refactor: rename sdk to workload in SLO workflows and metrics

polRk · polRk · commit 79aab75a55c9 · 2024-11-26T00:43:21.000+03:00
diff --git a/.github/workflows/slo-report.yml b/.github/workflows/slo-report.yml
@@ -18,5 +18,5 @@ jobs:
       - name: Publish YDB SLO Report
         uses: ydb-platform/ydb-slo-action/report@main
         with:
-          token: ${{ secrets.GITHUB_TOKEN }}
-          run_id: ${{ github.event.workflow_run.id }}
+          github_token: ${{ secrets.GITHUB_TOKEN }}
+          github_run_id: ${{ github.event.workflow_run.id }}
diff --git a/.github/workflows/slo.yml b/.github/workflows/slo.yml
@@ -30,12 +30,12 @@ jobs:
 
     strategy:
       matrix:
-        sdk:
-          - py-sync-table
-          - py-sync-query
+        workload:
+          - sync-table
+          - sync-query
 
     concurrency:
-      group: slo-${{ github.ref }}-${{ matrix.sdk }}
+      group: slo-${{ github.ref }}-${{ matrix.workload }}
       cancel-in-progress: true
 
     steps:
@@ -59,7 +59,7 @@ jobs:
         with:
           github_pull_request_number: ${{ github.event.inputs.github_pull_request_number }}
           github_token: ${{ secrets.GITHUB_TOKEN }}
-          sdk_name: ${{ matrix.sdk }}
+          workload_name: ${{ matrix.workload }}
           ydb_database_node_count: 5
 
       - name: Prepare SLO Database
@@ -69,7 +69,7 @@ jobs:
       - name: Run SLO Tests
         env:
           REF: '${{ github.head_ref || github.ref }}'
-          SDK_SERVICE: '${{ matrix.sdk }}'
+          WORKLOAD: '${{ matrix.workload }}'
         run: |
           python ./tests/slo/src run grpc://localhost:2135 /Root/testdb \
             --prom-pgw localhost:9091 \
@@ -78,7 +78,12 @@ jobs:
             --read-rps ${{inputs.slo_workload_read_max_rps || 1000}} \
             --write-rps ${{inputs.slo_workload_write_max_rps || 100}} \
             --read-timeout 1000 \
-            --write-timeout 1000 || true
+            --write-timeout 1000
+
+      - if: always()
+        name: Cleanup SLO Database
+        run: |
+          python ./tests/slo/src cleanup grpc://localhost:2135 /Root/testdb
 
       - if: always()
         name: Store ydb chaos testing logs
@@ -88,11 +93,6 @@ jobs:
       - if: always()
         uses: actions/upload-artifact@v4
         with:
-          name: ${{ matrix.sdk }}-chaos-ydb.log
+          name: ${{ matrix.workload }}-chaos-ydb.log
           path: ./chaos-ydb.log
           retention-days: 1
-
-      - if: always()
-        name: Cleanup SLO Database
-        run: |
-          python ./tests/slo/src cleanup grpc://localhost:2135 /Root/testdb || true
diff --git a/tests/slo/src/metrics.py b/tests/slo/src/metrics.py
@@ -13,7 +13,7 @@
 OP_STATUS_SUCCESS, OP_STATUS_FAILURE = "success", "err"
 
 REF = environ.get("REF", "main")
-SDK_SERVICE_NAME = environ.get("SDK_SERVICE", "py-sync-table")
+WORKLOAD = environ.get("WORKLOAD", "sync-query")
 
