Chapter 2. Analyzing Metrics

Kibana offers two ways of analyzing metrics:

Red Hat suggests that you start off by using the predefined visualizations. Each set is known as a dashboard. Dashboards have the advantage of enabling you to quickly access a wide range of metrics while offering the flexibility of changing them to match your individual needs.

2.1. Using Dashboards

A dashboard displays a set of saved visualizations. Dashboards have the advantage of enabling you to quickly access a wide range of metrics while offering the flexibility of changing them to match your individual needs.

You can use the Dashboard tab to create your own dashboards. Alternatively, Red Hat provides the following dashboard examples, which you can import into Kibana and use as is or customize to suit your specific needs:

  • System dashboard
  • Hosts dashboard
  • VMs dashboard

Importing Dashboard Examples

  1. Copy the /etc/ovirt-engine-metrics/dashboards-examples directory from the Manager virtual machine to your local machine.
  2. Open Kibana and click the Settings tab.
  3. Click the Indices tab.
  4. Click the Objects tab.
  5. Click Import and import Searches from your local copy of /etc/ovirt-engine-metrics/dashboards-examples.

  6. Click Import and import Visualizations.

    Note

    If you see an error message while importing the visualizations, check your hosts to ensure that Collectd and Fluentd are running without errors.

  7. Click Import and import Dashboards.
  8. Select project.ovirt-metrics-<ovirt-env-name>.<uuid> in the Index Patterns pane and click the Refresh field list refresh button.

  9. Select the project.ovirt-logs-<ovirt-env-name>.<uuid> index and click Refresh field list.

    The imported dashboards are now stored in the system.

Loading Saved Dashboards

Once you have created and saved a dashboard, or imported Red Hat’s sample dashboards, you can display them in the Dashboard tab:

  1. Click the Dashboard tab.
  2. Click the Load Saved Dashboard loadSavedDashboard button to display a list of saved dashboards.
  3. Click a saved dashboard to load it.

2.2. Creating a New Visualization

Use the Visualize page to design data visualizations based on the metrics or log data collected by Metrics Store. You can save these visualizations, use them individually, or combine visualizations into a dashboard. A visualization can be based on one of the following data source types:

  • A new interactive search
  • A saved search
  • An existing saved visualization

Visualizations are based on Elasticsearch’s aggregation feature.

Creating a New Visualization

Kibana guides you through the creation process with the help of a visualization wizard.

  1. To start the new visualization wizard, click the Visualize tab.
  2. In step 1, Create a new visualization table, select the type of visualization you want to create.

  3. In step 2, Select a search source, select whether you want to create a new search or reuse a saved search:

    • To create a new search, select From a new search and enter the indexes to use as the source. Use project.ovirt-logs prefix for log data or project.ovirt-metrics prefix for metric data.
    • To create a visualization from a saved search, select From a saved search and enter the name of the search.
      The visualization editor appears.

2.3. Graphic User Interface Elements

The visualization editor consists of three main areas:

  • 1 The toolbar
  • 2 The aggregation builder
  • 3 The preview pane

Visualization Editor

visualize

2.4. Using the Visualization Editor

Use the visualization editor to create visualizations by:

2.4.1. Submitting Search Queries

Use the toolbar to perform search queries based on the Lucene query parser syntax. For a detailed explanation of this syntax, see Apache Lucene - Query Parser Syntax.

2.4.2. Selecting Metrics and Aggregations

Use the aggregation builder to define which metrics to display, how to aggregate the data, and how to group the results.

The aggregation builder performs two types of aggregations, metric and bucket, which differ depending on the type of visualization you are creating:

  • Bar, line, or area chart visualizations use metrics for the y-axis and buckets for the x-axis, segment bar colors, and row/column splits.
  • Pie charts use metrics for the slice size and buckets to define the number of slices.

To define a visualization from the aggregation bar:

  1. Select the metric aggregation for your visualization’s y-axis from the Aggregation drop-down list in the metrics section, for example, count, average, sum, min, max, or unique count. For more information about how these aggregations are calculated, see Metrics Aggregation in the Elasticsearch Reference documentation.

  2. Use the buckets area to select the aggregations for the visualization’s x-axis, color slices, and row/column splits:

    1. Use the Aggregation drop-down list to define how to aggregate the bucket. Common bucket aggregations include date histogram, range, terms, filters, and significant terms.

      The order in which you define the buckets determines the order in which they will be executed, so the first aggregation determines the data set for any subsequent aggregations. For more information, see Aggregation Builder in the Kibana documentation.

