Archival and Retrieval Flow¶

Archival is split into two subflows, Create datablocks and Move datablocks to LTSAn archival task can contain multiple datasets; for simplicity the case with only one is depicted here.

Nomenclature¶

Critical Sections¶

A sequence of tasks that need to succeed atomically (all succeed or all are rolled back) are depicted as critical tasks below. They have a specific error handling associated in order to roll back all the task in that sequence.

Retries¶

All task interacting with external services (i.e. Scicat) have retries implemented. Since they are not done in a task level but on the http request level, they are not depicted below.

Reading from the LTS has retries implemented on a task level and is therefore depicted below.

Sequence Diagrams¶

Archival¶

The diagramm below shows the different steps in the archiving flow. Each step executed by the Worker is implemented as its own task (expect steps executed in error handling).

sequenceDiagram
  autonumber
  participant AS as Archival Service
  participant L as Landing Zone
  participant W as Worker
  participant ST as Staging
  participant S as SciCat
  participant LTS as LTS


  Note over AS, S: Subflow: Create datablocks

  S --) AS: Archive: POST /api/v1/jobs
  Note right of AS: Job {"id" : "id", "type":"archive", "datasetlist": [], ... } as defined in Scicat

  AS --) W: Create archival flow for each dataset
  AS -->> S: Respond to Scicat
  Note left of S: {"uuid": "<flow uuid>", "name": "<flow name>"}

  W --) S: PATCH /api/v4/Jobs/{JobId}
  Note left of S: {"jobStatusMessage": "inProgress"}, 

  critical Job for datasetlist
    Note over L,LTS: Subflow: Move dataset to LTS
  critical Subflow for {DatasetId}
              W --) S:  PATCH /api/v4/Dataset/{DatasetId}
              Note left of S: {"datasetlifecycle": {"archiveStatusMessage": "started"}}  
              W --> S: GET /api/v4/Datasets/{dataset_id}/origdatablocks
        W -->> L: Create datablocks
        L -->> ST: Move datablocks to staging
        W -->> S: Register datablocks POST /api/v4/Datasets/{DatasetId}/datablocks
        W -->> L: Cleanup dataset files, datablocks
        Note right of L: Dataset files only get cleaned up when everythings succeeds
    option Failure user error
        W --) S: Report error: PATCH /api/v4/Dataset/{DatasetId}
        Note left of S: {"datasetlifecycle:{"archiveStatusMessage": "missingFileserror", "archivable": False, "retrievable":False}}
        W -->> ST: Cleanup Datablocks
        W -->> L: Cleanup Datablocks
        Note right of L: No cleanup of dataset files
    option Failure system error
        W --) S: Report error: PATCH /api/v4/Dataset/{DatasetId}
        Note left of S: {"datasetlifecycle:{"archiveStatusMessage": "scheduleArchiveJobFailed", "archivable": "false", "retrievable":"false"}}
        W -->> ST: Cleanup datablocks
        W -->> L: Cleanup datablocks
        Note right of L: No cleanup of dataset files
    end
    option Failure user error
        W --) S: Report error: PATCH /api/v4/Jobs/{JobId}
        Note left of S: {"jobStatusMessage": "finishedWithDataseterrors"}
    option Failure system error
        W --) S: Report error: PATCH /api/v4/Jobs/{JobId}
        Note left of S: {"jobStatusMessage": "finishedUnsuccessful"}
  end
Note over W, LTS: Subflow: Move datablocks to LTS
loop for each datablock 
  critical
      ST -->> LTS: Move datablock to LTS
      loop Retry
        ST -->> LTS: Verify checksum of datablock to LTS with retries
      end
  option Moving datablock to LTS Failed
      W -->> LTS: Cleanup LTS folder
      W -->> S: Report error
      Note left of S: {"jobStatusMessage": "scheduleArchiveJobFailed"}
  end
  critical
      W -->> LTS: Verify datablock in LTS
      W -->> L: Cleanup dataset files, datablock
  option Datablock validation failed
      W -->> S: Report error
      Note left of S: {"jobStatusMessage": "scheduleArchiveJobFailed"}
      W -->> LTS: Cleanup datablocks
  end
end
    W -->> S:  PATCH /api/v4/Datasets/{DatasetId}
    Note left of S: {"datasetlifecycle": {"archivable": False, "retrievable": True, "archiveStatusMessage": "datasetOnArchiveDisk"}} 
    W -->> S: PATCH /api/v4/Jobs/{JobId}
    Note left of S: {"jobStatusMessage": "finishedSuccessful"} 

Retrieval¶

The diagramm below shows the different steps in the retrieval flow. Each step executed by the Worker is implemented as its own task (expect steps executed in error handling).

sequenceDiagram
  autonumber
  participant AS as Archival Service
  participant J as Worker
  participant ST as Staging
  participant S as SciCat
  participant LTS as LTS

  S --) AS: Retrieve: POST /api/v1/jobs
  Note right of AS: Job {"id" : "id", "type":"retrieve", "datasetlist": [], ... } as defined in Scicat
  AS --) J: Create Retrieval Flow
  AS -->> S: Respond to Scicat
  Note left of S: {"uuid": "<flow uuid>", "name": "<flow name>"}

    J -->> S: PATCH /api/v4/Jobs/{JobId}
    Note left of S: {"jobStatusMessage": "inProgress"}  
    J -->> S: PATCH /api/v4/Datasets/{DatasetId}
    Note left of S: {"datasetlifecycle": {"retrieveStatusMessage": "started"}}  
  critical
      LTS -->> ST: Download Datablocks from LTS
  option Retrieval Failure
      J --) S: Report Error: PATCH /api/v4/Dataset/{DatasetId}
          Note left of S: {"retrieveStatusMessage": datasetRetrievalFailed,<br> "retrieveReturnMessage": "retrievalFailed"}
      J --) S: Report Error: PATCH /api/v4/Jobs/{JobId}
         Note left of S: {"jobStatusMessage": "finishedWithDatasetErrors",<br> "jobResultObject": [JobResultEntry]}
      J -->> ST: Cleanup Files
  end
  critical
    J -->> ST: Validate Datablocks
  option Validation Failure
      J -->> S: Report Error: PATCH /api/v4/Dataset/{DatasetId}
        Note left of S: {"retrieveStatusMessage": datasetRetrievalFailed,<br> "retrieveReturnMessage": "retrievalFailed"}
      J -->> ST: Cleanup Files
  end
      J -->> S: PATCH /api/v4/Jobs/{JobId}
      Note left of S: {"jobStatusMessage": "finishedSuccessful"} 
      J -->> S: PATCH /api/v4/Datasets/{DatasetId}
      Note left of S: {"datasetlifecycle": {"retrieveStatusMessage": "datasetRetrieved",<br>  "retrievable": True }} 

Note: updatedBy and updatedAt are omitted for brevity but need to be set for every update of the job status and datsetlifecycle as well.

JobResult¶

The result of a retrieval job contains a list of JobResultEntry objects, which contain a url in order for end users to download the datablocks.

JobResultEntry¶

 {
    "datasetId": "<datasetId>",
    "name": "str",
    "size": "str",
    "archiveId": "str",
    "url": "str"
 }