Introduce repository integrity verification API

docs/changelog/112348.yaml

@@ -0,0 +1,6 @@
+pr: 112348
+summary: Introduce repository integrity verification API
+area: Snapshot/Restore
+type: enhancement
+issues:
+ - 52622

docs/reference/snapshot-restore/apis/snapshot-restore-apis.asciidoc

@@ -28,6 +28,7 @@ For more information, see <<snapshot-restore>>.
 include::put-repo-api.asciidoc[]
 include::verify-repo-api.asciidoc[]
 include::repo-analysis-api.asciidoc[]
+include::verify-repo-integrity-api.asciidoc[]
 include::get-repo-api.asciidoc[]
 include::delete-repo-api.asciidoc[]
 include::clean-up-repo-api.asciidoc[]

docs/reference/snapshot-restore/apis/verify-repo-integrity-api.asciidoc

@@ -0,0 +1,185 @@
+[role="xpack"]
+[[verify-repo-integrity-api]]
+=== Verify repository integrity API
+++++
+<titleabbrev>Verify repository integrity</titleabbrev>
+++++
+
+Verifies the integrity of the contents of a snapshot repository.
+
+////
+[source,console]
+----
+PUT /_snapshot/my_repository
+{
+  "type": "fs",
+  "settings": {
+    "location": "my_backup_location"
+  }
+}
+----
+// TESTSETUP
+////
+
+[source,console]
+----
+POST /_snapshot/my_repository/_verify_integrity
+----
+
+[[verify-repo-integrity-api-request]]
+==== {api-request-title}
+
+`POST /_snapshot/<repository>/_verify_integrity`
+
+[[verify-repo-integrity-api-prereqs]]
+==== {api-prereq-title}
+
+* If the {es} {security-features} are enabled, you must have the `manage`
+<<privileges-list-cluster,cluster privilege>> to use this API. For more
+information, see <<security-privileges>>.
+
+[[verify-repo-integrity-api-desc]]
+==== {api-description-title}
+
+This API allows you to perform a comprehensive check of the contents of a
+repository, looking for any anomalies in its data or metadata which might
+prevent you from restoring snapshots from the repository or which might cause
+future snapshot create or delete operations to fail.
+
+The default values for the parameters of this API are designed to limit the
+impact of the integrity verification on other activities in your cluster. For
+instance, by default it will only use at most half of the `snapshot_meta`
+threads to verify the integrity of each snapshot, allowing other snapshot
+operations to use the other half of this thread pool.
+
+You should avoid any operations which write to the repository while this API is
+running. If something changes the repository contents while an integrity
+verification is running then {es} may incorrectly report having detected some
+anomalies in its contents due to the concurrent writes. It may also incorrectly
+fail to report some anomalies that the concurrent writes prevented it from
+detecting.
+
+NOTE: This API is intended for exploratory use by humans. You should expect the
+request parameters and the response format to vary in future versions.
+
+NOTE: This API may not work correctly in a mixed-version cluster.
+
+[[verify-repo-integrity-api-path-params]]
+==== {api-path-parms-title}
+
+`<repository>`::
+(Required, string)
+Name of the snapshot repository whose integrity to verify.
+
+[[verify-repo-integrity-api-query-params]]
+==== {api-query-parms-title}
+
+`snapshot_verification_concurrency`::
+(Optional, integer) Specifies the number of snapshots to verify concurrently.
+Defaults to `0` which means to use at most half of the `snapshot_meta` thread
+pool at once.
+
+`index_verification_concurrency`::
+(Optional, integer) Specifies the number of indices to verify concurrently.
+Defaults to `0` which means to use the entire `snapshot_meta` thread pool.
+
+`meta_thread_pool_concurrency`::
+(Optional, integer) Specifies the maximum number of snapshot metadata
+operations to execute concurrently. Defaults to `0` which means to use at most
+half of the `snapshot_meta` thread pool at once.
+
+`index_snapshot_verification_concurrency`::
+(Optional, integer) Specifies the maximum number of index snapshots to verify
+concurrently within each index verification. Defaults to `1`.
+
+`max_failed_shard_snapshots`::
+(Optional, integer) Limits the number of shard snapshot failures to track
+during integrity verification, in order to avoid excessive resource usage. If
+your repository contains more than this number of shard snapshot failures then
+the verification will fail. Defaults to `10000`.
+
+`verify_blob_contents`::
+(Optional, boolean) Specifies whether to verify the checksum of every data blob
+in the repository. Defaults to `false`. If this feature is enabled, {es} will
+read the entire repository contents, which may be extremely slow and expensive.
+
+`blob_thread_pool_concurrency`::
+(Optional, integer) If `?verify_blob_contents` is `true`, this parameter
+specifies how many blobs to verify at once. Defaults to `1`.
+
+`max_bytes_per_sec`::
+(Optional, <<size-units, size units>>)
+If `?verify_blob_contents` is `true`, this parameter specifies the maximum
+amount of data that {es} will read from the repository every second. Defaults
+to `10mb`.
+
+[role="child_attributes"]
+[[verify-repo-integrity-api-response-body]]
+==== {api-response-body-title}
+
+The response exposes implementation details of the analysis which may change
+from version to version. The response body format is therefore not considered
+stable and may be different in newer versions.
+
+`log`::
+(array) A sequence of objects that report the progress of the analysis.
++
+.Properties of `log`
+[%collapsible%open]
+====
+`snapshot`::
+(object) If the log entry pertains to a particular snapshot then the snapshot
+will be described in this object.
+
+`index`::
+(object) If the log entry pertains to a particular index then the index will be
+described in this object.
+
+`index`::
+(object) If the log entry pertains to a particular index then the index will be
+described in this object.
+
+`snapshot_restorability`::
+(object) If the log entry pertains to the restorability of an index then the
+details will be described in this object.
+
+`anomaly`::
+(string) If the log entry pertains to an anomaly in the repository contents then
+this string will describe the anomaly.
+
+`exception`::
+(object) If the log entry pertains to an exception that {es} encountered during
+the verification then the details will be included in this object.
+
+====
+
+`results`::
+(object) An object which describes the final results of the analysis.
++
+.Properties of `results`
+[%collapsible%open]
+====
+`status`::
+(object) The final status of the analysis task.
+
+`final_repository_generation`::
+(integer) The repository generation at the end of the analysis. If there were
+any writes to the repository during the analysis then this value will be
+different from the `generation` reported in the task status, and the analysis
+may have detected spurious anomalies due to the concurrent writes, or may even
+have failed to detect some anomalies in the repository contents.
+
+`total_anomalies`::
+(integer) The total number of anomalies detected during the analysis.
+
+`result`::
+(string) The final result of the analysis. If the repository contents appear to
+be intact then this will be the string `pass`. If this field is missing, or
+contains some other value, then the repository contents were not fully
+verified.
+
+====
+
+`exception`::
+(object) If the analysis encountered an exception which prevented it from
+completing successfully then this exception will be reported here.

