high level REST api: cancel task #30745

pcsanwald · 2018-05-20T14:38:16Z

The goal of this PR is to expose the cancel task functionality from the high level REST client, as part of #27205.

Still a few things to do on this, but wanted to put it up early in case I've gone off the rails with any of this. There are three things in particular I am unsure about:

1 - I'm unhappy with my integration testCase in ClusterClientIT, some options to make it better include using a similar technique to TasksIT.testTasksCancellation (which feels like duplication but an ok approach), or alternatively, passing something through that I can write better assertions against somehow. would love thoughts here.

2 - The response for cancelTask is the same as listTasks, for the documentation, should I copy paste and change the verb, or just say "the response is the same"?

3 - In terms of testing this locally against a running server, it seems like a valuable thing to do, but does need a task that will block. Is there a suitable task or do I need to do it via plugin somehow?

elasticmachine · 2018-05-20T14:38:18Z

Pinging @elastic/es-core-infra

nik9000 · 2018-05-21T15:17:32Z

client/rest-high-level/src/test/java/org/elasticsearch/client/ClusterClientIT.java

+
+        CancelTasksRequest request = new CancelTasksRequest();
+        request.setTaskId(new TaskId(node, firstTask.getId()));
+        request.setReason("persistent action was removed");


I'd probably do something like "testreason" so no one gets confused.

nik9000 · 2018-05-21T15:20:30Z

client/rest-high-level/src/test/java/org/elasticsearch/client/ClusterClientIT.java

+            highLevelClient().cluster()::cancelTasksAsync
+        );
+
+        assertThat(response, notNullValue());


It'd certainly be nice to have more here but you can't really be sure that you are going to actually cancel the task or it is going to finish. Since we already have an integration test for the act of cancelling such a task I'd probably be satisfied with something fairly basic here and good unit testing on the parsing side of the responses. We know the cancel works because we have TaskIT. Reindex even has some tests to make sure that it cancels itself in a sane way. So I think we're ok if we make sure we can parse here.

I do think it is worth leaving a comment about how this might cancel a task and it might not, depending on how quickly the task executes.

nik9000 · 2018-05-21T15:31:13Z

.../main/java/org/elasticsearch/action/admin/cluster/node/tasks/cancel/CancelTasksResponse.java

 /**
 * Returns the list of tasks that were cancelled
 */
 public class CancelTasksResponse extends ListTasksResponse {

+    private static final ConstructingObjectParser<CancelTasksResponse, Void> PARSER =


Do you think you can make something like:

protected static <T> ConstructingObjectParser<T, Void> setupParser(String name, TriFunction<List<TaskInfo>, List<TaskOperationFailure>, List<ElasticsearchException>> ctor)

in ListTasksResponse? That way you wouldn't need to copy this stuff.

great idea! will do

@nik9000 I went to do this and ran into a problem: setupParser would maybe need to instantiate a response: return new ListTasksResponse(tasks, tasksFailures, nodeFailures);, someplace, were you thinking that the calling method would supply a lambda to do this?

because otherwise, unless I'm missing something, couldn't call new on a generic type. Am I missing something?

I'd supply a lambda, yeah.

@nik9000 nice suggestion, this is implemented now

nik9000 · 2018-05-21T15:32:08Z

.../src/main/java/org/elasticsearch/action/admin/cluster/node/tasks/list/ListTasksResponse.java

@@ -50,9 +50,9 @@
 * Returns the list of tasks currently running on the nodes
 */
 public class ListTasksResponse extends BaseTasksResponse implements ToXContentObject {
-    private static final String TASKS = "tasks";
-    private static final String TASK_FAILURES = "task_failures";
-    private static final String NODE_FAILURES = "node_failures";


If you made that method I mentioned about then these could stay private which is kind of nice.

nik9000 · 2018-05-21T15:43:50Z

2 - The response for cancelTask is the same as listTasks, for the documentation, should I copy paste and change the verb, or just say "the response is the same"?

I'd say the response is the same and leave a link.