    2. Select the metric you want to display from the Field drop-down list. For details about each of the available metrics, see Metrics Schema.
    3. Select the required interval from the Interval field.
  3. Click Apply Changes create .

2.5. Metrics Schema

The following sections describe the metrics that are available from the Field menu when creating visualizations.

Note

All metric values are collected at 10 second intervals.

2.5.1. Aggregation Metrics

The Aggregation metric aggregates several values into one using aggregation functions such as sum, average, min, and max. It is used to provide a combined value for average and total CPU statistics.

The following table describes the aggregation metrics reported by the Aggregation plugin.

Metric Namecollectd.type_instanceDescription

collectd.aggregation.percent

  • interrupt
  • user
  • wait
  • nice
  • softirq
  • system
  • idle
  • steal

The average and total CPU usage, as an aggregated percentage, for each of the collectd.type_instance states.

Additional Values

  • collectd.plugin: Aggregation
  • collectd.type_instance: cpu-average / cpu-sum
  • collectd.plugin_instance:
  • collectd.type: percent
  • ovirt.entity: host
  • ovirt.cluster.name.raw: The cluster’s name
  • ovirt.engine_fqdn.raw: The Manager’s FQDN
  • hostname: The host’s FQDN
  • ipaddr4: IP address
  • interval: 10
  • collectd.dstypes: Gauge

2.5.2. CPU Metrics

CPU metrics display the amount of time spent by the hosts' CPUs, as a percentage.

The following table describes CPU metrics as reported by the CPU plugin.

Table 2.1. CPU Metrics

Metric Namecollectd.type_instanceDescription

collectd.cpu.percent

  • interrupt
  • user
  • wait
  • nice
  • softirq
  • system
  • idle
  • steal

The percentage of time spent, per CPU, in the collectd.type_instance states.

Additional Values

  • collectd.plugin: CPU
  • collectd.plugin_instance: The CPU’s number
  • collectd.type: percent
  • ovirt.entity: host
  • ovirt.cluster.name.raw: The cluster’s name
  • ovirt.engine_fqdn.raw: The Manager’s FQDN
  • hostname: The host’s FQDN
  • ipaddr4: IP address
  • interval: 10
  • collectd.dstypes: Gauge

2.5.3. CPU Load Average Metrics

CPU load represents CPU contention, that is, the average number of schedulable processes at any given time. This is reported as an average value for all CPU cores on the host. Each CPU core can only execute one process at a time. Therefore, a CPU load average above 1.0 indicates that the CPUs have more work than they can perform, and the system is overloaded.

CPU load is reported over short term (last one minute), medium term (last five minutes) and long term (last fifteen minutes). While it is normal for a host’s short term load average to exceed 1.0 (for a single CPU), sustained load average above 1.0 on a host may indicate a problem.

On multi-processor systems, the load is relative to the number of processor cores available. The "100% utilization" mark is 1.00 on a single-core, 2.00 on a dual-core, 4.00 on a quad-core system.

Red Hat recommends looking at CPU load in conjunction with CPU Metrics.

The following table describes the CPU load metrics reported by the Load plugin.

Table 2.2. CPU Load Average Metrics

Metric NameDescription

collectd.load.load.longterm

Average number of schedulable processes per CPU core over the last 15 minutes. A value above 1.0 indicates the system was overloaded during the last 15 minutes.

collectd.load.load.midterm

Average number of schedulable processes per CPU core over the last five minutes. A value above 1.0 indicates the system was overloaded during the last 5 minutes.

collectd.load.load.shortterm

Average number of schedulable processes per CPU core over the last one minute. A value above 1.0 indicates the system was overloaded during the last minute.

Additional Values

  • collectd.plugin: Load
  • collectd.type: load
  • collectd.type_instance: None
  • collectd.plugin_instance: None
  • ovirt.entity: host
  • ovirt.cluster.name.raw: The cluster’s name
  • ovirt.engine_fqdn.raw: The Manager’s FQDN
  • hostname: The host’s FQDN
  • ipaddr4: IP address
  • interval: 10
  • collectd.dstypes: Gauge

2.5.4. Disk Consumption Metrics

Disk consumption (DF) metrics enable you to monitor metrics about disk consumption, such as the used, reserved, and free space for each mounted file system.

The following table describes the disk consumption metrics reported by the DF plugin.

Metric NameDescription

collectd.df.df_complex

The amount of free, used, and reserved disk space, in bytes, on this file system.

collectd.df.percent_bytes

The amount of free, used, and reserved disk space, as a percentage of total disk space, on this file system.

