mongo 2.15.0.alpha → 2.15.0

data/spec/spec_tests/data/sdam_integration/{isMaster-command-error.yml → hello-command-error.yml}

@@ -4,19 +4,19 @@ runOn:
     - minServerVersion: "4.4"
 
 database_name: &database_name "sdam-tests"
-collection_name: &collection_name "isMaster-command-error"
+collection_name: &collection_name "hello-command-error"
 
 data: []
 
 tests:
   - description: Command error on Monitor handshake
     # Configure the initial handshake to fail with a command error.
-    # Use times: 2 so that the RTT isMaster fails as well.
+    # Use times: 2 so that the RTT hello fails as well.
     failPoint:
       configureFailPoint: failCommand
       mode: { times: 2 }
       data:
-          failCommands: ["isMaster"]
+          failCommands: ["hello", "isMaster"]
           appName: commandErrorHandshakeTest
           closeConnection: false
           errorCode: 91 # ShutdownInProgress
@@ -33,12 +33,6 @@ tests:
         arguments:
           event: ServerMarkedUnknownEvent
           count: 1
-# https://jira.mongodb.org/browse/DRIVERS-1314
-#       - name: waitForEvent
-#         object: testRunner
-#         arguments:
-#           event: PoolClearedEvent
-#           count: 1
       # Perform an operation to ensure the node is discovered.
       - name: insertMany
         object: collection
@@ -47,7 +41,7 @@ tests:
             - _id: 1
             - _id: 2
       # We cannot assert the server was marked Unknown and pool was cleared an
-      # exact number of times because the RTT isMaster may or may not have
+      # exact number of times because the RTT hello may or may not have
       # triggered this failpoint as well.
       # - name: assertEventCount
       #   object: testRunner
@@ -90,9 +84,9 @@ tests:
           documents:
             - _id: 1
             - _id: 2
-      # Configure the next streaming isMaster check to fail with a command
+      # Configure the next streaming hello check to fail with a command
       # error.
-      # Use times: 2 so that the RTT isMaster is blocked as well.
+      # Use times: 2 so that the RTT hello is blocked as well.
       - name: configureFailPoint
         object: testRunner
         arguments:
@@ -100,7 +94,7 @@ tests:
             configureFailPoint: failCommand
             mode: { times: 2 }
             data:
-                failCommands: ["isMaster"]
+                failCommands: ["hello", "isMaster"]
                 appName: commandErrorCheckTest
                 closeConnection: false
                 blockConnection: true
@@ -112,14 +106,12 @@ tests:
         object: testRunner
         arguments:
           event: ServerMarkedUnknownEvent
-          # Ruby driver marks server unknown for each fail point hit
-          count: 2
-# Ruby driver normally produces 2 pool clears, but sometimes only 1
-#       - name: waitForEvent
-#         object: testRunner
-#         arguments:
-#           event: PoolClearedEvent
-#           count: 1
+          count: 1
+      - name: waitForEvent
+        object: testRunner
+        arguments:
+          event: PoolClearedEvent
+          count: 1
       # Perform an operation to ensure the node is rediscovered.
       - name: insertMany
         object: collection
@@ -132,14 +124,12 @@ tests:
         object: testRunner
         arguments:
           event: ServerMarkedUnknownEvent
-          # Ruby driver marks server unknown for each fail point hit
-          count: 2
-# Ruby driver normally produces 2 pool clears, but sometimes only 1
-#       - name: waitForEvent
-#         object: testRunner
-#         arguments:
-#           event: PoolClearedEvent
-#           count: 1
+          count: 1
+      - name: assertEventCount
+        object: testRunner
+        arguments:
+          event: PoolClearedEvent
+          count: 1
 
     expectations:
       - command_started_event:

data/spec/spec_tests/data/sdam_integration/{isMaster-network-error.yml → hello-network-error.yml}