3 - In terms of testing this locally against a running server, it seems like a valuable thing to do, but does need a task that will block. Is there a suitable task or do I need to do it via plugin somehow?

Reindex can be made to take a long time if you index a bunch of documents. It is my go-to for manual testing for things like this.

javanna · 2018-05-22T11:49:28Z

docs/java-rest/high-level/cluster/cancel_tasks.asciidoc

+
+==== Cancel Tasks Response
+
+TBD


I'd probably paste what we have in the other page, for simplicity.

javanna · 2018-05-22T11:50:34Z

server/src/test/java/org/elasticsearch/tasks/CancelTaskResponseTests.java

+
+    @Override
+    protected boolean supportsUnknownFields() {
+        return false;


this should be set to true so we test forward compatibility. Did you set it to false because you encountered problems with this?

@javanna I did have problems with this value set to true, here's a gist with the error I saw.

I used the ListTask stuff as kind of a template for my changes, and I noticed we also set this value to false in ListTasksResponseTest.java, so I figured I'd need to do the same thing here. Is there a problem in both places? I'm not sure I fully understand the reason behind the failure, actually. I'll have a closer look at why this is happening, if the setting seems incorrect (there's a fair amount of me looking at other examples in the codebase for things on this PR)

the gist looks about the xcontent equivalence checks more than unknown fields. That is kind of expected whenever the response holds an exception, as we parse any exception back into a generic ElasticsearchException, which has a different toXContent output compared to the original exception.

Did a bit more debugging here: The problem, which also exists in ListTasksResponseTests, is that the headers field isn't being parsed correctly. I initially thought that this was because we were not overriding getRandomFieldsExcludeFilter (which TaskInfoTest does do) on either ListTasksResponseTests or CancelTasksResponseTests, but, I still am getting a parse error on the headers field.

The error seems to happen most often when the headers field is an empty map (which obv happens in a subset of the cases because it's random). Any thoughts on where the problem might be? I'm still not familiar enough with this code to triangulate, although I'm getting there :).

I had a look and reworked a bit tests for list tasks here: #31001 . That should clarify what should and should not work in these tests, your cancel response tests should end up looking very similar. headers and status need to be both excluded, then we can enable the insertion of random fields by setting supportsUnknownFields to true. Also, xcontent equivalence works only when there are no failures, hence I split the two scenarios (with and without failures).

That doesn't really answer your question on the empty map problem that you saw, I don't quite follow on the empty map issue but I think it may have to do with the exclusion predicate. I have return field -> field.endsWith("status") || field.endsWith("headers");. Note the endsWith here compared to the equals in TaskInfoTests, that is because such fields are at the top-level for TaskInfo but deeper in the ListTasksResponse or CancelTasksResponse output.

javanna · 2018-05-22T11:51:40Z

server/src/test/java/org/elasticsearch/tasks/CancelTaskResponseTests.java

+
+    @Override
+    protected boolean assertToXContentEquivalence() {
+        return false;


this is due to parsing exceptions back I guess?

This is another case of me doing the same thing as ListTasksResponseTest (perhaps incorrectly): here's a gist with the error when the value is set to true. I'm happy to dig deeper into this error if the setting seems incorrect.

this should be ok!

javanna · 2018-05-28T13:04:54Z

client/rest-high-level/src/main/java/org/elasticsearch/client/ClusterClient.java

+     * <a href="https://www.elastic.co/guide/en/elasticsearch/reference/current/tasks.html"> Task Management API on elastic.co</a>
+     * </p>
+     */
+    public CancelTasksResponse cancelTasks(CancelTasksRequest cancelTasksRequest, Header... headers) throws IOException {


I just realized that these API should belong to the tasks namespace rather than cluster, according to our API spec, see #30906 .

…k namespace

pcsanwald · 2018-05-30T03:50:08Z

@javanna @nik9000 I've still got a bit more work to do to address your comments, wanted to get namespace changes done first. will drop a note (maybe tomorrow) when this is ready for re-review.