Additional Values

  • collectd.plugin: DF
  • collectd.type_instance: free, used, reserved
  • collectd.plugin_instance: A mounted partition
  • ovirt.entity: host
  • ovirt.cluster.name.raw: The cluster’s name
  • ovirt.engine_fqdn.raw: The Manager’s FQDN
  • hostname: The host’s FQDN
  • ipaddr4: IP address
  • interval: 10
  • collectd.dstypes: Gauge

2.5.5. Disk Operation Metrics

Disk operation metrics are reported per physical disk on the host, and per partition.

The following table describes the disk operation metrics reported by the Disk plugin.

Table 2.3. Disk Operation Metrics

Metric NameDescriptioncollectd.dstypes

collectd.disk.disk_ops.read

The number of disk read operations.

Derive

collectd.disk.disk_ops.write

The number of disk write operations.

Derive

collectd.disk.disk_merged.read

The number of disk reads that have been merged into single physical disk access operations. In other words, this metric measures the number of instances in which one physical disk access served multiple disk reads. The higher the number, the better.

Derive

collectd.disk.disk_merged.write

The number of disk writes that were merged into single physical disk access operations. In other words, this metric measures the number of instances in which one physical disk access served multiple write operations. The higher the number, the better.

Derive

collectd.disk.disk_time.read

The average amount of time it took to do a read operation, in milliseconds.

Derive

collectd.disk.disk_time.write

The average amount of time it took to do a write operation, in milliseconds.

Derive

collectd.disk.pending_operations

The queue size of pending I/O operations.

Gauge

collectd.disk.disk_io_time.io_time

The time spent doing I/Os in milliseconds. This can be used as a device load percentage, where a value of 1 second of time spent represents a 100% load.

Derive

collectd.disk.disk_io_time.weighted_io_time

A measure of both I/O completion time and the backlog that may be accumulating.

Derive

Additional Values

  • collectd.plugin: Disk
  • collectd.type_instance: None
  • collectd.plugin_instance: The disk’s name
  • ovirt.entity: host
  • ovirt.cluster.name.raw: The cluster’s name
  • ovirt.engine_fqdn.raw: The Manager’s FQDN
  • hostname: The host’s FQDN
  • ipaddr4: IP address
  • interval: 10

2.5.6. Entropy Metrics

Entropy metrics display the available entropy pool size on the host. Entropy is important for generating random numbers, which are used for encryption, authorization, and similar tasks.

The following table describes the entropy metrics reported by the Entropy plugin.

Table 2.4. Entropy Metrics

Metric NameDescription

collectd.entropy.entropy

The entropy pool size, in bits, on the host.

Additional Values

  • collectd.plugin: Entropy
  • collectd.type_instance: None
  • collectd.plugin_instance: None
  • ovirt.entity: host
  • ovirt.cluster.name.raw: The cluster’s name
  • ovirt.engine_fqdn.raw: The Manager’s FQDN
  • hostname: The host’s FQDN
  • ipaddr4: IP address
  • interval: 10
  • collectd.dstypes: Gauge

2.5.7. Network Interface Metrics

The following types of metrics are reported from physical and virtual network interfaces on the host:

  • Bytes (octets) transmitted and received (total, or per second)
  • Packets transmitted and received (total, or per second)
  • Interface errors (total, or per second)

The following table describes the network interface metrics reported by the Interface plugin.

Table 2.5. Network Interface Metrics

collectd.typeMetric NameDescription

if_octets

collectd.interface.if_octets.rx

A count of the bytes received by the interface. You can view this metric as a Rate/sec or a cumulative count (Max):

* Rate/sec: Provides the current traffic level on the interface in bytes/sec.

* Max: Provides the cumulative count of bytes received. Note that since this metric is a cumulative counter, its value will periodically restart from zero when the maximum possible value of the counter is exceeded.

if_octets

collectd.interface.if_octets.tx

A count of the bytes transmitted by the interface. You can view this metric as a Rate/sec or a cumulative count (Max):

* Rate/sec: Provides the current traffic level on the interface in bytes/sec.

* Max: Provides the cumulative count of bytes transmitted. Note that since this metric is a cumulative counter, its value will periodically restart from zero when the maximum possible value of the counter is exceeded.

if_packets

collectd.interface.if_packets.rx

A count of the packets received by the interface.

You can view this metric as a Rate/sec or a cumulative count (Max):

* Rate/sec: Provides the current traffic level on the interface in bytes/sec.

