Health API
editHealth API
editAn experimental API that returns the health status of an Elasticsearch cluster.
This API is currently experimental for internal use by Elastic software only.
This feature is designed for indirect use by Elasticsearch Service, Elastic Cloud Enterprise, and Elastic Cloud on Kubernetes. Direct use is not supported.
Prerequisites
edit-
If the Elasticsearch security features are enabled, you must have the
monitorormanagecluster privilege to use this API.
Description
editThe health API returns the health status of an Elasticsearch cluster. It returns a list of indicators that compose Elasticsearch functionality.
Each indicator has a health status of: green, unknown, yellow or red. The indicator will
provide an explanation and metadata describing the reason for its current health status.
The cluster’s status is controlled by the worst indicator status.
In the event that an indicator’s status is non-green, a list of impacts may be present in the indicator result which detail the functionalities that are negatively affected by the health issue. Each impact carries with it a severity level, an area of the system that is affected, and a simple description of the impact on the system.
Some health indicators can determine the root cause of a health problem and prescribe a set of
steps that can be performed in order to improve the health of the system. The root cause and remediation
steps are encapsulated in a diagnosis.
A diagnosis contains a cause detailing a root cause analysis, an action containing a brief description
of the steps to take to fix the problem, the list of affected resources (if applicable), and a detailed
step by step troubleshooting guide to fix the diagnosed problem.
Path parameters
edit-
<indicator> -
(Optional, string) Limit the information returned to a specific indicator. Supported indicators are:
-
master_is_stable - Reports health issues regarding the stability of the node that is seen as the master by the node handling the health request. In case of enough observed master changes in a short period of time this indicator will aim to diagnose and report back useful information regarding the cluster formation issues it detects.
-
shards_availability - Reports health issues regarding shard assignments.
-
disk - Reports health issues caused by lack of disk space.
-
ilm - Reports health issues related to Indexing Lifecycle Management.
-
repository_integrity - Tracks repository integrity and reports health issues that arise if repositories become corrupted.
-
slm - Reports health issues related to Snapshot Lifecycle Management.
-
Query parameters
edit-
explain -
(Optional, Boolean) If
true, the response includes additional details that help explain the status of each non-green indicator. These details include additional troubleshooting metrics and sometimes a root cause analysis of a health status. Defaults totrue.
Some health indicators perform root cause analysis of non-green health statuses. This can
be computationally expensive when called frequently. When setting up automated polling of the API
for health status set explain to false to disable the more expensive analysis logic.
Response body
edit-
cluster_name - (string) The name of the cluster.
-
status -
(Optional, string) Health status of the cluster, based on the aggregated status of all indicators in the cluster. If the health of a specific indicator is being requested, this top level status will be omitted. Statuses are:
-
green - The cluster is healthy.
-
unknown - The health of the cluster could not be determined.
-
yellow -
The functionality of a cluster is in a degraded state and may need remediation
to avoid the health becoming
red. -
red - The cluster is experiencing an outage or certain features are unavailable for use.
-
-
indicators -
(object) Information about the health of the cluster indicators.
Properties of
indicators-
<indicator> -
(object) Contains health results for an indicator.
Properties of
<indicator>-
status -
(string) Health status of the indicator. Statuses are:
-
green - The indicator is healthy.
-
unknown - The health of the indicator could not be determined.
-
yellow -
The functionality of an indicator is in a degraded state and may need remediation
to avoid the health becoming
red. -
red - The indicator is experiencing an outage or certain features are unavailable for use.
-
-
symptom - (string) A message providing information about the current health status.
-
details -
(Optional, object) An object that contains additional information about the cluster that
has lead to the current health status result. This data is unstructured, and each
indicator returns a unique set of details. Details will not be calculated if the
explainproperty is set to false. -
impacts -
(Optional, array) If a non-healthy status is returned, indicators may include a list of impacts that this health status will have on the cluster.
Properties of
impacts-
severity - (integer) How important this impact is to the functionality of the cluster. A value of 1 is the highest severity, with larger values indicating lower severity.
-
description - (string) A description of the impact on the cluster.
-
impact_areas -
(array of strings) The areas of cluster functionality that this impact affects. Possible values are:
-
search -
ingest -
backup -
deployment_management
-
-
-
diagnosis -
(Optional, array) If a non-healthy status is returned, indicators may include a list of diagnosis that encapsulate the cause of the health issue and an action to take in order to remediate the problem. The diagnosis will not be calculated if the
explainproperty is false.Properties of
diagnosis-
cause - (string) A description of a root cause of this health problem.
-
action -
(string) A brief description the steps that should be taken to remediate the problem.
A more detailed step by step guide to remediate the problem is provided by the
help_urlfield. -
affected_resources - (Optional, array of strings) If the root cause pertains to multiple resources in the cluster (like indices, shards, nodes, etc…) this will hold all resources that this diagnosis is applicable for.
