Skip to content

A Prometheus exporter which uses XenServer's XAPI to derive various metrics.

License

Notifications You must be signed in to change notification settings

solipsist01/xapi_exporter

 
 

Repository files navigation

xapi_exporter

A Prometheus exporter which uses XenServer's XAPI to derive various metrics.

Usage

xapi_exporter [-config <configFile>]

If no configFile is specified, xapi_exporter will attempt to read its configuration from "xapi_exporter.yml" in the working directory.

Config

YML config file should look something like:

bindaddress: ":9290"
namespace: "xenstats"
timeoutlogin: 15

pools:
  "xen-pool-a":
  - "xen-pool-a-host-1"
  - "xen-pool-a-host-2"
  "xen-pool-b":
  - "xen-pool-b-host-1"
  - "xen-pool-b-host-2"
  - "xen-pool-b-host-3"
  - "xen-pool-b-host-4"

auth:
  username: "your_xapi_user"
  password: "your_xapi_password"

Note that you can optionally select which metrics are reported by including an "enabledmetrics" key to an array of strings. Without the "enabledmetrics" option, all metrics are enabled. If one or more metrics are specified with this option, only those metrics will be reported.

enabledmetrics:
- "memory_free"
- "memory_total"

The above would only generate memory_free and memory_total metrics.

Metrics

Below are available metrics.

Pool Metrics

These metrics are relevant to XenServer pools, and will have a "pool" label to distinguish them.

ha_allow_overcommit: If set to false then operations which would cause the Pool to become overcommitted will be blocked.

ha_enabled: true if HA is enabled on the pool, false otherwise

ha_host_failures_to_tolerate: Number of host failures to tolerate before the Pool is declared to be overcommitted

ha_overcommitted: True if the Pool is considered to be overcommitted i.e. if there exist insufficient physicalk resources to tolerate the configured number of host failures

wlb_enabled: true if workload balancing is enabled on the pool

Host Metrics

These metrics are relevant to XenServer hosts, and will have a "host" label to distinguish them.

cpu_pct_allocated: percent vCPUs over total CPUs on this host

cpu_count: the number of physical CPUs on the host

memory_pct_allocated: percent used memory over total memory on this host

memory_free: Free host memory (bytes)

memory_total: Total host memory (bytes)

resident_vcpu_count: count of vCPUs on VMs running on this host

resident_vm_count: count of VMs running on this host

Storage Metrics

These metrics are relevant to XenServer SRs, and will have "pool", "uuid", and "type" labels to distinguish them, as well as a "name_label" label for informational purposes.

default_storage: true if SR is a default SR for a pool, otherwise false

physical_pct_allocated: percent of physical_utilisation over physical_size

physical_size: total physical size of the repository (in bytes)

physical_utilisation: physical space currently utilised on this storage repository (in bytes). Note that for sparse disk formats, physical_utilisation may be less than virtual_allocation.

virtual_allocation: sum of virtual_sizes of all VDIs in this SR (in bytes)

Docker

A script exists to assist in generating a Docker container for xapi_exporter.

./build_container.sh xapi_exporter
  • creates a build image to compile the go application
  • copies the static binary out of the build image
  • deletes the build image
  • creates a run image using the static binary

The dual-image approach allows the final run image to be a tidy 35 MB in size.

The final image expects to be provided with /xapi_exporter/config.yml at run time.

docker run -v /local/xapi_exporter:/xapi_exporter -p 9290:9290 -t xapi_exporter

About

A Prometheus exporter which uses XenServer's XAPI to derive various metrics.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Go 91.9%
  • Shell 6.1%
  • Dockerfile 2.0%