* Max: Provides the cumulative count of packets received. Note that since this metric is a cumulative counter, its value will periodically restart from zero when the maximum possible value of the counter is exceeded.

if_packets

collectd.interface.if_packets.tx

A count of the packets transmitted by the interface.

You can view this metric as a Rate/sec or a cumulative count (Max):

* Rate/sec: Provides the current traffic level on the interface in packets/sec.

* Max: Provides the cumulative count of packets transmitted. Note that since this metric is a cumulative counter, its value will periodically restart from zero when the maximum possible value of the counter is exceeded.

if_errors

collectd.interface.if_errors.rx

A count of errors received on the interface.

You can view this metric as a Rate/sec or a cumulative count (Max).

* Rate/sec rollup provides the current rate of errors received on the interface in errors/sec.

* Max rollup provides the total number of errors received since the beginning. Note that since this is a cumulative counter, its value will periodically restart from zero when the maximum possible value of the counter is exceeded.

if_errors

collectd.interface.if_errors.tx

A count of errors transmitted on the interface.

You can view this metric as a Rate/sec or a cumulative count (Max).

* Rate/sec rollup provides the current rate of errors transmitted on the interface in errors/sec.

* Max rollup provides the total number of errors transmitted since the beginning. Note that since this is a cumulative counter, its value will periodically restart from zero when the maximum possible value of the counter is exceeded.

if_dropped

collectd.interface.if_dropped.rx

 

if_dropped

collectd.interface.if_dropped.tx

 

Additional Values

  • collectd.plugin: Interface
  • collectd.type_instance: None
  • collectd.plugin_instance: The network’s name
  • ovirt.entity: host
  • ovirt.cluster.name.raw: The cluster’s name
  • ovirt.engine_fqdn.raw: The Manager’s FQDN
  • hostname: The host’s FQDN
  • ipaddr4: IP address
  • interval: 10
  • collectd.dstypes: Derive

2.5.8. Memory Metrics

Metrics collected about memory usage.

The following table describes the memory usage metrics reported by the Memory plugin.

Table 2.6. Memory Metrics

Metric Namecollectd.typecollectd.type_instanceDescription

collectd.memory.memory

memory

used

The total amount of memory used.

free

The total amount of unused memory.

cached

The amount of memory used for caching disk data for reads, memory-mapped files, or tmpfs data.

buffered

The amount of memory used for buffering, mostly for I/O operations.

slab_recl

The amount of reclaimable memory used for slab kernel allocations.

slab_unrecl

Amount of unreclaimable memory used for slab kernel allocations.

collectd.memory.percent

percent

used

The total amount of memory used, as a percentage.

free

The total amount of unused memory, as a percentage.

cached

The amount of memory used for caching disk data for reads, memory-mapped files, or tmpfs data, as a percentage.

buffered

The amount of memory used for buffering I/O operations, as a percentage.

slab_recl

The amount of reclaimable memory used for slab kernel allocations, as a percentage.

slab_unrecl

The amount of unreclaimable memory used for slab kernel allocations, as a percentage.

Additional Values

  • collectd.plugin: Memory
  • collectd.plugin_instance: None
  • ovirt.entity: Host
  • ovirt.cluster.name.raw: The cluster’s name
  • ovirt.engine_fqdn.raw: The Manager’s FQDN
  • hostname: The host’s FQDN
  • ipaddr4: IP address
  • interval: 10
  • collectd.dstypes: Gauge

2.5.9. NFS Metrics

NFS metrics enable you to analyze the use of NFS procedures.

The following table describes the NFS metrics reported by the NFS plugin.

Metric Namecollectd.type_instanceDescription

collectd.nfs.nfs_procedure

null / getattr / lookup / access / readlink / read / write / create / mkdir / symlink / mknod / rename / readdir / remove / link / fsstat / fsinfo / readdirplus / pathconf / rmdir / commit / compound / reserved / access / close / delegpurge / putfh / putpubfh putrootfh / renew / restorefh / savefh / secinfo

/ setattr / setclientid / setcltid_confirm / verify / open / openattr / open_confirm / exchange_id / create_session / destroy_session / bind_conn_to_session / delegreturn / getattr / getfh / lock / lockt / locku / lookupp / open_downgrade / nverify

/ release_lockowner / backchannel_ctl / free_stateid / get_dir_delegation / getdeviceinfo / getdevicelist / layoutcommit / layoutget / layoutreturn / secinfo_no_name / sequence / set_ssv / test_stateid / want_delegation / destroy_clientid / reclaim_complete

The number of processes per collectd.type_instance state.

