You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 6 Next »

Status

Current state: [Under Discussion]

Discussion thread: http://apache-flink-mailing-list-archive.1008284.n3.nabble.com/DISCUSS-FLIP-75-Flink-Web-UI-Improvement-Proposal-td33540.html

JIRA:  Unable to render Jira issues macro, execution error.

Released: <Flink Version>

Please keep the discussion on the mailing list rather than commenting on the wiki (wiki discussions get unwieldy fast).

Motivation

According to the docs, there may exist more than one attempt in a subtask, but there is no way to get the attempt history list in the REST API, users have no way to know if the subtask has failed before.

In the timeline page, the Web UI can only get and visualize the latest execution attempt of a subtask timeline, there is no way to get a failed attempt timeline in the current REST API. 

Proposed Changes

Add the attempt history

  • We can add the attempt history under the subtasks drawer on the job vertex page.
  • Display all attempt timeline.

Frontend Design

Add view attempt history menu in the subtask, when popup an attempts history modal when user clicking it.

When users click the vertex timeline, display all the subtask attempts timeline with subtaskId-host-attemptId.

REST API Design

  • get prior execution attempt

AccessExecution execution = executionVertex.getCurrentExecutionAttempt();

int currentAttemptNum = execution.getAttemptNumber();

JobID jobID = request.getPathParameter(JobIDPathParameter.class);

JobVertexID jobVertexID = request.getPathParameter(JobVertexIdPathParameter.class);

List<SubtaskExecutionAttemptDetailsInfo> allAttempts = new ArrayList<>();

allAttempts.add(SubtaskExecutionAttemptDetailsInfo.create(execution, metricFetcher, jobID, jobVertexID));

if (currentAttemptNum > 0) {

  for (int i = currentAttemptNum - 1; i >= 0; i--) {

    AccessExecution currentExecution = executionVertex.getPriorExecutionAttempt(i);

    if (currentExecution != null) {

      allAttempts.add(SubtaskExecutionAttemptDetailsInfo.create(currentExecution, metricFetcher, jobID, jobVertexID));

    }

  }

}


  • add SubtaskAllExecutionAttemptsDetailsHandler for failed attempt
  • url /jobs/:jobid/vertices/:vertexid/subtasks/:subtaskIndex/attempts
  • response:

{

  "attempts" : {

    "type" : "array",

    "items" : {

      "type" : "object",

      "id" : "urn:jsonschema:org:apache:flink:runtime:rest:messages:job:SubtaskExecutionAttemptDetailsInfo",

      "properties" : {

        "subtask" : {

          "type" : "integer"

        },

        "status" : {

          "type" : "string",

          "enum" : [ "CREATED", "SCHEDULED", "DEPLOYING", "RUNNING", "FINISHED", "CANCELING", "CANCELED", "FAILED", "RECONCILING" ]

        },

        "attempt" : {

          "type" : "integer"

        },

        "host" : {

          "type" : "string"

        },

        "start-time" : {

          "type" : "integer"

        },

        "end-time" : {

          "type" : "integer"

        },

        "duration" : {

          "type" : "integer"

        },

        "metrics" : {

          "type" : "object",

          "id" : "urn:jsonschema:org:apache:flink:runtime:rest:messages:job:metrics:IOMetricsInfo",

          "properties" : {

            "read-bytes" : {

              "type" : "integer"

            },

            "read-bytes-complete" : {

              "type" : "boolean"

            },

            "write-bytes" : {

              "type" : "integer"

            },

            "write-bytes-complete" : {

              "type" : "boolean"

            },

            "read-records" : {

              "type" : "integer"

            },

            "read-records-complete" : {

              "type" : "boolean"

            },

            "write-records" : {

              "type" : "integer"

            },

            "write-records-complete" : {

              "type" : "boolean"

            }

          }

        }

      }

    }

  }

}

  • add Collection<SubtaskTimeInfo> in SubtasksTimesHandler result, which comes from ExecutionVertex’s prior executions.
  • url: /jobs/:jobid/vertices/:vertexid/subtasktimes
  • response

{

  "type" : "object",

  "id" : "urn:jsonschema:org:apache:flink:runtime:rest:messages:SubtasksTimesInfo",

  "properties" : {

    "id" : {

      "type" : "string"

    },

    "name" : {

      "type" : "string"

    },

    "now" : {

      "type" : "integer"

    },

    "subtasks" : {

      "type" : "array",

      "items" : {

        "type" : "object",

        "id" : "urn:jsonschema:org:apache:flink:runtime:rest:messages:SubtasksTimesInfo:SubtaskTimeInfo",

        "properties" : {

          "subtask" : {

            "type" : "integer"

          },

          "host" : {

            "type" : "string"

          },

          "duration" : {

            "type" : "integer"

          },

          "timestamps" : {

            "type" : "object",

            "additionalProperties" : {

              "type" : "integer"

            }

          },

          "attempt-num": {

            "type" : "integer"

          },

          "attempts-time-info": {

            "type": "array",

            "items" : {

              "type" : "object",

              "id" : "urn:jsonschema:org:apache:flink:runtime:rest:messages:SubtaskAttemptTimeInfo",

              "properties" : {

                "subtask" : {

                  "type" : "integer"

                },

                "host" : {

                  "type" : "string"

                },

                "duration" : {

                  "type" : "integer"

                },

                "timestamps" : {

                  "type" : "object",

                  "additionalProperties" : {

                    "type" : "integer"

                  }

                },

                "attempt-num": {

                  "type" : "integer"

                },

              }

            }

          } 

        }

      }

    }

  }

}

Test Plan

Everything can be tested with unit tests.

  • No labels