 
 class Metrics:
@@ -75,24 +75,24 @@ def __init__(self, push_gateway):
                 labelnames=("operation_type",),
                 registry=self._registry,
             ),
-            retry_attempts_total=Counter(
-                "sdk_retry_attempts_total",
-                "Total number of retry attempts, categorized by operation type.",
-                labelnames=("operation_type",),
-                registry=self._registry,
-            ),
-            retries_success_total=Counter(
-                "sdk_retries_success_total",
-                "Total number of successful retries, categorized by operation type.",
-                labelnames=("operation_type",),
-                registry=self._registry,
-            ),
-            retries_failure_total=Counter(
-                "sdk_retries_failure_total",
-                "Total number of failed retries, categorized by operation type.",
-                labelnames=("operation_type",),
-                registry=self._registry,
-            ),
+            # retry_attempts_total=Counter(
+            #     "sdk_retry_attempts_total",
+            #     "Total number of retry attempts, categorized by operation type.",
+            #     labelnames=("operation_type",),
+            #     registry=self._registry,
+            # ),
+            # retries_success_total=Counter(
+            #     "sdk_retries_success_total",
+            #     "Total number of successful retries, categorized by operation type.",
+            #     labelnames=("operation_type",),
+            #     registry=self._registry,
+            # ),
+            # retries_failure_total=Counter(
+            #     "sdk_retries_failure_total",
+            #     "Total number of failed retries, categorized by operation type.",
+            #     labelnames=("operation_type",),
+            #     registry=self._registry,
+            # ),
             pending_operations=Gauge(
                 "sdk_pending_operations",
                 "Current number of pending operations, categorized by type.",
@@ -135,28 +135,30 @@ def stop(self, labels, start_time, attempts=1, error=None):
         self.retry_attempts.labels(*labels).set(attempts)
         self.operations_total.labels(*labels).inc()
         self.pending_operations.labels(*labels).dec()
-        self.retry_attempts_total.labels(*labels).inc(attempts)
+        # self.retry_attempts_total.labels(*labels).inc(attempts)
 
         if error:
             self.errors_total.labels(*labels, type(error).__name__).inc()
-            self.retries_failure_total.labels(*labels).inc(attempts)
+            # self.retries_failure_total.labels(*labels).inc(attempts)
             self.operations_failure_total.labels(*labels).inc()
             self.operation_latency_seconds.labels(*labels, OP_STATUS_FAILURE).observe(duration)
             return
 
-        self.retries_success_total.labels(*labels).inc(attempts)
+        # self.retries_success_total.labels(*labels).inc(attempts)
         self.operations_success_total.labels(*labels).inc()
         self.operation_latency_seconds.labels(*labels, OP_STATUS_SUCCESS).observe(duration)
 
     def push(self):
         push_to_gateway(
             self._push_gtw,
-            job=f"workload-{SDK_SERVICE_NAME}",
+            job=f"workload-{WORKLOAD}",
             registry=self._registry,
             grouping_key={
                 "ref": REF,
-                "sdk": SDK_SERVICE_NAME,
+                "sdk": "ydb-python-sdk",
                 "sdk_version": version("ydb"),
+                "workload": WORKLOAD,
+                "workload_version": "0.0.0",
             },
         )
 
diff --git a/tests/slo/src/runner.py b/tests/slo/src/runner.py
@@ -14,7 +14,7 @@
     run_write_jobs_query,
     run_metric_job,
 )
-from metrics import Metrics, SDK_SERVICE_NAME
+from metrics import Metrics, WORKLOAD
 
 logger = logging.getLogger(__name__)
 
@@ -91,20 +91,20 @@ def run_slo(args, driver, tb_name):
     logger.info("Max ID: %s", max_id)
 
     metrics = Metrics(args.prom_pgw)
-    if SDK_SERVICE_NAME == "py-sync-table":
+    if WORKLOAD == "sync-table":
         futures = (
             *run_read_jobs(args, driver, tb_name, max_id, metrics),
             *run_write_jobs(args, driver, tb_name, max_id, metrics),
             run_metric_job(args, metrics),
         )
-    elif SDK_SERVICE_NAME == "py-sync-query":
+    elif WORKLOAD == "sync-query":
         futures = (
             *run_read_jobs_query(args, driver, tb_name, max_id, metrics),
             *run_write_jobs_query(args, driver, tb_name, max_id, metrics),
             run_metric_job(args, metrics),
         )
     else:
-        raise ValueError(f"Unsupported service: {SDK_SERVICE_NAME}")
+        raise ValueError(f"Unsupported service: {WORKLOAD}")
 
     for future in futures:
         future.join()