Additional Values

  • collectd.plugin: NFS
  • collectd.plugin_instance: File system + server or client (for example: v3client)
  • collectd.type: nfs_procedure
  • ovirt.entity: host
  • ovirt.cluster.name.raw: The cluster’s name
  • ovirt.engine_fqdn.raw: The Manager’s FQDN
  • hostname: The host’s FQDN
  • ipaddr4: IP address
  • interval: 10
  • collectd.dstypes: Derive

2.5.10. PostgreSQL Metrics

PostgreSQL data collected by executing SQL statements on a PostgreSQL database.

The following table describes the PostgreSQL metrics reported by the PostgreSQL plugin.

Table 2.7. PostgreSQL Metrics

Metric Namecollectd.type_instanceDescription

collectd.postgresql.pg_numbackends

N/A

How many server processes this database is using.

collectd.postgresql.pg_n_tup_g

live

The number of live rows in the database.

dead

The number of dead rows in the database. Rows that are deleted or obsoleted by an update are not physically removed from their table; they remain present as dead rows until a VACUUM is performed.

collectd.postgresql.pg_n_tup_c

del

The number of delete operations.

upd

The number of update operations.

hot_upd

The number of update operations that have been performed without requiring an index update.

ins

The number of insert operations.

collectd.postgresql.pg_xact

num_deadlocks

The number of deadlocks that have been detected by the database. Deadlocks are caused by two or more competing actions that are unable to finish because each is waiting for the other’s resources to be unlocked.

collectd.postgresql.pg_db_size

N/A

The size of the database on disk, in bytes.

collectd.postgresql.pg_blks

heap_read

How many disk blocks have been read.

heap_hit

How many read operations were served from the buffer in memory, so that a disk read was not necessary. This only includes hits in the PostgreSQL buffer cache, not the operating system’s file system cache.

idx_read

How many disk blocks have been read by index access operations.

idx_hit

How many index access operations have been served from the buffer in memory.

toast_read

How many disk blocks have been read on TOAST tables.

toast_hit

How many TOAST table reads have been served from buffer in memory.

tidx_read

How many disk blocks have been read by index access operations on TOAST tables.

Additional Values

  • collectd.plugin: Postgresql
  • collectd.plugin_instance: Database’s Name
  • ovirt.entity: engine
  • ovirt.cluster.name.raw: The cluster’s name
  • ovirt.engine_fqdn.raw: The Manager’s FQDN
  • hostname: The host’s FQDN
  • ipaddr4: IP address
  • interval: 10
  • collectd.dstypes: Gauge

2.5.11. Process Metrics

The following table describes the process metrics reported by the Processes plugin.

Metric Namecollectd.typecollectd.dstypes

collectd.processes.ps_state

ps_state

Gauge

collectd.processes.ps_disk_ops.read

ps_disk_ops

Derive

collectd.processes.ps_disk_ops.write

ps_disk_ops

Derive

collectd.processes.ps_vm

ps_vm

Gauge

collectd.processes.ps_rss

ps_rss

Gauge

collectd.processes.ps_data

ps_data

Gauge

collectd.processes.ps_code

ps_code

Gauge

collectd.processes.ps_stacksize

ps_stacksize

Gauge

collectd.processes.ps_cputime.syst

ps_cputime

Derive

collectd.processes.ps_cputime.user

ps_cputime

Derive

collectd.processes.ps_count.processes

ps_count

Gauge

collectd.processes.ps_count.threads

ps_count

Gauge

collectd.processes.ps_pagefaults.majfltadd

ps_pagefaults

Derive

collectd.processes.ps_pagefaults.minflt

ps_pagefaults

Derive

collectd.processes.ps_disk_octets.write

ps_disk_octets

Derive

collectd.processes.ps_disk_octets.read

ps_disk_octets

Derive

collectd.processes.fork_rate

fork_rate

Derive

Additional Values

  • collectd.plugin: Processes
  • collectd.plugin_instance: _The process’s name (except for collectd.processes.fork_rate=N/A) collectd.type_instance:* N/A (except for collectd.processes.ps_state=running/ zombies/ stopped/ paging/ blocked/ sleeping)
  • ovirt.entity: host
  • ovirt.cluster.name.raw: The cluster’s name
  • ovirt.engine_fqdn.raw: The Manager’s FQDN
  • hostname: The host’s FQDN
  • ipaddr4: IP address
  • interval: 10

2.5.12. Swap Metrics

Swap metrics enable you to view the amount of memory currently written onto the hard disk, in bytes, according to available, used, and cached swap space.

