Apache
/
lucene
mirror of https://github.com/apache/lucene.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Add documentation for searchRate trigger.

This commit is contained in:
Andrzej Bialecki 2018-02-07 14:43:50 -08:00
parent 466c1ef4da
commit d2e8e10233
1 changed files with 51 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

51
solr/solr-ref-guide/src/solrcloud-autoscaling-triggers.adoc
View File

@ -35,6 +35,7 @@ Currently the following event types (and corresponding trigger implementations)
* `nodeAdded` - generated when a new node joins the cluster * `nodeAdded` - generated when a new node joins the cluster
* `nodeLost` - generated when a node leaves the cluster * `nodeLost` - generated when a node leaves the cluster
* `metric` - generated when the configured metric crosses a configured lower or upper threshold value * `metric` - generated when the configured metric crosses a configured lower or upper threshold value
* `searchRate` - generated when the 1 min average search rate exceeds configured upper threshold
Events are not necessarily generated immediately after the corresponding state change occurred - the Events are not necessarily generated immediately after the corresponding state change occurred - the
maximum rate of events is controlled by the `waitFor` configuration parameter (see below). maximum rate of events is controlled by the `waitFor` configuration parameter (see below).
@ -87,6 +88,56 @@ This trigger supports the following configuration:
} }
---- ----
== Search Rate trigger
The search rate trigger can be used for monitoring 1-min average search rates in a selected
collection, and request that either replicas be moved to different nodes or new replicas be added
to reduce the per-replica search rate for a collection / shard with search rate hot spots.
(Note: future versions of Solr will also be able to automatically remove some replicas
when search rate falls below the configured lower threshold).
This trigger support the following configuration:
* `collection` - (string, optional) collection name to monitor, or any collection if empty
* `shard` - (string, optional) shard name within the collection (requires `collection` to be set), or any shard if empty
* `node` - (string, optional) node name to monitor, or any if empty
* `handler` - (string, optional) handler name whose request rate represents the search rate
(default is `/select`). This name is used for creating the full metric key, in
this case `solr.core.<coreName>:QUERY./select.requestTimes:1minRate`
* `rate` - (double, required) the upper bound for the request rate metric value.
If a rate is exceeded for a node (but not for individual replicas placed on this node) then
the action requested by this event is to move one replica (with the highest rate) to another
node. If a rate is exceeded for a collection / shard then the action requested is to add some
replicas - currently at least 1 and at most 3, depending on how much the rate is exceeded, proportional to
the threshold rate and the current request rate.
.Example: a trigger configuration that monitors collection "test" and adds new replicas if 1-min average request rate of "/select" handler exceeds 100 reqs/sec:
[source,json]
----
{
"set-trigger": {
"name" : "search_rate_trigger",
"event" : "searchRate",
"collection" : "test",
"handler" : "/select",
"rate" : 100.0,
"waitFor" : "1m",
"enabled" : true,
"actions" : [
{
"name" : "compute_plan",
"class": "solr.ComputePlanAction"
},
{
"name" : "execute_plan",
"class": "solr.ExecutePlanAction"
}
]
}
}
----
== Trigger Configuration == Trigger Configuration
Trigger configurations are managed using the Autoscaling Write API and the commands `set-trigger`, `remove-trigger`, Trigger configurations are managed using the Autoscaling Write API and the commands `set-trigger`, `remove-trigger`,
`suspend-trigger`, and `resume-trigger`. `suspend-trigger`, and `resume-trigger`.