feat: measuring compute efficiency per job

.github/workflows/announce-release.yml

@@ -16,7 +16,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Post to Mastodon
-        uses: snakemake/mastodon-release-post-action@v1.4.0
+        uses: snakemake/mastodon-release-post-action@v1.4.1
         with:
           access-token: ${{ secrets.MASTODONBOT }}
           pr-title: ${{ github.event.head_commit.message }}

docs/further.md

@@ -618,6 +618,10 @@ This configuration directs SLURM logs to a centralized location, making them eas
 Running Snakemake within an active SLURM job can lead to unpredictable behavior, as the execution environment may not be properly configured for job submission.
 To mitigate potential issues, the SLURM executor plugin detects when it's operating inside a SLURM job and issues a warning, pausing for 5 seconds before proceeding.
 
+### Getting Job Efficiency Information
+
+With `--slurm-efficiency-report` you can generate a table of all efficiency data. A logfile `efficiency_report_<workflow_id>.log` will be generated in your current directory. This is equivalent to the information with `seff <jobid>` for individual jobs. It works best if "comments" are stored as a job property on your cluster as this plugin uses the "comment" parameter to store the rule name.
+
 ### Frequently Asked Questions
 
 #### Should I run Snakemake on the Login Node of my Cluster?

pyproject.toml

@@ -26,6 +26,7 @@ flake8 = "^6.1.0"
 coverage = "^7.3.1"
 pytest = "^8.3.5"
 snakemake = "^9.4.0"
+pandas = "^2.2.3"
 
 [tool.coverage.run]
 omit = [".*", "*/site-packages/*", "Snakefile"]

snakemake_executor_plugin_slurm/__init__.py

@@ -3,7 +3,6 @@
 __email__ = "johannes.koester@uni-due.de"
 __license__ = "MIT"
 
-import atexit
 import csv
 from io import StringIO
 import os
@@ -16,6 +15,9 @@
 from datetime import datetime, timedelta
 from typing import List, Generator, Optional
 import uuid
+
+import pandas as pd
+
 from snakemake_interface_executor_plugins.executors.base import SubmittedJobInfo
 from snakemake_interface_executor_plugins.executors.remote import RemoteExecutor
 from snakemake_interface_executor_plugins.settings import (
@@ -27,7 +29,12 @@
 )
 from snakemake_interface_common.exceptions import WorkflowError
 
-from .utils import delete_slurm_environment, delete_empty_dirs, set_gres_string
+from .utils import (
+    delete_slurm_environment,
+    delete_empty_dirs,
+    set_gres_string,
+)
+from .efficiency_report import time_to_seconds, parse_maxrss, parse_reqmem
 from .submit_string import get_submit_command
 
 
@@ -106,6 +113,15 @@ class ExecutorSettings(ExecutorSettingsBase):
             "required": False,
         },
     )
+    efficiency_report: bool = field(
+        default=False,
+        metadata={
+            "help": "Generate an efficiency report at the end of the workflow. "
+            "This flag has no effect, if not set.",
+            "env_var": False,
+            "required": False,
+        },
+    )
 
 
 # Required:
@@ -149,7 +165,21 @@ def __post_init__(self, test_mode: bool = False):
             if self.workflow.executor_settings.logdir
             else Path(".snakemake/slurm_logs").resolve()
         )
-        atexit.register(self.clean_old_logs)
+
+    def shutdown(self) -> None:
+        """
+        Shutdown the executor.
+        This method is overloaded, to include the cleaning of old log files
+        and to optionally create an efficiency report.
+        """
+        # First, we invoke the original shutdown method
+        super().shutdown()
+
+        # Next, clean up old log files, unconditionally.
+        self.clean_old_logs()
+        # If the efficiency report is enabled, create it.
+        if self.workflow.executor_settings.efficiency_report:
+            self.create_efficiency_report()
 
     def clean_old_logs(self) -> None:
         """Delete files older than specified age from the SLURM log directory."""
@@ -160,20 +190,23 @@ def clean_old_logs(self) -> None:
             return
         cutoff_secs = age_cutoff * 86400
         current_time = time.time()
-        self.logger.info(f"Cleaning up log files older than {age_cutoff} day(s)")
+        self.logger.info(f"Cleaning up log files older than {age_cutoff} day(s).")
+
         for path in self.slurm_logdir.rglob("*.log"):
             if path.is_file():
                 try:
                     file_age = current_time - path.stat().st_mtime
                     if file_age > cutoff_secs:
                         path.unlink()
                 except (OSError, FileNotFoundError) as e:
-                    self.logger.warning(f"Could not delete logfile {path}: {e}")
+                    self.logger.error(f"Could not delete logfile {path}: {e}")
         # we need a 2nd iteration to remove putatively empty directories
         try:
             delete_empty_dirs(self.slurm_logdir)
         except (OSError, FileNotFoundError) as e:
-            self.logger.warning(f"Could not delete empty directory {path}: {e}")
+            self.logger.error(
+                f"Could not delete empty directories in {self.slurm_logdir}: {e}"
+            )
 