@@ -1,22 +1,22 @@
-# Test SDAM error handling.
+#hello Test SDAM error handling.
 runOn:
     # failCommand appName requirements
     - minServerVersion: "4.4"
 
 database_name: &database_name "sdam-tests"
-collection_name: &collection_name "isMaster-network-error"
+collection_name: &collection_name "hello-network-error"
 
 data: []
 
 tests:
   - description: Network error on Monitor handshake
     # Configure the initial handshake to fail with a network error.
-    # Use times: 2 so that the RTT isMaster fails as well.
+    # Use times: 2 so that the RTT hello fails as well.
     failPoint:
       configureFailPoint: failCommand
       mode: { times: 2 }
       data:
-          failCommands: ["isMaster"]
+          failCommands: ["hello", "isMaster"]
           appName: networkErrorHandshakeTest
           closeConnection: true
     clientOptions:
@@ -32,12 +32,6 @@ tests:
         arguments:
           event: ServerMarkedUnknownEvent
           count: 1
-# https://jira.mongodb.org/browse/DRIVERS-1314
-#       - name: waitForEvent
-#         object: testRunner
-#         arguments:
-#           event: PoolClearedEvent
-#           count: 1
       # Perform an operation to ensure the node is discovered.
       - name: insertMany
         object: collection
@@ -46,7 +40,7 @@ tests:
             - _id: 1
             - _id: 2
       # We cannot assert the server was marked Unknown and pool was cleared an
-      # exact number of times because the RTT isMaster may or may not have
+      # exact number of times because the RTT hello may or may not have
       # triggered this failpoint as well.
       # - name: assertEventCount
       #   object: testRunner
@@ -89,9 +83,9 @@ tests:
           documents:
             - _id: 1
             - _id: 2
-      # Configure the next streaming isMaster check to fail with a non-timeout
+      # Configure the next streaming hello check to fail with a non-timeout
       # network error. Use times: 2 to ensure that the the Monitor check fails
-      # since the RTT isMaster may trigger this failpoint as well.
+      # since the RTT hello may trigger this failpoint as well.
       - name: configureFailPoint
         object: testRunner
         arguments:
@@ -99,7 +93,7 @@ tests:
             configureFailPoint: failCommand
             mode: { times: 2 }
             data:
-                failCommands: ["isMaster"]
+                failCommands: ["hello", "isMaster"]
                 appName: networkErrorCheckTest
                 closeConnection: true
       # The network error on the next check should mark the server Unknown and
@@ -122,7 +116,7 @@ tests:
             - _id: 3
             - _id: 4
       # We cannot assert the server was marked Unknown and pool was cleared an
-      # exact number of times because the RTT isMaster may or may not have
+      # exact number of times because the RTT hello may or may not have
       # triggered this failpoint as well.
       # - name: assertEventCount
       #   object: testRunner

data/spec/spec_tests/data/sdam_integration/{isMaster-timeout.yml → hello-timeout.yml}

@@ -4,19 +4,19 @@ runOn:
     - minServerVersion: "4.4"
 
 database_name: &database_name "sdam-tests"
-collection_name: &collection_name "isMaster-timeout"
+collection_name: &collection_name "hello-timeout"
 
 data: []
 
 tests:
   - description: Network timeout on Monitor handshake
     # Configure the initial handshake to fail with a timeout.
-    # Use times: 2 so that the RTT isMaster is blocked as well.
+    # Use times: 2 so that the RTT hello is blocked as well.
     failPoint:
       configureFailPoint: failCommand
       mode: { times: 2 }
       data:
-          failCommands: ["isMaster"]
+          failCommands: ["hello", "isMaster"]
           appName: timeoutMonitorHandshakeTest
           blockConnection: true
           blockTimeMS: 1000
@@ -33,12 +33,6 @@ tests:
         arguments:
           event: ServerMarkedUnknownEvent
           count: 1
-# https://jira.mongodb.org/browse/DRIVERS-1314
-#       - name: waitForEvent
-#         object: testRunner
-#         arguments:
-#           event: PoolClearedEvent
-#           count: 1
       # Perform an operation to ensure the node is discovered.
       - name: insertMany
         object: collection
@@ -47,7 +41,7 @@ tests:
             - _id: 1
             - _id: 2
       # We cannot assert the server was marked Unknown and pool was cleared an
-      # exact number of times because the RTT isMaster may or may not have
+      # exact number of times because the RTT hello may or may not have
       # triggered this failpoint as well.
       # - name: assertEventCount
       #   object: testRunner
@@ -90,8 +84,8 @@ tests:
           documents:
             - _id: 1
             - _id: 2
-      # Configure the next streaming isMaster check to fail with a timeout
-      # Use times: 2 so that the RTT isMaster is blocked as well.
+      # Configure the next streaming hello check to fail with a timeout
+      # Use times: 2 so that the RTT hello is blocked as well.
       - name: configureFailPoint
         object: testRunner
         arguments:
@@ -99,7 +93,7 @@ tests:
             configureFailPoint: failCommand
             mode: { times: 2 }
             data:
-                failCommands: ["isMaster"]
+                failCommands: ["hello", "isMaster"]
                 appName: timeoutMonitorCheckTest
                 blockConnection: true
                 # blockTimeMS is evaluated after the waiting for heartbeatFrequencyMS server-side, so this value only