This commit reworks testing for `ListTasksResponse` so that random fields insertion can be tested and xcontent equivalence can be checked too. Proper exclusions need to be configured, and failures need to be tested separately. This helped finding a little problem, whenever there is a node failure returned, the nodeId was lost as it was never printed out as part of the exception toXContent.

javanna · 2018-05-31T14:01:31Z

server/src/test/java/org/elasticsearch/tasks/CancelTaskResponseTests.java

+import static org.hamcrest.Matchers.equalTo;
+import static org.hamcrest.Matchers.notNullValue;
+
+public class CancelTaskResponseTests  extends AbstractXContentTestCase<CancelTasksResponse> {


nit: would you mind calling it CancelTasksResponseTests to align it with the class that we are testing? and remove one of the whitespaces between the class name and extends? :)

…rom non-failure cases

javanna · 2018-06-05T09:27:26Z

client/rest-high-level/src/main/java/org/elasticsearch/client/ClusterClient.java

+    public void deletePipelineAsync(DeletePipelineRequest request, ActionListener<WritePipelineResponse> listener, Header... headers) {
+        restHighLevelClient.performRequestAsyncAndParseEntity( request, RequestConverters::deletePipeline,
+            WritePipelineResponse::fromXContent, listener, emptySet(), headers);
+    }


These methods should be gone now, they have been moved to IngestClient upstream

javanna · 2018-06-05T09:28:54Z

client/rest-high-level/src/main/java/org/elasticsearch/client/TasksClient.java

+     * <a href="https://www.elastic.co/guide/en/elasticsearch/reference/current/tasks.html"> Task Management API on elastic.co</a>
+     * </p>
+     */
+    public void cancelTasksAsync(CancelTasksRequest cancelTasksRequest, ActionListener<CancelTasksResponse> listener, Header... headers) {


now that #30490 is in, could you replace the header argument with the RequestOptions one? In the method that accept a listener we add RequestOptions between the request and the listener though.

javanna · 2018-06-05T09:29:36Z

client/rest-high-level/src/test/java/org/elasticsearch/client/ClusterClientIT.java

+        WritePipelineResponse response =
+            execute(request, highLevelClient().cluster()::deletePipeline, highLevelClient().cluster()::deletePipelineAsync);
+        assertTrue(response.isAcknowledged());
+    }


the changes should not be here, these tests have been moved to IngestClientIT upstream

javanna · 2018-06-05T09:32:20Z

client/rest-high-level/src/main/java/org/elasticsearch/client/RequestConverters.java

+            cancelTasksRequest.getTimeout())
+            .withTaskId(cancelTasksRequest.getTaskId())
+            .withNodes(cancelTasksRequest.getNodes())
+            .withActions(cancelTasksRequest.getActions());


the REST action seems to also support parent_task_id, the parameter is also declared in the REST spec so it should be here too?

javanna · 2018-06-05T09:33:23Z

client/rest-high-level/src/main/java/org/elasticsearch/client/TasksClient.java

+     * <a href="https://www.elastic.co/guide/en/elasticsearch/reference/current/tasks.html"> Task Management API on elastic.co</a>
+     * </p>
+     */
+    public CancelTasksResponse cancelTasks(CancelTasksRequest cancelTasksRequest, Header... headers) throws IOException {


can we call the methods just cancel given that they now belong to the TasksClient?

javanna · 2018-06-05T09:33:58Z

client/rest-high-level/src/test/java/org/elasticsearch/client/RequestConvertersTests.java

+        CancelTasksRequest request = new CancelTasksRequest();
+        Map<String, String> expectedParams = new HashMap<>();
+        TaskId taskId = new TaskId(randomAlphaOfLength(5), randomNonNegativeLong());
+        request.setTaskId(taskId);


shall we also test the other supported parameters?

javanna · 2018-06-05T09:36:25Z

server/src/test/java/org/elasticsearch/tasks/CancelTasksResponseTests.java

+
+    @Override
+    protected boolean assertToXContentEquivalence() {
+        return false;


in theory this could be left as default (true) given that failures are tested separately?