     def warn_on_jobcontext(self, done=None):
         if not done:
@@ -730,10 +763,134 @@ def check_slurm_extra(self, job):
         jobname = re.compile(r"--job-name[=?|\s+]|-J\s?")
         if re.search(jobname, job.resources.slurm_extra):
             raise WorkflowError(
-                "The --job-name option is not allowed in the 'slurm_extra' "
-                "parameter. The job name is set by snakemake and must not be "
-                "overwritten. It is internally used to check the stati of the "
-                "all submitted jobs by this workflow."
+                "The --job-name option is not allowed in the 'slurm_extra' parameter. "
+                "The job name is set by snakemake and must not be overwritten. "
+                "It is internally used to check the stati of the all submitted jobs "
+                "by this workflow."
                 "Please consult the documentation if you are unsure how to "
                 "query the status of your jobs."
             )
+
+    def create_efficiency_report(self, efficiency_threshold=0.8):
+        """
+        Fetch sacct job data for a Snakemake workflow
+        and compute efficiency metrics.
+        """
+        cmd = f"sacct --name={self.run_uuid} --parsable2 --noheader"
+        cmd += (
+            " --format=JobID,JobName,Comment,Elapsed,TotalCPU,"
+            "NNodes,NCPUS,MaxRSS,ReqMem"
+        )
+
+        try:
+            result = subprocess.run(
+                shlex.split(cmd), capture_output=True, text=True, check=True
+            )
+            lines = result.stdout.strip().split("\n")
+        except subprocess.CalledProcessError:
+            self.logger.error(
+                f"Failed to retrieve job data for workflow {self.run_uuid}."
+            )
+            return None
+
+        # Convert to DataFrame
+        df = pd.DataFrame(
+            (line.split("|") for line in lines),
+            columns=[
+                "JobID",
+                "JobName",
+                "Comment",
+                "Elapsed",
+                "TotalCPU",
+                "NNodes",
+                "NCPUS",
+                "MaxRSS",
+                "ReqMem",
+            ],
+        )
+
+        # If the "Comment" column is empty,
+        # a) delete the column
+        # b) issue a warning
+        if df["Comment"].isnull().all():
+            self.logger.warning(
+                f"No comments found for workflow {self.run_uuid}. "
+                "This field is used to store the rule name. "
+                "Please ensure that the 'comment' field is set for your cluster. "
+                "Administrators can set this up in the SLURM configuration."
+            )
+            df.drop(columns=["Comment"], inplace=True)
+            # remember, that the comment column is not available
+            nocomment = True
+        # else: rename the column to 'RuleName'
+        else:
+            df.rename(columns={"Comment": "RuleName"}, inplace=True)
+            nocomment = False
+        # Convert types
+        df["NNodes"] = pd.to_numeric(df["NNodes"], errors="coerce")
+        df["NCPUS"] = pd.to_numeric(df["NCPUS"], errors="coerce")
+
+        # Convert time fields
+        df["Elapsed_sec"] = df["Elapsed"].apply(time_to_seconds)
+        df["TotalCPU_sec"] = df["TotalCPU"].apply(time_to_seconds)
+
+        # Compute CPU efficiency
+        df["CPU Efficiency (%)"] = (
+            df["TotalCPU_sec"] / (df["Elapsed_sec"] * df["NCPUS"])
+        ) * 100
+        df["CPU Efficiency (%)"] = df["CPU Efficiency (%)"].fillna(0).round(2)
+
+        # Convert MaxRSS
+        df["MaxRSS_MB"] = df["MaxRSS"].apply(parse_maxrss)
+
+        # Convert ReqMem and calculate memory efficiency
+        df["RequestedMem_MB"] = df.apply(
+            lambda row: parse_reqmem(row["ReqMem"], row["NNodes"]), axis=1
+        )
+        df["Memory Usage (%)"] = df.apply(
+            lambda row: (
+                (row["MaxRSS_MB"] / row["RequestedMem_MB"] * 100)
+                if row["RequestedMem_MB"] > 0
+                else 0
+            ),
+            axis=1,
+        )
+
+        df["Memory Usage (%)"] = df["Memory Usage (%)"].fillna(0).round(2)
+
+        # Drop all rows containing "batch" or "extern" as job names
+        df = df[~df["JobName"].str.contains("batch|extern")]
+
+        # Log warnings for low efficiency
+        for _, row in df.iterrows():
+            if row["CPU Efficiency (%)"] < efficiency_threshold:
+                if nocomment:
+                    self.logger.warning(
+                        f"Job {row['JobID']} ({row['JobName']}) "
+                        f"has low CPU efficiency: {row['CPU Efficiency (%)']}%."
+                    )
+                else:
+                    # if the comment column is available, we can use it to
+                    # identify the rule name
+                    self.logger.warning(
+                        f"Job {row['JobID']} for rule '{row['RuleName']}' "
+                        f"({row['JobName']}) has low CPU efficiency: "
+                        f"{row['CPU Efficiency (%)']}%."
+                    )
+
+        # we construct a path object to allow for a customi
+        # logdir, if specified
+        p = Path()
+
+        # Save the report to a CSV file
+        logfile = f"efficiency_report_{self.run_uuid}.csv"
+        if self.workflow.executor_settings.logdir:
+            logfile = Path(self.workflow.executor_settings.logdir) / logfile
+        else:
+            logfile = p.cwd() / logfile
+        df.to_csv(logfile)
+
+        # write out the efficiency report at normal verbosity in any case
+        self.logger.info(
+            f"Efficiency report for workflow {self.run_uuid} saved to {logfile}."
+        )

