Stale Data

The stale data error occurs when Corteza detects that a request attempted to change a resource that was already changed by another request.

The diagram illustrates how the stale data error occurs.

@startuml
concise "Alice" as alc
concise "Bob" as bb

@0
alc is "Fetching resource R"
bb is "Fetching resource R"

@500
alc is "Editing resource R"
bb is "Editing resource R"

@900
alc is "Updating resource R"

@1600
alc is {hidden}

@1800
bb is "Updating resource R"

@2200
bb is {hidden}


highlight 0 to 500 #Gold;line:DimGrey : \t\tBoth Alice and Bob fetch\nthe same version of resource
highlight 900 to 1600 #Gold;line:DimGrey : Alice caused the resource on the server to change
highlight 1800 to 2200 #Pink;line:DimGrey : Bob tried to update a resource\ndetermined as stale

@enduml

Example: Bulk record update

The example utilizes batch record creation where we are trying to update the record $RECORD_ID twice.

curl -X POST "$BASE_URL/api/compose/namespace/$NAMESPACE_ID/module/$MODULE_ID/record/" \
  -H 'accept: application/json, text/plain, */*' \
  -H "authorization: Bearer $JWT" \
  -H 'content-type: application/json' \
  --data-raw "{
    \"records\": [{
      \"set\": [{
        \"recordID\": \"$RECORD_ID\",
        \"moduleID\": \"$MODULE_ID\",
        \"values\": [{ \"name\": \"name\", \"value\": \"Some value\" }],
        \"namespaceID\": \"$NAMESPACE_ID\",
        \"createdAt\": \"2022-02-17T11:44:20Z\",
        \"updatedAt\": \"2022-02-17T17:13:42Z\"
      },
      {
        \"recordID\": \"$RECORD_ID\",
        \"moduleID\": \"$MODULE_ID\",
        \"values\": [{ \"name\": \"name\", \"value\": \"Some OTHER value\" }],
        \"namespaceID\": \"$NAMESPACE_ID\",
        \"createdAt\": \"2022-02-17T11:44:20Z\",
        \"updatedAt\": \"2022-02-17T17:13:42Z\"
      }]
    }]
  }" \
  --compressed

The stale data error occurs because the first record causes the resource to change on the server.

When the second resource is handled, the current resource on the server is no longer the one that the second record is referencing.

To solve the issue, either split the request in two or determine the latest version locally before sending the request.

Example: Workflow execution

The example utilizes the fork gateway where we first fetch the record and then attempt to update the same instance twice. Because one resolves before the other, the second one raises a stale data error

Use the two fork branches to update the values without saving the record to solve the issue.

Save the record at the very end when all of the values are set to what they should be.