javanna

I left some more comments but this is very close.

pcsanwald · 2018-06-06T02:17:58Z

...nt/rest-high-level/src/test/java/org/elasticsearch/client/ESRestHighLevelClientTestCase.java

@@ -70,6 +70,17 @@ protected static RestHighLevelClient highLevelClient() {
        }
    }

+    protected static <Req, Resp> Resp executeWithOptions(Req request, SyncMethodWithRequestOptions<Req, Resp> syncMethod,
+                                              AsyncMethodWithRequestOptions<Req, Resp> asyncMethod, RequestOptions options) throws IOException {


after some discussion with @nik9000, I tried refactoring the existing execute and associated functional interfaces, and it seems like a big enough refactor that I would rather not include in this PR. I'm happy to follow up on a separate PR, but these changes make it possible to migrate the *IT.java tests piecemeal.

gotcha, this has to do with #31069 then, which I will get in today hopefully, then no refactorings should be needed.

…an match 0 tasks

pcsanwald · 2018-06-06T22:39:28Z

@javanna @nik9000 I've merged latest changes from master; this should be ready for final review!

javanna

left some super minor nits, LGTM. Feel free to merge once those are addressed, all good on my end. Thanks!

javanna · 2018-06-07T09:14:04Z

client/rest-high-level/src/main/java/org/elasticsearch/client/RequestConverters.java

+        Request request = new Request(HttpPost.METHOD_NAME, "/_tasks/_cancel");
+        Params params = new Params(request);
+        params.withTimeout(
+            cancelTasksRequest.getTimeout())


nit: can you adjust the indentation?

javanna · 2018-06-07T09:14:21Z

client/rest-high-level/src/main/java/org/elasticsearch/client/ClusterClient.java

@@ -95,4 +95,5 @@ public void putSettingsAsync(ClusterUpdateSettingsRequest clusterUpdateSettingsR
        restHighLevelClient.performRequestAsyncAndParseEntity(clusterUpdateSettingsRequest, RequestConverters::clusterPutSettings,
                ClusterUpdateSettingsResponse::fromXContent, listener, emptySet(), headers);
    }
+


nit: remove empty line?

javanna · 2018-06-07T09:14:59Z

client/rest-high-level/src/main/java/org/elasticsearch/client/TasksClient.java

@@ -65,4 +67,45 @@ public void listAsync(ListTasksRequest request, RequestOptions options, ActionLi
        restHighLevelClient.performRequestAsyncAndParseEntity(request, RequestConverters::listTasks, options,
                ListTasksResponse::fromXContent, listener, emptySet());
    }
+
+    /**
+     * Cancel one or more cluster tasks using the Task Management API


can you add a punctuation mark at the end of the sentence? It will make the generated html slightly better

javanna · 2018-06-07T09:15:13Z

client/rest-high-level/src/main/java/org/elasticsearch/client/TasksClient.java

+    }
+
+    /**
+     * Asynchronously cancel one or more cluster tasks using the Task Management API


Same here, can you add the punctuation mark

javanna · 2018-06-07T09:15:29Z

client/rest-high-level/src/test/java/org/elasticsearch/client/ClusterClientIT.java

@@ -108,4 +108,5 @@ public void testClusterUpdateSettingNonExistent() {
        assertThat(exception.getMessage(), equalTo(
                "Elasticsearch exception [type=illegal_argument_exception, reason=transient setting [" + setting + "], not recognized]"));
    }
+


nit: remove empty line?

javanna · 2018-06-07T09:18:08Z

client/rest-high-level/src/test/java/org/elasticsearch/client/TasksIT.java

+            highLevelClient().tasks()::listAsync
+        );
+
+        // TODO[PCS] submit a task that is cancellable and assert it's cancelled


is this TODO still valid? do we need to address it before merging?

javanna · 2018-06-07T09:18:14Z

client/rest-high-level/src/test/java/org/elasticsearch/client/TasksIT.java

+        cancelTasksRequest.setReason("testreason");
+
+        CancelTasksResponse response = execute(cancelTasksRequest, highLevelClient().tasks()::cancel, highLevelClient().tasks()::cancelAsync
+        );


nit: adjust indentation?