data/spec/spec_tests/data/sdam_integration/rediscover-quickly-after-step-down.yml

@@ -31,19 +31,30 @@ tests:
             - _id: 4
       - name: recordPrimary
         object: testRunner
+      # Unfreeze a secondary with replSetFreeze:0 to ensure a speedy election.
+      - name: runAdminCommand
+        object: testRunner
+        command_name: replSetFreeze
+        arguments:
+          command:
+            replSetFreeze: 0
+          readPreference:
+            mode: Secondary
       # Run replSetStepDown on the meta client.
       - name: runAdminCommand
         object: testRunner
         command_name: replSetStepDown
         arguments:
           command:
-            replSetStepDown: 1
-            secondaryCatchUpPeriodSecs: 1
+            replSetStepDown: 30
+            secondaryCatchUpPeriodSecs: 30
             force: false
       - name: waitForPrimaryChange
         object: testRunner
         arguments:
-          timeoutMS: 5000
+          # We use a relatively large timeout here to workaround slow
+          # elections on Windows, possibly caused by SERVER-48154.
+          timeoutMS: 15000
       # Rediscover the new primary.
       - name: insertMany
         object: collection

data/spec/spec_tests/data/sdam_monitoring/discovered_standalone.yml

@@ -5,7 +5,7 @@ phases:
     responses:
       -
         - "a:27017"
-        - { ok: 1, ismaster: true, minWireVersion: 0, maxWireVersion: 4 }
+        - { ok: 1, helloOk: true, isWritablePrimary: true, minWireVersion: 0, maxWireVersion: 4 }
 
     outcome:
       events:

data/spec/spec_tests/data/sdam_monitoring/replica_set_with_no_primary.yml

@@ -7,7 +7,8 @@ phases:
         - "a:27017"
         - 
           ok: 1
-          ismaster: false
+          helloOk: true
+          isWritablePrimary: false
           secondary: true
           setName: "rs"
           setVersion: 1

data/spec/spec_tests/data/sdam_monitoring/replica_set_with_primary.yml

@@ -7,7 +7,8 @@ phases:
         - "a:27017"
         - 
           ok: 1
-          ismaster: true
+          helloOk: true
+          isWritablePrimary: true
           setName: "rs"
           setVersion: 1
           primary: "a:27017"

data/spec/spec_tests/data/sdam_monitoring/replica_set_with_removal.yml

@@ -43,7 +43,8 @@ phases:
         - "a:27017"
         - {
             ok: 1,
-            ismaster: true,
+            helloOk: true,
+            isWritablePrimary: true,
             setName: "rs",
             setVersion: 1.0,
             primary: "a:27017",
@@ -53,7 +54,7 @@ phases:
           }
       -
         - "b:27017"
-        - { ok: 1, ismaster: true }
+        - { ok: 1, helloOk: true, isWritablePrimary: true }
     outcome:
       events:
         -

data/spec/spec_tests/data/sdam_monitoring/required_replica_set.yml

@@ -7,7 +7,8 @@ phases:
         - "a:27017"
         - {
             ok: 1,
-            ismaster: true,
+            helloOk: true,
+            isWritablePrimary: true,
             setName: "rs",
             setVersion: 1.0,
             primary: "a:27017",

data/spec/spec_tests/data/sdam_monitoring/standalone.yml

@@ -5,7 +5,7 @@ phases:
     responses:
       -
         - "a:27017"
-        - { ok: 1, ismaster: true, minWireVersion: 0, maxWireVersion: 4 }
+        - { ok: 1, helloOk: true, isWritablePrimary: true, minWireVersion: 0, maxWireVersion: 4 }
 
     outcome:
       events:

data/spec/spec_tests/data/sdam_monitoring/standalone_suppress_equal_description_changes.yml

