From 11d67aff9c835d6aaabe26e8b2371465dae9f403 Mon Sep 17 00:00:00 2001
From: Nitin Garg <gargnitin@google.com>
Date: Thu, 3 Oct 2024 20:22:13 +0000
Subject: [PATCH] Add try-catch for exception during parsing

Adds row with "ERROR" for all values
rather than crashing during printing in
CSV file.
---
 .../examples/dlio/parse_logs.py               | 53 +++++++++++--------
 .../testing_on_gke/examples/fio/parse_logs.py | 53 +++++++++++--------
 2 files changed, 60 insertions(+), 46 deletions(-)

diff --git a/perfmetrics/scripts/testing_on_gke/examples/dlio/parse_logs.py b/perfmetrics/scripts/testing_on_gke/examples/dlio/parse_logs.py
index 58f76223cd..34e8d5df5b 100644
--- a/perfmetrics/scripts/testing_on_gke/examples/dlio/parse_logs.py
+++ b/perfmetrics/scripts/testing_on_gke/examples/dlio/parse_logs.py
@@ -288,29 +288,36 @@ def writeOutput(
           )
           continue
 
-        new_row = (
-            record_set["mean_file_size"],
-            record_set["num_files_train"],
-            total_size,
-            record_set["batch_size"],
-            scenario,
-            r["epoch"],
-            r["duration"],
-            r["train_au_percentage"],
-            r["train_throughput_samples_per_second"],
-            r["train_throughput_mb_per_second"],
-            r["throughput_over_local_ssd"],
-            r["lowest_memory"],
-            r["highest_memory"],
-            r["lowest_cpu"],
-            r["highest_cpu"],
-            r["pod_name"],
-            r["start"],
-            r["end"],
-            f'"{r["gcsfuse_mount_options"].strip()}"',  # need to wrap in quotes to encapsulate commas in the value.
-            args.instance_id,
-        )
-        rows.append(new_row)
+        try:
+          new_row = (
+              record_set["mean_file_size"],
+              record_set["num_files_train"],
+              total_size,
+              record_set["batch_size"],
+              scenario,
+              r["epoch"],
+              r["duration"],
+              r["train_au_percentage"],
+              r["train_throughput_samples_per_second"],
+              r["train_throughput_mb_per_second"],
+              r["throughput_over_local_ssd"],
+              r["lowest_memory"],
+              r["highest_memory"],
+              r["lowest_cpu"],
+              r["highest_cpu"],
+              r["pod_name"],
+              r["start"],
+              r["end"],
+              f'"{r["gcsfuse_mount_options"].strip()}"',  # need to wrap in quotes to encapsulate commas in the value.
+              args.instance_id,
+          )
+          rows.append(new_row)
+        except Exception as e:
+          print(
+              f"Error while creating new output row for key={key},"
+              f" scenario={scenario}, epoch={i}, r={r}: {e}"
+          )
+          rows.append((("ERROR",) * len(_HEADER)))
 
   export_to_csv(output_file_path=args.output_file, header=_HEADER, rows=rows)
   export_to_gsheet(
diff --git a/perfmetrics/scripts/testing_on_gke/examples/fio/parse_logs.py b/perfmetrics/scripts/testing_on_gke/examples/fio/parse_logs.py
index 8424a30adf..f4053fabb7 100644
--- a/perfmetrics/scripts/testing_on_gke/examples/fio/parse_logs.py
+++ b/perfmetrics/scripts/testing_on_gke/examples/fio/parse_logs.py
@@ -322,29 +322,36 @@ def writeOutput(
           )
           continue
 
-        new_row = (
-            record_set["mean_file_size"],
-            record_set["read_type"],
-            scenario,
-            r["epoch"],
-            r["duration"],
-            r["throughput_mb_per_second"],
-            r["IOPS"],
-            r["throughput_over_local_ssd"],
-            r["lowest_memory"],
-            r["highest_memory"],
-            r["lowest_cpu"],
-            r["highest_cpu"],
-            r["pod_name"],
-            r["start"],
-            r["end"],
-            f'"{r["gcsfuse_mount_options"].strip()}"',  # need to wrap in quotes to encapsulate commas in the value.
-            r["blockSize"],
-            r["filesPerThread"],
-            r["numThreads"],
-            args.instance_id,
-        )
-        rows.append(new_row)
+        try:
+          new_row = (
+              record_set["mean_file_size"],
+              record_set["read_type"],
+              scenario,
+              r["epoch"],
+              r["duration"],
+              r["throughput_mb_per_second"],
+              r["IOPS"],
+              r["throughput_over_local_ssd"],
+              r["lowest_memory"],
+              r["highest_memory"],
+              r["lowest_cpu"],
+              r["highest_cpu"],
+              r["pod_name"],
+              r["start"],
+              r["end"],
+              f'"{r["gcsfuse_mount_options"].strip()}"',  # need to wrap in quotes to encapsulate commas in the value.
+              r["blockSize"],
+              r["filesPerThread"],
+              r["numThreads"],
+              args.instance_id,
+          )
+          rows.append(new_row)
+        except Exception as e:
+          print(
+              f"Error while creating new output row for key={key},"
+              f" scenario={scenario}, epoch={i}, r={r}: {e}"
+          )
+          rows.append((("ERROR",) * len(_HEADER)))
 
   export_to_csv(output_file_path=args.output_file, header=_HEADER, rows=rows)
   export_to_gsheet(