-
help_url - (string) A link to the troubleshooting guide that’ll fix the healh problem.
-
-
-
Indicator Details
editEach health indicator in the health api returns a set of details that further explains the state of the system. The details have contents and a structure that is unique to each indicator.
master_is_stable
edit-
current_master -
(object) Information about the currently elected master.
Properties of
current_master-
node_id - (string) The node id of the currently elected master, or null if no master is elected.
-
name - (string) The node name of the currently elected master, or null if no master is elected.
-
-
recent_masters -
(Optional, array) A list of nodes that have been elected or replaced as master in a recent time window. This field is present if the master is changing rapidly enough to cause problems, and also present as additional information when the indicator is
green. This array includes only elected masters, and does not include empty entries for periods when there was no elected master.Properties of
recent_masters-
node_id - (string) The node id of a recently active master node.
-
name - (string) The node name of a recently active master node.
-
-
exception_fetching_history -
(Optional, object) If the node being queried sees that the elected master has stepped down repeatedly, the master history is requested from the most recently elected master node for diagnosis purposes. If fetching this remote history fails, the exception information is returned in this detail field.
Properties of
exception_fetching_history-
message - (string) The exception message for the failed history fetch operation.
-
stack_trace - (string) The stack trace for the failed history fetch operation.
-
-
cluster_formation -
(Optional, array) If there has been no elected master node recently, the node being queried attempts to gather information about why the cluster has been unable to form, or why the node being queried has been unable to join the cluster if it has formed. This array could contain any entry for each master eligible node’s view of cluster formation.
Properties of
cluster_formation-
node_id - (string) The node id of a master-eligible node
-
name - (Optional, string) The node name of a master-eligible node
-
cluster_formation_message - (string) A detailed description explaining what went wrong with cluster formation, or why this node was unable to join the cluster if it has formed.
-
shards_availability
edit-
unassigned_primaries - (int) The number of primary shards that are unassigned for reasons other than initialization or relocation.
-
initializing_primaries - (int) The number of primary shards that are initializing or recovering.
-
creating_primaries - (int) The number of primary shards that are unassigned because they have been very recently created.
-
restarting_primaries - (int) The number of primary shards that are relocating because of a node shutdown operation.
-
started_primaries - (int) The number of primary shards that are active and available on the system.
-
unassigned_replicas - (int) The number of replica shards that are unassigned for reasons other than initialization or relocation.
-
initializing_replicas - (int) The number of replica shards that are initializing or recovering.
-
restarting_replicas - (int) The number of replica shards that are relocating because of a node shutdown operation.
-
started_replicas - (int) The number of replica shards that are active and available on the system.
disk
edit-
indices_with_readonly_block -
(int) The number of indices the system enforced a read-only index block (
index.blocks.read_only_allow_delete) on because the cluster is running out of space. -
nodes_with_enough_disk_space - (int) The number of nodes that have enough available disk space to function.
-
nodes_over_high_watermark - (int) The number of nodes that are running low on disk and it is likely that they will run out of space. Their disk usage has tripped the high watermark threshold.
-
nodes_over_flood_stage_watermark - (int) The number of nodes that have run out of disk. Their disk usage has tripped the flood stage watermark threshold.
-
unknown_nodes - (int) The number of nodes for which it was not possible to determine their disk health.
repository_integrity
edit-
total_repositories - (Optional, int) The number of currently configured repositories on the system. If there are no repositories configured then this detail is omitted.
-
corrupted_repositories - (Optional, int) The number of repositories on the system that have been determined to be corrupted. If there are no corrupted repositories detected, this detail is omitted.
-
corrupted - (Optional, array of strings) If corrupted repositories have been detected in the system, the names of up to ten of them are displayed in this field. If no corrupted repositories are found, this detail is omitted.
ilm
edit-
ilm_status -
(string) The current status of the Indexing Lifecycle Management feature. Either
STOPPED,STOPPING, orRUNNING. -
policies - (int) The number of index lifecycle policies that the system is managing.
slm
edit-
slm_status -
(string) The current status of the Snapshot Lifecycle Management feature. Either
STOPPED,STOPPING, orRUNNING. -
policies - (int) The number of snapshot policies that the system is managing.
-
unhealthy_policies -
(map) A detailed view on the policies that are considered unhealthy due to having
several consecutive unssuccesful invocations.
The
countkey represents the number of unhealthy policies (int). Theinvocations_since_last_successkey will report a map where the unhealthy policy name is the key and it’s corresponding number of failed invocations is the value.
Examples
editGET _internal/_health
The API returns a response with all the indicators regardless of current status.
GET _internal/_health/shards_availability
The API returns a response for just the shard availability indicator.
GET _internal/_health?explain=false
The API returns a response with all health indicators but will not calculate details or root cause analysis for the response. This is helpful if you would like to monitor the health API and do not want the overhead of calculating additional troubleshooting details each call.