Make setting prometheus query interval minute configurable

We have some clusters where we would like to change the frequency at
which we collect metrics and generate the report based on that.

To accomplish this, we need to make 2 changes. We set an environment
variable that is passed on to the metrics collector scripts. And we
insert the collection interval into the json file.

The merge scripts will read this interval when loading these files. For
backwards compatibility with our old metrics file, we'll default to 15
minutes when processing these files.

Author: naved001

openshift_metrics/config.py

@@ -14,3 +14,5 @@
 S3_SECRET_ACCESS_KEY = os.getenv("S3_OUTPUT_SECRET_ACCESS_KEY")
 S3_INVOICE_BUCKET = os.getenv("S3_INVOICE_BUCKET", "nerc-invoicing")
 S3_METRICS_BUCKET = os.getenv("S3_METRICS_BUCKET", "openshift_metrics")
+PROM_QUERY_INTERVAL_MINUTES = int(os.getenv("PROM_QUERY_INTERVAL_MINUTES", 15))
+assert PROM_QUERY_INTERVAL_MINUTES >= 1, "Query interval must be at least 1 minute"

openshift_metrics/merge.py

@@ -2,6 +2,7 @@
 Merges metrics from files and produces reports by pod and by namespace
 """
 
+import sys
 import logging
 import argparse
 from datetime import datetime, UTC
@@ -12,7 +13,7 @@
 
 from openshift_metrics import utils, invoice
 from openshift_metrics.metrics_processor import MetricsProcessor
-from openshift_metrics.config import S3_INVOICE_BUCKET
+from openshift_metrics.config import S3_INVOICE_BUCKET, PROM_QUERY_INTERVAL_MINUTES
 
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
@@ -106,7 +107,26 @@ def main():
     report_start_date = None
     report_end_date = None
     cluster_name = None
-    processor = MetricsProcessor()
+    interval_minutes = None
+
+    for file in files:
+        with open(file, "r") as jsonfile:
+            metrics_from_file = json.load(jsonfile)
+            if interval_minutes is None:
+                interval_minutes = metrics_from_file.get("interval_minutes")
+            else:
+                interval_minutes_from_file = metrics_from_file["interval_minutes"]
+                if interval_minutes != interval_minutes_from_file:
+                    sys.exit(f"Prometheus interval minutes differ in the given set of files {interval_minutes} != {interval_minutes_from_file}")
+
+    if interval_minutes is None:
+        logger.info(f"No prometheus query interval minutes found in the given set of files. Using the provided interval: {PROM_QUERY_INTERVAL_MINUTES} minute(s)")
+        interval_minutes = PROM_QUERY_INTERVAL_MINUTES
+    else:
+        logger.info(f"Prometheus Query interval set to {interval_minutes} minute(s) from file")
+
+
+    processor = MetricsProcessor(interval_minutes)
 
     for file in files:
         with open(file, "r") as jsonfile:

openshift_metrics/openshift_prometheus_metrics.py

@@ -26,6 +26,7 @@
     OPENSHIFT_PROMETHEUS_URL,
     OPENSHIFT_TOKEN,
     S3_METRICS_BUCKET,
+    PROM_QUERY_INTERVAL_MINUTES,
 )
 
 logging.basicConfig(level=logging.INFO)
@@ -89,14 +90,15 @@ def main():
         output_file = f"metrics-{report_start_date}-to-{report_end_date}.json"
 
     logger.info(
-        f"Generating report starting {report_start_date} and ending {report_end_date} in {output_file}"
+        f"Generating report starting {report_start_date} and ending {report_end_date} in {output_file} with interval {PROM_QUERY_INTERVAL_MINUTES} minute"
     )
 
-    prom_client = PrometheusClient(openshift_url, OPENSHIFT_TOKEN)
+    prom_client = PrometheusClient(openshift_url, OPENSHIFT_TOKEN, PROM_QUERY_INTERVAL_MINUTES)
 
     metrics_dict = {}
     metrics_dict["start_date"] = report_start_date
     metrics_dict["end_date"] = report_end_date
+    metrics_dict["interval_minutes"] = PROM_QUERY_INTERVAL_MINUTES
     metrics_dict["cluster_name"] = URL_CLUSTER_NAME_MAPPING.get(
         args.openshift_url, args.openshift_url
     )