it creates the ESXi host and also a site an places the virtual machines below that site
The VMware collector can automatically synchronize the configuration in PVSR with the configuration in the vCenter. The synchronization is configurable with these parameters in the CONFIG_INI.pm file:
· $PVSR_AUTO_VMWARE_TEMPLATE: if it is set to a PVSR equipment template and there is a new virtual machine, host or datastore in the vCenter then PVSR automatically creates it. The default value of the parameter after installation is “Default VMware”
· $PVSR_AUTO_VMWARE_VMNAME_PREFIX: if it is set then PVSR uses its value as the prefix when automatically creating a virtual machine. The “New equipments from template” page uses this parameter as well for the same purpose. The default value of the parameter after installation is “VM ”
· $PVSR_AUTO_VMWARE_VMNAME_POSTFIX: if it is set then PVSR uses its value as the postfix when automatically creating a virtual machine. The “New equipments from template” page uses this parameter as well for the same purpose. The default value of the parameter after installation is empty
· $PVSR_AUTO_VMWARE_MOVE_DELETED_SITE: if it is set then PVSR moves the non-existing virtual machines, hosts and datastores into this site (next to the datacenter sites). If the site does not exist then PVSR automatically creates it. The default value of the parameter after installation is “Deleted”
· $PVSR_AUTO_VMWARE_MANAGE_DATA_COLL: if it is set to “Y” then PVSR automatically deactivates the data collection of a virtual machine if it is not running. The default value of the parameter after installation is empty
· $PVSR_AUTO_VMWARE_MOVE_OFFLINE_VM_SITE: if it is set then PVSR moves the shut-down virtual machines into this site (next to the datacenter sites). If the site does not exist then PVSR automatically creates it. The default value of the parameter after installation is empty
· $PVSR_AUTO_VMWARE_SKIP_ENTRY: if it is set then the synchronizer concatenates the type (HostSystem, VirtualMachine, Datastore) and the name of every entry seen in the vCenter and if for an entry the regular expression in the parameter matches this concatenation then it will not be created. For example if the value of the parameter is “VirtualMachine:NotNeeded” then no vm with the name beginning NotNeeded will be created. Other example: if the value is “Datastore:” then no datastore will be created
PVSR offers three collector pages for VMware equipments: Virtual Machines, Hosts and Datastores.
This page shows the virtual machines in the VMware environment. If the page is opened from an ESXi host then if automatically filters to show only the virtual machines on that host. If the page is opened from a virtual machines then it filters out the other virtual machines.
· Name: name of the virtual machine
· Status: the current status of the virtual machine: stopped, running, suspended or template
· Host name: name of the physical host running the virtual machines
· vCPU: number of vCPUs configured for the virtual machines
· CPU usage: the CPU usage of the virtual machine in MHz and in percentage where 100% represents the maximum allowable CPU usage configured for the virtual machine
· Memory usage: the memory usage of the virtual machine in MB and in percentage where 100% represents the maximum allowable memory usage configured for the virtual machine
· Storage: the storage configured for the virtual machine and the really used storage (thin provisioning). This is only supported since VMware 4.1
· Other: If the VMware Tools is configured then it shows the IP address of the virtual machine by default. After the cell is expanded other informations are also available:
o VMware Tools: Not installed, Not running, Ok or Old version
o IP address: If the VMware Tools is configured then it shows the IP address of the virtual machine
o Computer name: If the VMware Tools is configured then it shows the host name of the virtual machine
o Remark: the comment set for the virtual machine
The page shows the physical hosts in the VMware environment. The parameters for each host:
· Name
· Status: current status: OK, Might have a problem, Definitely has a problem, Unknown
· CPU cores: the total number of cores in the server
· Memory: the total MB of memory in the server
· VMs: how many virtual machines are configured on that host
· CPU usage: the total CPU usage in MHz and in percentage
· Memory usage: the total memory usage in MB and in percentage
The list of datastores. This page only works for VMware 4.1 or later and shows the following parameters:
· Name: name of the datastore
· Accessible: whether it is available or not
· Type: VMFS, NFS or CIFS
· Capacity: the usable storage space in GB
· Freespace: the free storage space in GB
· Uncommited: the storage space not yet commited (thin provisioning)
· VMs: the number of virtual machines on the datastore
· VM names: the names of these virtual machines
When opened from a server the page shows all the servers in the virtual environment, when opened from a virtual machine then it shows all the virtual machines. The user can select a measurement type and specify the selected timespan with the widgets in the Time span browser area group. PVSR calculates the averages of the collected data based on these input parameters and colors the different items based on their value. The blue end of the color spectrum has the minimum value and the red end of the color spectrum has the maximum value. If an item does not have a color then it means that it does not have any collected data for the specified time span. The page also shows the actual time span for which there is collected data in the system.
PVSR is capable of collecting any parameter provided by VMware, although not every one is defined as a measurement after the installation. In the VMware environment a parameter is either real-time (one value every 20 seconds) or historical (5 minutes, 30 minutes, 2 hours or 24 hours resolution). In each case PVSR gets the data with the least resolution and if necessary it calculates the average to convert the data resolution into PVSR measurement cycle.
Every variable has an identifier which consists of the variable group name and the variable name. The formatting of a variable name is the following:
· The prefix M_ indicating that this is a performance indicator
· The name of the group, for example cpu or memory
· The _ character
· The name of the variable, for example usagemhz or system. If the name contains a . character that it must be replaced with the _ character, for example capacity_usage instead of capacity.usage
· If the variable belongs to an object type which can have multiple instances then either the .PORT PVSR specific postfix must end the variable name or an instance specific .inst where inst refers to the instance identifier. For example some of the object types have a global cumulative instance identified by _Total
Examples:
· Active memory: it does not have multiple instances, it is in the group mem and its name is active, so in PVSR it can be referred to as M_mem_active
· Used processor MHz: it has multiple instances, it is in the group cpu and its name is usagemhz
o As a multiple-instance measurement: M_cpu_usagemhz.PORT
o Total usage using the _Total instance: M_cpu_usagemhz._Total
· Some of the measurements are only calculated by Vmware with a specific interval, for example every 30 minutes. In this case the MI[interval_in_sec]_[group]_[parameter] format must be used, for example MI1800_disk_used for disk usage measurements
If the unit of a parameter is percentage then VMware multiplies the value with 100 and gives it back as an integer, so the received value must be divided by 100, for example:
FLOAT_OUT=#M_mem_usage#/100;
For other possible performance indicators please refer to the Managed Object - PerformanceManager chapter in the VMware documentation.