limit() function execution order in Azure Cosmos GraphDb?

Question

I have encountered a problem when using Cosmos GraphDb. Vertex "EXO-OAuth-DailyAvailabilityPuller-JobDefinition" has edges named "JobDefinitionToJobHistory," and the target of these edges are the job history vertexes. In 2022-02-12, vertex "EXO-OAuth-DailyAvailabilityPuller-JobDefinition" has two job history vertexes, the first status of job history vertex is failed (ExecutionStartTime is 2022-02-12T11:58:47.3459708Z) and the second status of job history vertex is succeed (ExecutionStartTime is 2022-02-12T14:03:02.8664959Z).

So, when I execute the Gremlin query below, I get the expected result which is a vertex with a succeed status(2 means succeed).

   g.V("EXO-OAuth-DailyAvailabilityPuller-JobDefinition").out('JobDefinitionToJobHistory').has("ExecutionStartTime", between("2022-02-12T00:00:00Z", "2022-02-13T00:00:00Z")).order().by('ExecutionStartTime', decr).limit(1)  

But when I execute the Gremlin query below, I get the incomprehensible result which is a vertex with a failed status (1 means failed). The expected result of this query I think should be empty or null.

   g.V("EXO-OAuth-DailyAvailabilityPuller-JobDefinition").out('JobDefinitionToJobHistory').has("ExecutionStartTime", between("2022-02-12T00:00:00Z", "2022-02-13T00:00:00Z")).order().by('ExecutionStartTime', decr).limit(1).has("Status", 1)  

I am very confused about the second Gremlin query result. Because I can't recreate this problem in TinkerGraph. So I don't know if this result is related to Azure Cosmos GraphDb. Hope you can give me some help, thanks!