[DOCS] Clarify definition of max_size (elastic#56561)

Author: debadair

docs/reference/ilm/actions/ilm-rollover.asciidoc

@@ -43,19 +43,24 @@ PUT my_index-000001
 You must specify at least one rollover condition. 
 An empty rollover action is invalid.
 
-`max_size`::
-(Optional, <<byte-units, byte units>>)
-Triggers roll over after the specified maximum primary shard index storage size is reached.  
+`max_age`::
+(Optional,  <<time-units, time units>>)  
+Triggers roll over after the maximum elapsed time from index creation is reached. 
 
 `max_docs`::
 (Optional, integer)
 Triggers roll over after the specified maximum number of documents is reached.
 Documents added since the last refresh are not included in the document count.
 The document count does *not* include documents in replica shards.
 
-`max_age`::
-(Optional,  <<time-units, time units>>)  
-Triggers roll over after the maximum elapsed time from index creation is reached. 
+`max_size`::
+(Optional, <<byte-units, byte units>>)
+Triggers roll over when the index reaches a certain size.
+This is the total size of all primary shards in the index. 
+Replicas are not counted toward the maximum index size.
++
+TIP: To see the current index size, use the <<cat-indices, _cat indices>> API. 
+The `pri.store.size` value shows the combined size of all primary shards.
 
 [[ilm-rollover-ex]]
 ==== Example

docs/reference/ilm/using-policies-rollover.asciidoc

@@ -4,10 +4,17 @@
 == Configure rollover
 [[using-policies-rollover]]
 You control when the rollover action is triggered by specifying one or more
-rollover parameters. The rollover is performed once any of the criteria are
-met. Because the criteria are checked periodically, the index might grow
-slightly beyond the specified threshold. To control how often the criteria are
-checked, specify the `indices.lifecycle.poll_interval` cluster setting.
+rollover criteria:
+
+* Maximum size (the combined size of all primary shards in the index)
+* Maximum document count
+* Maximum age
+
+The rollover is performed once any of the criteria are met. 
+Because the criteria are checked periodically, the index might grow
+slightly beyond the specified threshold. 
+To control how often the criteria are checked, 
+specify the `indices.lifecycle.poll_interval` cluster setting.
 
 IMPORTANT: New indices created via rollover will not automatically inherit the
 policy used by the old index, and will not use any policy by default. Therefore,
@@ -16,24 +23,6 @@ it is highly recommended to apply the policy via
 setting, for your indices which specifies the policy you wish to use for each
 new index.
 
-The rollover action takes the following parameters:
-
-[[rollover-action-params]]
-.`rollover` Action Parameters
-[options="header"]
-|===
-|Name |Description
-|max_size |The maximum estimated size the primary shard of the index is allowed
-to grow to. Defaults to `null`. Optional.
-|max_docs |The maximum number of document the index should
-contain. Defaults to `null`. Optional.
-|max_age |The maximum age of the index. Defaults to `null`. Optional.
-|===
-
-These parameters are used to determine when the index is considered "full" and
-a rollover should be performed. Where multiple criteria are defined the
-rollover operation will be performed once any of the criteria are met.
-
 The following request defines a policy with a rollover action that triggers
 when the index size reaches 25GB. The old index is subsequently deleted after
 30 days.
@@ -127,7 +116,7 @@ the new index, enabling indexing to continue uninterrupted.
 [[skipping-rollover]]
 === Skipping Rollover
 
-The `index.lifecycle.indexing_complete` setting indicates to {ilm} whether this
+The `index.lifecycle.indexing_complete` setting indicates to {ilm-init} whether this
 index has already been rolled over. If it is set to `true`, that indicates that
 this index has already been rolled over and does not need to be rolled over
 again. Therefore, {ilm} will skip any Rollover Action configured in the

docs/reference/indices/rollover-index.asciidoc

@@ -145,7 +145,12 @@ The document count does *not* include documents in replica shards.
 
 `max_size`::
 (Optional, <<byte-units, byte units>>)
-Maximum estimated size of the primary shard of the index.
+Maximum index size. 
+This is the total size of all primary shards in the index. 
+Replicas are not counted toward the maximum index size.
+
+TIP: To see the current index size, use the <<cat-indices, _cat indices>> API. 
+The `pri.store.size` value shows the combined size of all primary shards. 
 --
 
 include::{docdir}/rest-api/common-parms.asciidoc[tag=mappings]