parallel-workload: Introduce bad queries

Author: def-

ci/nightly/pipeline.template.yml

@@ -1903,8 +1903,20 @@ steps:
   - group: "Parallel Workload"
     key: parallel-workload
     steps:
+      - id: parallel-workload-bad-queries
+        label: "Parallel Workload + Bad Queries"
+        depends_on: build-x86_64
+        artifact_paths: [parallel-workload-queries.log.zst]
+        timeout_in_minutes: 90
+        agents:
+          queue: hetzner-x86-64-8cpu-16gb
+        plugins:
+          - ./ci/plugins/mzcompose:
+              composition: parallel-workload
+              args: [--runtime=1500, --complexity=bad, --threads=8]
+
       - id: parallel-workload-read-only
-        label: "Parallel Workload (Read-Only)"
+        label: "Parallel Workload + Read-Only"
         depends_on: build-x86_64
         artifact_paths: [parallel-workload-queries.log.zst]
         timeout_in_minutes: 90
@@ -1916,7 +1928,7 @@ steps:
               args: [--runtime=1500, --complexity=read, --threads=8]
 
       - id: parallel-workload-dml
-        label: "Parallel Workload (DML)"
+        label: "Parallel Workload + DML"
         depends_on: build-x86_64
         artifact_paths: [parallel-workload-queries.log.zst]
         timeout_in_minutes: 90
@@ -1928,7 +1940,7 @@ steps:
               args: [--runtime=1500, --complexity=dml, --threads=8]
 
       - id: parallel-workload-ddl
-        label: "Parallel Workload (DDL)"
+        label: "Parallel Workload + DDL"
         depends_on: build-x86_64
         artifact_paths: [parallel-workload-queries.log.zst]
         timeout_in_minutes: 90
@@ -1940,7 +1952,7 @@ steps:
               args: [--runtime=1500, --threads=8]
 
       - id: parallel-workload-ddl-only
-        label: "Parallel Workload (DDL Only)"
+        label: "Parallel Workload + DDL Only"
         depends_on: build-x86_64
         artifact_paths: [parallel-workload-queries.log.zst]
         timeout_in_minutes: 90
@@ -1952,7 +1964,7 @@ steps:
               args: [--runtime=1500, --complexity=ddl-only, --threads=8]
 
       - id: parallel-workload-many-threads
-        label: "Parallel Workload (many threads)"
+        label: "Parallel Workload + many threads"
         depends_on: build-x86_64
         artifact_paths: [parallel-workload-queries.log.zst]
         timeout_in_minutes: 90
@@ -1966,7 +1978,7 @@ steps:
         skip: "Too unstable at the moment"
 
       - id: parallel-workload-rename-naughty
-        label: "Parallel Workload (rename + naughty identifiers)"
+        label: "Parallel Workload + rename + naughty identifiers"
         depends_on: build-x86_64
         artifact_paths: [parallel-workload-queries.log.zst]
         timeout_in_minutes: 90
@@ -1978,7 +1990,7 @@ steps:
               args: [--runtime=1500, --scenario=rename, --naughty-identifiers, --threads=8]
 
       - id: parallel-workload-rename
-        label: "Parallel Workload (rename)"
+        label: "Parallel Workload + rename"
         depends_on: build-x86_64
         artifact_paths: [parallel-workload-queries.log.zst]
         timeout_in_minutes: 90
@@ -1990,7 +2002,7 @@ steps:
               args: [--runtime=1500, --scenario=rename, --threads=8]
 
       - id: parallel-workload-cancel
-        label: "Parallel Workload (cancel)"
+        label: "Parallel Workload + cancel"
         depends_on: build-x86_64
         artifact_paths: [parallel-workload-queries.log.zst]
         timeout_in_minutes: 90
@@ -2002,7 +2014,7 @@ steps:
               args: [--runtime=1500, --scenario=cancel, --threads=8]
 
       - id: parallel-workload-kill
-        label: "Parallel Workload (kill)"
+        label: "Parallel Workload + kill"
         depends_on: build-x86_64
         artifact_paths: [parallel-workload-queries.log.zst]
         timeout_in_minutes: 90
@@ -2014,7 +2026,7 @@ steps:
               args: [--runtime=1500, --scenario=kill, --threads=8]
 