javanna · 2018-06-07T09:18:27Z

...level/src/test/java/org/elasticsearch/client/documentation/ClusterClientDocumentationIT.java

@@ -178,4 +178,5 @@ public void onFailure(Exception e) {
            assertTrue(latch.await(30L, TimeUnit.SECONDS));
        }
    }
+


nit: remove empty line?

javanna · 2018-06-07T09:18:51Z

.../main/java/org/elasticsearch/action/admin/cluster/node/tasks/cancel/CancelTasksResponse.java

+
+    @Override
+    public String toString() {
+        return Strings.toString(this);


should we pretty print the output?

javanna · 2018-06-07T09:19:18Z

server/src/test/java/org/elasticsearch/tasks/CancelTasksResponseTests.java

+        return true;
+    }
+
+


nit: remove one of the two empty lines?

* elastic/master: (53 commits) Painless: Restructure/Clean Up of Spec Documentation (#31013) Update ignore_unmapped serialization after backport Add back dropped substitution on merge high level REST api: cancel task (#30745) Enable engine factory to be pluggable (#31183) Remove vestiges of animal sniffer (#31178) Rename elasticsearch-nio to nio (#31186) Rename elasticsearch-core to core (#31185) Move cli sub-project out of server to libs (#31184) [DOCS] Fixes broken link in auditing settings QA: Better seed nodes for rolling restart [DOCS] Moves ML content to stack-docs [DOCS] Clarifies recommendation for audit index output type (#31146) Add nio-transport as option for http smoke tests (#31162) QA: Set better node names on rolling restart tests Add support for ignore_unmapped to geo sort (#31153) Share common parser in some AcknowledgedResponses (#31169) Fix random failure on SearchQueryIT#testTermExpansionExceptionOnSpanFailure Remove reference to multiple fields with one name (#31127) Remove BlobContainer.move() method (#31100) ...

javanna · 2018-06-08T06:34:23Z

@pcsanwald are you going to backport this to 6.x?

pcsanwald · 2018-06-08T13:16:49Z

Yes, great point. Will do today

On Thu, Jun 7, 2018 at 11:35 PM Luca Cavanna ***@***.***> wrote: @pcsanwald <https://github.com/pcsanwald> are you going to backport this to 6.x? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#30745 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAJ96vJss-NWUDs04og5thqX0rTfpDxPks5t6hsugaJpZM4UGGJP> .

-- --paul

* Initial commit of rest high level exposure of cancel task * fix javadocs * address some code review comments * update branch to use tasks namespace instead of cluster * High-level client: list tasks failure to not lose nodeId This commit reworks testing for `ListTasksResponse` so that random fields insertion can be tested and xcontent equivalence can be checked too. Proper exclusions need to be configured, and failures need to be tested separately. This helped finding a little problem, whenever there is a node failure returned, the nodeId was lost as it was never printed out as part of the exception toXContent. * added comment * merge from master * re-work CancelTasksResponseTests to separate XContent failure cases from non-failure cases * remove duplication of logic in parser creation * code review changes * refactor TasksClient to support RequestOptions * add tests for parent task id * address final PR review comments, mostly formatting and such

pcsanwald · 2018-06-08T15:24:55Z

@javanna backported into 6.x with 540b117

javanna · 2018-06-08T15:26:41Z

nice thanks @pcsanwald !

