[ML] Avoid unnecessary calls to Task#getDescription in model download

[davidkyle]

The model download code checks for an existing download task by iterating over the currently running tasks comparing the task description. The call to getDescription() can sometimes throw- see the stack trace below.

This change rearranges the conditional check so that getDescription() is only called on instances of ModelDownloadTask.

Caused by: java.lang.UnsupportedOperationException: This should not be XContent serialized
	at [email protected]/org.elasticsearch.xpack.inference.rank.textsimilarity.TextSimilarityRankBuilder.doXContent(TextSimilarityRankBuilder.java:106)
	at [email protected]/org.elasticsearch.search.rank.RankBuilder.toXContent(RankBuilder.java:70)
	at [email protected]/org.elasticsearch.xcontent.XContentBuilder.value(XContentBuilder.java:993)
	at [email protected]/org.elasticsearch.xcontent.XContentBuilder.value(XContentBuilder.java:982)
	at [email protected]/org.elasticsearch.xcontent.XContentBuilder.field(XContentBuilder.java:974)
	at [email protected]/org.elasticsearch.search.builder.SearchSourceBuilder.innerToXContent(SearchSourceBuilder.java:1713)
	at [email protected]/org.elasticsearch.search.builder.SearchSourceBuilder.toXContent(SearchSourceBuilder.java:1852)
	at [email protected]/org.elasticsearch.common.xcontent.XContentHelper.toXContent(XContentHelper.java:653)
	at [email protected]/org.elasticsearch.common.xcontent.XContentHelper.toXContent(XContentHelper.java:633)
	at [email protected]/org.elasticsearch.search.builder.SearchSourceBuilder.toString(SearchSourceBuilder.java:2177)
	at [email protected]/org.elasticsearch.action.search.SearchRequest.buildDescription(SearchRequest.java:747)
	at java.base/java.util.stream.ReferencePipeline$3$1.accept(ReferencePipeline.java:215)
	at java.base/java.util.ArrayList$ArrayListSpliterator.forEachRemaining(ArrayList.java:1709)
	at java.base/java.util.stream.AbstractPipeline.copyInto(AbstractPipeline.java:570)
	at java.base/java.util.stream.AbstractPipeline.wrapAndCopyInto(AbstractPipeline.java:560)
	at java.base/java.util.stream.ReduceOps$ReduceOp.evaluateSequential(ReduceOps.java:921)
	at java.base/java.util.stream.AbstractPipeline.evaluate(AbstractPipeline.java:265)
	at java.base/java.util.stream.ReferencePipeline.collect(ReferencePipeline.java:727)
Caused by: java.lang.UnsupportedOperationException: This should not be XContent serialized

	at [email protected]/org.elasticsearch.action.search.MultiSearchRequest$1.getDescription(MultiSearchRequest.java:381)
	at org.elasticsearch.xpack.ml.packageloader.action.TransportLoadTrainedModelPackage.handleDownloadInProgress(TransportLoadTrainedModelPackage.java:154)

[elasticsearchmachine]

Pinging @elastic/ml-core (Team:ML)

[elasticsearchmachine]

Hi @davidkyle, I've created a changelog YAML for you.

[kderusso]

Thanks for the fix! Could we please also add a test for this case?

[Mikep86]

Thanks for the quick turnaround on this! Agree with Kathleen that a test would be nice, but this is such a straightforward change that we could do it in a follow-up if we just want to get the fix live ASAP.

[Mikep86]

💚 All backports created successfully

Status	Branch	Result
✅	8.x
✅	9.0
✅	8.18

Questions ?

Please refer to the Backport tool documentation