       - id: parallel-workload-backup-restore
-        label: "Parallel Workload (backup & restore)"
+        label: "Parallel Workload + backup & restore"
         depends_on: build-x86_64
         artifact_paths: [parallel-workload-queries.log.zst]
         timeout_in_minutes: 90
@@ -2027,7 +2039,7 @@ steps:
               args: [--runtime=1500, --scenario=backup-restore, --naughty-identifiers, --threads=8]
 
       - id: parallel-workload-0dt
-        label: "Parallel Workload (0dt deploy)"
+        label: "Parallel Workload + 0dt deploy"
         depends_on: build-x86_64
         artifact_paths: [parallel-workload-queries.log.zst]
         timeout_in_minutes: 90

misc/python/materialize/parallel_workload/action.py

@@ -396,6 +396,31 @@ def run(self, exe: Executor) -> bool:
         return True
 
 
+class BadAction(Action):
+    keywords: list[str]
+    objects: list[str]
+    def __init__(self, rng: random.Random, composition: Composition | None):
+        super().__init__(rng, composition)
+        self.keywords = []
+        with open("src/sql-lexer/src/keywords.txt", "r") as f:
+            for line in f:
+                line = line.strip()
+                if not line or line.startswith("#"):
+                    continue
+                self.keywords.append(line.upper())
+        self.objects = []
+
+    def errors_to_ignore(self, exe: Executor) -> list[str]:
+        # TODO: Ignore all errors except internal errors
+        result = super().errors_to_ignore(exe)
+        return result
+
+    def run(self, exe: Executor) -> bool:
+        
+        exe.execute(query, explainable=True, http=Http.RANDOM, fetch=True)
+        return True
+
+
 class SelectOneAction(Action):
     def run(self, exe: Executor) -> bool:
         exe.execute("SELECT 1", explainable=True, http=Http.RANDOM, fetch=True)
@@ -2781,10 +2806,19 @@ def __init__(
     autocommit=True,
 )
 
+# TODO: Should we have some "bad" actions in all scenarios or keep it isolated in one?
+bad_action_list = ActionList(
+    [
+        (BadAction, 100),
+    ],
+    autocommit=True,
+)
+
 action_lists = [
     read_action_list,
     fetch_action_list,
     write_action_list,
     dml_nontrans_action_list,
     ddl_action_list,
+    bad_action_list,
 ]

misc/python/materialize/parallel_workload/parallel_workload.py

@@ -32,6 +32,7 @@
     ZeroDowntimeDeployAction,
     action_lists,
     ddl_action_list,
+    bad_action_list,
     dml_nontrans_action_list,
     fetch_action_list,
     read_action_list,
@@ -166,13 +167,15 @@ def run(
     for i in range(num_threads):
         weights: list[float]
         if complexity == Complexity.DDL:
-            weights = [60, 30, 30, 30, 100]
+            weights = [60, 30, 30, 30, 100, 0]
         elif complexity == Complexity.DML:
-            weights = [60, 30, 30, 30, 0]
+            weights = [60, 30, 30, 30, 0, 0]
         elif complexity == Complexity.Read:
-            weights = [60, 30, 0, 0, 0]
+            weights = [60, 30, 0, 0, 0, 0]
         elif complexity == Complexity.DDLOnly:
-            weights = [0, 0, 0, 0, 100]
+            weights = [0, 0, 0, 0, 100, 0]
+        elif complexity == Complexity.Bad:
+            weights = [0, 0, 0, 0, 0, 100]
         else:
             raise ValueError(f"Unknown complexity {complexity}")
         worker_rng = random.Random(rng.randrange(SEED_RANGE))
@@ -183,6 +186,7 @@ def run(
                 write_action_list,
                 dml_nontrans_action_list,
                 ddl_action_list,
+                bad_action_list,
             ],
             weights,
         )[0]

misc/python/materialize/parallel_workload/settings.py

@@ -12,6 +12,7 @@
 
 
 class Complexity(Enum):
+    Bad = "bad"
     Read = "read"
     DML = "dml"
     DDL = "ddl"