docs/reference/snapshot-restore/register-repository.asciidoc

@@ -272,7 +272,9 @@ filesystem snapshot of this repository.
 When restoring a repository from a backup, you must not register the repository
 with {es} until the repository contents are fully restored. If you alter the
 contents of a repository while it is registered with {es} then the repository
-may become unreadable or may silently lose some of its contents.
+may become unreadable or may silently lose some of its contents. After
+restoring a repository from a backup, use the <<verify-repo-integrity-api>> API
+to verify its integrity before you start to use the repository.
 
 include::repository-azure.asciidoc[]
 include::repository-gcs.asciidoc[]

server/src/main/java/org/elasticsearch/repositories/RepositoryData.java

@@ -281,6 +281,13 @@ public Collection<SnapshotId> getSnapshotIds() {
         return snapshotIds.values();
     }
 
+    /**
+     * @return the number of index snapshots (i.e. the sum of the index count of each snapshot)
+     */
+    public long getIndexSnapshotCount() {
+        return indexSnapshots.values().stream().mapToLong(List::size).sum();
+    }
+
     /**
      * @return whether some of the {@link SnapshotDetails} of the given snapshot are missing, due to BwC, so that they must be loaded from
      * the {@link SnapshotInfo} blob instead.

x-pack/plugin/security/qa/operator-privileges-tests/src/javaRestTest/java/org/elasticsearch/xpack/security/operator/Constants.java

@@ -98,6 +98,7 @@ public class Constants {
         "cluster:admin/snapshot/restore",
         "cluster:admin/snapshot/status",
         "cluster:admin/snapshot/status[nodes]",
+        "cluster:admin/repository/verify_integrity",
         "cluster:admin/features/get",
         "cluster:admin/features/reset",
         "cluster:admin/tasks/cancel",