@@ -5,10 +5,10 @@ phases:
     responses:
       -
         - "a:27017"
-        - { ok: 1, ismaster: true, minWireVersion: 0, maxWireVersion: 4 }
+        - { ok: 1, helloOk: true, isWritablePrimary: true, minWireVersion: 0, maxWireVersion: 4 }
       -
         - "a:27017"
-        - { ok: 1, ismaster: true, minWireVersion: 0, maxWireVersion: 4 }
+        - { ok: 1, helloOk: true, isWritablePrimary: true, minWireVersion: 0, maxWireVersion: 4 }
 
     outcome:
       events:

data/spec/spec_tests/data/transactions/error-labels.yml

@@ -5,6 +5,9 @@ runOn:
     -
         minServerVersion: "4.1.8"
         topology: ["sharded"]
+        # serverless proxy doesn't append error labels to errors in transactions
+        # caused by failpoints (CLOUDP-88216)
+        serverless: "forbid"
 
 database_name: &database_name "transaction-tests"
 collection_name: &collection_name "test"

data/spec/spec_tests/data/transactions/mongos-pin-auto.yml

@@ -4,6 +4,9 @@ runOn:
     -
         minServerVersion: "4.1.8"
         topology: ["sharded"]
+        # serverless proxy doesn't append error labels to errors in transactions
+        # caused by failpoints (CLOUDP-88216)
+        serverless: "forbid"
 
 database_name: &database_name "transaction-tests"
 collection_name: &collection_name "test"

data/spec/spec_tests/data/transactions/mongos-recovery-token.yml

@@ -2,6 +2,8 @@ runOn:
     -
         minServerVersion: "4.1.8"
         topology: ["sharded"]
+        # serverless proxy doesn't use recovery tokens
+        serverless: "forbid"
 
 database_name: &database_name "transaction-tests"
 collection_name: &collection_name "test"

data/spec/spec_tests/data/transactions/pin-mongos.yml

@@ -16,6 +16,9 @@ runOn:
     -
         minServerVersion: "4.1.8"
         topology: ["sharded"]
+        # serverless proxy doesn't append error labels to errors in transactions
+        # caused by failpoints (CLOUDP-88216)
+        serverless: "forbid"
 
 database_name: &database_name "transaction-tests"
 collection_name: &collection_name "test"
@@ -461,7 +464,7 @@ tests:
         result:
           insertedId: 3
       # Enable the fail point only on the Mongos that session0 is pinned to.
-      # Fail isMaster to prevent the heartbeat requested directly after the
+      # Fail hello/legacy hello to prevent the heartbeat requested directly after the
       # insert error from racing with server selection for the commit.
       # Note: times: 7 is slightly artbitrary but it accounts for one failed
       # insert and some SDAM heartbeats. A test runner will have multiple
@@ -475,7 +478,7 @@ tests:
             configureFailPoint: failCommand
             mode: { times: 7 }
             data:
-              failCommands: ["insert", "isMaster"]
+              failCommands: ["insert", "isMaster", "hello"]
               closeConnection: true
       - name: insertOne
         object: collection
@@ -495,7 +498,7 @@ tests:
       #
       # Note that the commit attempt should not select the original mongos
       # because that server's SDAM state is reset by the connection error,
-      # heartbeatFrequencyMS is high, and subsequent isMaster heartbeats
+      # heartbeatFrequencyMS is high, and subsequent heartbeats
       # should fail.
       - name: commitTransaction
         object: session0

