mirror of
https://github.com/elastic/elasticsearch.git
synced 2025-04-25 07:37:19 -04:00
bfb85bcecb
Currently when the fleet global checkpoints API returns immediately if the index is not ready or shards are not ready. This commit modifies the API to wait on the index and primary shards active up until the timeout period. Related to #71449.
90 lines
3.1 KiB
Text
90 lines
3.1 KiB
Text
[role="xpack"]
|
|
[[get-global-checkpoints]]
|
|
=== Get global checkpoints API
|
|
++++
|
|
<titleabbrev>Get global checkpoints</titleabbrev>
|
|
++++
|
|
|
|
The purpose of the get global checkpoints api is to return the current global
|
|
checkpoints for an index. This API allows users to know the what sequence numbers
|
|
have been safely persisted in Elasticsearch.
|
|
|
|
[discrete]
|
|
[[polling-on-global-checkpoint]]
|
|
== Polling on global checkpoint advance
|
|
|
|
The API has an optional polling mode enabled by the `wait_for_advance` query
|
|
parameter. In polling mode, the API will only return after the global checkpoints
|
|
advance past the provided `checkpoints`. By default, `checkpoints` is an empty
|
|
array, which will lead to the API returning immediately.
|
|
|
|
If a timeout occurs before the global checkpoints advance past the provided
|
|
`checkpoints`, Elasticsearch will return the current global checkpoints and a
|
|
boolean indicating that the request timed out.
|
|
|
|
Currently the `wait_for_advance` parameter is only supported for one shard indices.
|
|
|
|
[discrete]
|
|
[[polling-on-index]]
|
|
== Polling on index ready
|
|
|
|
By default in polling mode, an exception will be returned if the index does not
|
|
exist or all the primary shards are not active. In polling mode, the
|
|
`wait_for_index` parameter can be used to modify this behavior. If `wait_for_index`
|
|
is set to true, the API will wait for the index to be created and all primary
|
|
shards to be active.
|
|
|
|
If a timeout occurs before these conditions are met, the relevant exception will be
|
|
returned.
|
|
|
|
Currently the `wait_for_index` parameter is only supported when `wait_for_advance`
|
|
is true.
|
|
|
|
[[get-global-checkpoints-api-request]]
|
|
==== {api-request-title}
|
|
|
|
`GET /<index>/_fleet/global_checkpoints`
|
|
|
|
[[get-global-checkpoints-api-path-params]]
|
|
==== {api-path-parms-title}
|
|
|
|
`<index>`::
|
|
(Required, string)
|
|
A single index or index alias that resolves to a single index.
|
|
|
|
[role="child_attributes"]
|
|
[[get-global-checkpoints-api-query-parms]]
|
|
==== {api-query-parms-title}
|
|
|
|
`wait_for_advance`::
|
|
(Optional, Boolean) A boolean value which controls whether to wait (until the
|
|
`timeout`) for the global checkpoints to advance past the provided
|
|
`checkpoints`. Defaults to `false`.
|
|
|
|
`wait_for_index`::
|
|
(Optional, Boolean) A boolean value which controls whether to wait (until the
|
|
`timeout`) for the target index to exist and all primary shards be active. Can
|
|
only be `true` when `wait_for_advance` is `true`. Defaults to `false`.
|
|
|
|
`checkpoints`::
|
|
(Optional, list) A comma separated list of previous global checkpoints.
|
|
When used in combination with `wait_for_advance`, the API will only return once
|
|
the global checkpoints advances past the `checkpoints`. Defaults to an empty list
|
|
which will cause Elasticsearch to immediately return the current global
|
|
checkpoints.
|
|
|
|
`timeout`::
|
|
(Optional, <<time-units, time units>>)
|
|
Period to wait for a global checkpoints to advance past `checkpoints`.
|
|
Defaults to `30s`.
|
|
|
|
[role="child_attributes"]
|
|
[[get-global-checkpoints-api-response-body]]
|
|
==== {api-response-body-title}
|
|
|
|
`global_checkpoints`::
|
|
(array of integers) The global checkpoints for the index.
|
|
|
|
`timed_out`::
|
|
(Boolean) If `false` the global checkpoints did not advance past the
|
|
`checkpoints` within the specified `timeout`.
|