The following table describes the Swap metrics reported by the Swap plugin.

Table 2.8. Swap Metrics

Metric Namecollectd.typecollectd.type_instancecollectd.dstypesDescription

collectd.swap.swap

swap

used / free / cached

Gauge

The used, available, and cached swap space (in bytes).

collectd.swap.swap_io

swap_io

in / out

Derive

The number of swap pages written and read per second.

collectd.swap.percent

percent

used / free / cached

Gauge

The percentage of used, available, and cached swap space.

Additional Fields

  • collectd.plugin: Swap
  • collectd.plugin_instance: None
  • ovirt.entity: host or Manager
  • ovirt.cluster.name.raw: The cluster’s name
  • ovirt.engine_fqdn.raw: The Manager’s FQDN
  • hostname: The host’s FQDN
  • ipaddr4: IP address
  • interval: 10

2.5.13. Virtual Machine Metrics

The following table describes the virtual machine metrics reported by the Virt plugin.

Metric Namecollectd.typecollectd.type_instancecollectd.dstypes

collectd.virt.ps_cputime.syst

ps_cputime.syst

N/A

Derive

collectd.virt.percent

percent

virt_cpu_total

Gauge

collectd.virt.ps_cputime.user

ps_cputime.user

N/A

Derive

collectd.virt.virt_cpu_total

virt_cpu_total

CPU number

Derive

collectd.virt.virt_vcpu

virt_vcpu

CPU number

Derive

collectd.virt.disk_octets.read

disk_octets.read

disk name

Gauge

collectd.virt.disk_ops.read

disk_ops.read

disk name

Gauge

collectd.virt.disk_octets.write

disk_octets.write

disk name

Gauge

collectd.virt.disk_ops.write

disk_ops.write

disk name

Gauge

collectd.virt.if_octets.rx

if_octets.rx

network name

Derive

collectd.virt.if_dropped.rx

if_dropped.rx

network name

Derive

collectd.virt.if_errors.rx

if_errors.rx

network name

Derive

collectd.virt.if_octets.tx

if_octets.tx

network name

Derive

collectd.virt.if_dropped.tx

if_dropped.tx

network name

Derive

collectd.virt.if_errors.tx

if_errors.tx

network name

Derive

collectd.virt.if_packets.rx

if_packets.rx

network name

Derive

collectd.virt.if_packets.tx

if_packets.tx

network name

Derive

collectd.virt.memory

memory

rss / total /actual_balloon / available / unused / usable / last_update / major_fault / minor_fault / swap_in / swap_out

Gauge

collectd.virt.total_requests

total_requests

flush-DISK

Derive

collectd.virt.total_time_in_ms

total_time_in_ms

flush-DISK

Derive

collectd.virt.total_time_in_ms

total_time_in_ms

flush-DISK

Derive

Additional Values

  • collectd.plugin: virt
  • collectd.plugin_instance: The virtual machine’s name
  • ovirt.entity: vm
  • ovirt.cluster.name.raw: The cluster’s name
  • ovirt.engine_fqdn.raw: The Manager’s FQDN
  • hostname: The host’s FQDN
  • ipaddr4: IP address
  • interval: 10

2.5.14. Gauge and Derive Data Source Types

Each metric includes a collectd.dstypes value that defines the data source’s type:

  • Gauge: A gauge value is simply stored as-is and is used for values that may increase or decrease, such as the amount of memory used.

  • Derive: These data sources assume that the change of the value is interesting, i.e., the derivative. Such data sources are very common for events that can be counted, for example the number of disk read operations. The total number of disk read operations is not interesting, but rather the change since the value was last read. The value is therefore converted to a rate using the following formula: 

    rate = value(new)-value(old)\
        time(new)-time(old)
    Note

    If value(new) is less than value (old), the resulting rate will be negative. If the minimum value to zero, such data points will be discarded.

2.6. Working with Metrics Store Indexes

Metrics Store creates the following two indexes per day:

  • project.ovirt-metrics-<ovirt-env-name>.uuid.yyyy.mm.dd
  • project.ovirt-logs-<ovirt-env-name>.uuid.yyyy.mm.dd

When using the Discover page, select the index named project.ovirt-logs-<ovirt-env-name>.uuid.

In the Visualization page select project.ovirt-metrics-<ovirt-env-name>.uuid for metrics data or project.ovirt-logs-<ovirt-env-name>.uuid for log data.