* 6.x: Move default location of dependencies report (#31228) Remove dependencies report task dependencies (#31227) Add recognition of MPL 2.0 (#31226) Fix unknown licenses (#31223) Fully encapsulate LocalCheckpointTracker inside of the engine (#31213) Remove version from license file name for GCS SDK (#31221) Remove DocumentFieldMappers#simpleMatchToFullName. (#31041) [DOCS] Removes 6.3.1 release notes [DOCS] Splits release notes by major version Remove DocumentFieldMappers#smartNameFieldMapper, as it is no longer needed. (#31018) Remove extraneous references to 'tokenized' in the mapper code. (#31010) SQL: Make a single JDBC driver jar (#31012) QA: Fix rolling restart tests some more Allow to trim all ops above a certain seq# with a term lower than X high level REST api: cancel task (#30745) Mute TokenBackwardsCompatibilityIT.testMixedCluster Mute WatchBackwardsCompatibilityIT.testWatcherRestart Enhance license detection for various licenses (#31198) [DOCS] Add note about long-lived idle connections (#30990) Add high-level client methods that accept RequestOptions (#31069) Remove RestGetAllMappingsAction (#31129) Move RestGetSettingsAction to RestToXContentListener (#31101) Move number of language analyzers to analysis-common module (#31143) flush job to ensure all results have been written (#31187)

Initial commit of rest high level exposure of cancel task

c7e221f

pcsanwald added WIP :Core/Java High Level REST Client labels May 20, 2018

Paul Sanwald added 2 commits May 20, 2018 16:57

fix javadocs

1df3544

Merge branch 'master' into rest-high-level-cancel-task

332f4dd

nik9000 reviewed May 21, 2018

View reviewed changes

javanna reviewed May 22, 2018

View reviewed changes

address some code review comments

9495732

javanna mentioned this pull request May 28, 2018

Java high-level REST client completeness #27205

Closed

80 tasks

javanna reviewed May 28, 2018

View reviewed changes

Paul Sanwald added 2 commits May 29, 2018 15:17

merge in changes from master that refactor list from cluster into tas…

9869569

…k namespace

update branch to use tasks namespace instead of cluster

5654c21

javanna reviewed May 31, 2018

View reviewed changes

javanna and others added 8 commits May 31, 2018 17:12

added comment

3ed1829

merge work from elastic#31001

62b161a

merge from master

314e1ff

merge from master

58e9539

re-work CancelTasksResponseTests to separate XContent failure cases f…

c51eb3c

…rom non-failure cases

merge from master

9e748ee

remove duplication of logic in parser creation

a511490

Merge branch 'master' into rest-high-level-cancel-task

96e3fa2

javanna reviewed Jun 5, 2018

View reviewed changes

javanna added >enhancement v7.0.0 v6.4.0 and removed WIP labels Jun 5, 2018

javanna reviewed Jun 5, 2018

View reviewed changes

Paul Sanwald added 3 commits June 5, 2018 11:05

code review changes

4260549

Merge branch 'master' into rest-high-level-cancel-task

f646fa4

refactor TasksClient to support RequestOptions

31f08ef

pcsanwald commented Jun 6, 2018

View reviewed changes

merge in master and change tasks param to optional, as cancel tasks c…

0c05fe7

…an match 0 tasks

pcsanwald force-pushed the rest-high-level-cancel-task branch from 36e9fcf to 0c05fe7 Compare June 6, 2018 22:35

add tests for parent task id

783d00a

javanna approved these changes Jun 7, 2018

View reviewed changes

address final PR review comments, mostly formatting and such

4e34cfd

pcsanwald merged commit e82e5cc into elastic:master Jun 7, 2018

pcsanwald deleted the rest-high-level-cancel-task branch June 7, 2018 21:02

colings86 added v7.0.0-beta1 and removed v7.0.0 labels Feb 7, 2019

high level REST api: cancel task #30745

high level REST api: cancel task #30745

Conversation

pcsanwald commented May 20, 2018

elasticmachine commented May 20, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

nik9000 commented May 21, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

pcsanwald May 23, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

pcsanwald commented May 30, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

javanna Jun 5, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

javanna left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

pcsanwald commented Jun 6, 2018

javanna left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

javanna commented Jun 8, 2018

pcsanwald commented Jun 8, 2018 via email

pcsanwald commented Jun 8, 2018 • edited Loading

javanna commented Jun 8, 2018

pcsanwald May 23, 2018 •

edited

Loading

javanna Jun 5, 2018 •

edited

Loading

pcsanwald commented Jun 8, 2018 •

edited

Loading