data/spec/spec_tests/data/transactions_unified/mongos-unpin.yml

@@ -0,0 +1,172 @@
+description: mongos-unpin
+
+schemaVersion: '1.4'
+
+runOnRequirements:
+  - minServerVersion: '4.2'
+    topologies: [ sharded-replicaset ]
+
+createEntities:
+  - client:
+      id: &client0 client0
+      useMultipleMongoses: true
+  - database:
+      id: &database0 database0
+      client: *client0
+      databaseName: &database0Name mongos-unpin-db
+  - collection:
+      id: &collection0 collection0
+      database: *database0
+      collectionName: &collection0Name test
+  - session:
+      id: &session0 session0
+      client: *client0
+
+initialData:
+  - collectionName: *collection0Name
+    databaseName: *database0Name
+    documents: []
+
+_yamlAnchors:
+  anchors:
+    # LockTimeout will cause the server to add a TransientTransactionError label. It is not retryable.
+    &lockTimeoutErrorCode 24
+
+tests:
+  - description: unpin after TransientTransactionError error on commit
+    runOnRequirements:
+        # serverless proxy doesn't append error labels to errors in transactions
+        # caused by failpoints (CLOUDP-88216)
+      - serverless: "forbid"
+    operations:
+      - &startTransaction
+        name: startTransaction
+        object: *session0
+      - &insertOne
+        name: insertOne
+        object: *collection0
+        arguments:
+          document: { x: 1 }
+          session: *session0
+      - name: targetedFailPoint
+        object: testRunner
+        arguments:
+          session: *session0
+          failPoint:
+            configureFailPoint: failCommand
+            mode: { times: 1 }
+            data:
+              failCommands: [ commitTransaction ]
+              errorCode: *lockTimeoutErrorCode
+      - name: commitTransaction
+        object: *session0
+        expectError:
+          # LockTimeout is not retryable, so the commit fails.
+          errorCode: *lockTimeoutErrorCode
+          errorLabelsContain: [ TransientTransactionError ]
+          errorLabelsOmit: [ UnknownTransactionCommitResult ]
+      - &assertNoPinnedServer
+        name: assertSessionUnpinned
+        object: testRunner
+        arguments:
+          session: *session0
+      # Cleanup the potentionally open server transaction by starting and
+      # aborting a new transaction on the same session.
+      - *startTransaction
+      - *insertOne
+      - &abortTransaction
+        name: abortTransaction
+        object: *session0
+
+  - description: unpin on successful abort
+    operations:
+      - *startTransaction
+      - *insertOne
+      - *abortTransaction
+      - *assertNoPinnedServer
+
+  - description: unpin after non-transient error on abort
+    runOnRequirements:
+        # serverless proxy doesn't append error labels to errors in transactions
+        # caused by failpoints (CLOUDP-88216)
+      - serverless: "forbid"
+    operations:
+      - *startTransaction
+      - *insertOne
+      - name: targetedFailPoint
+        object: testRunner
+        arguments:
+          session: *session0
+          failPoint:
+            configureFailPoint: failCommand
+            mode: { times: 1 }
+            data:
+              failCommands: [ abortTransaction ]
+              errorCode: *lockTimeoutErrorCode
+      - *abortTransaction
+      - *assertNoPinnedServer
+      # Cleanup the potentionally open server transaction by starting and
+      # aborting a new transaction on the same session.
+      - *startTransaction
+      - *insertOne
+      - *abortTransaction
+
+  - description: unpin after TransientTransactionError error on abort
+    operations:
+      - *startTransaction
+      - *insertOne
+      - name: targetedFailPoint
+        object: testRunner
+        arguments:
+          session: *session0
+          failPoint:
+            configureFailPoint: failCommand
+            mode: { times: 1 }
+            data:
+              failCommands: [ abortTransaction ]
+              errorCode: 91 # ShutdownInProgress
+      - *abortTransaction
+      - *assertNoPinnedServer
+      # Cleanup the potentionally open server transaction by starting and
+      # aborting a new transaction on the same session.
+      - *startTransaction
+      - *insertOne
+      - *abortTransaction
+
+  - description: unpin when a new transaction is started
+    operations:
+      - *startTransaction
+      - *insertOne
+      - name: commitTransaction
+        object: *session0
+      - &assertPinnedServer
+        name: assertSessionPinned
+        object: testRunner
+        arguments:
+          session: *session0
+      - *startTransaction
+      - *assertNoPinnedServer
+
+  - description: unpin when a non-transaction write operation uses a session
+    operations:
+      - *startTransaction
+      - *insertOne
+      - name: commitTransaction
+        object: *session0
+      - *assertPinnedServer
+      - *insertOne
+      - *assertNoPinnedServer
+
+  - description: unpin when a non-transaction read operation uses a session
+    operations:
+      - *startTransaction
+      - *insertOne
+      - name: commitTransaction
+        object: *session0
+      - *assertPinnedServer
+      - name: find
+        object: *collection0
+        arguments:
+          filter: { x: 1 }
+          session: *session0
+      - *assertNoPinnedServer