snakemake_executor_plugin_slurm/efficiency_report.py

@@ -0,0 +1,50 @@
+import re
+import pandas as pd
+
+
+def time_to_seconds(time_str):
+    """Convert SLURM time format to seconds."""
+    if pd.isna(time_str) or time_str.strip() == "":
+        return 0
+    parts = time_str.split(":")
+
+    if len(parts) == 3:  # H:M:S
+        return int(parts[0]) * 3600 + int(parts[1]) * 60 + float(parts[2])
+    elif len(parts) == 2:  # M:S
+        return int(parts[0]) * 60 + float(parts[1])
+    elif len(parts) == 1:  # S
+        return float(parts[0])
+    return 0
+
+
+def parse_maxrss(maxrss):
+    """Convert MaxRSS to MB."""
+    if pd.isna(maxrss) or maxrss.strip() == "" or maxrss == "0":
+        return 0
+    match = re.match(r"(\d+)([KMG]?)", maxrss)
+    if match:
+        value, unit = match.groups()
+        value = int(value)
+        unit_multipliers = {"K": 1 / 1024, "M": 1, "G": 1024}
+        return value * unit_multipliers.get(unit, 1)
+    return 0
+
+
+def parse_reqmem(reqmem, number_of_nodes=1):
+    """Convert requested memory to MB."""
+    if pd.isna(reqmem) or reqmem.strip() == "":
+        return 0
+    match = re.match(
+        r"(\d+)([KMG])?(\S+)?", reqmem
+    )  # Handles "4000M" or "4G" or "2G/node"
+    if match:
+        value, unit, per_unit = match.groups()
+        value = int(value)
+        unit_multipliers = {"K": 1 / 1024, "M": 1, "G": 1024}
+        mem_mb = value * unit_multipliers.get(unit, 1)
+        if per_unit and "/node" in per_unit:
+            # the memory values is per node, hence we need to
+            # multiply with the number of nodes
+            return mem_mb * number_of_nodes
+        return mem_mb  # Default case (per CPU or total)
+    return 0

tests/tests.py

@@ -1,3 +1,5 @@
+import os
+import re
 from typing import Optional
 import snakemake.common.tests
 from snakemake_interface_executor_plugins.settings import ExecutorSettingsBase
@@ -17,7 +19,43 @@ def get_executor(self) -> str:
         return "slurm"
 
     def get_executor_settings(self) -> Optional[ExecutorSettingsBase]:
-        return ExecutorSettings(init_seconds_before_status_checks=1)
+        return ExecutorSettings(
+            init_seconds_before_status_checks=2,
+            # seconds_between_status_checks=5,
+        )
+
+
+class TestEfficiencyReport(snakemake.common.tests.TestWorkflowsLocalStorageBase):
+    __test__ = True
+
+    def get_executor(self) -> str:
+        return "slurm"
+
+    def get_executor_settings(self) -> Optional[ExecutorSettingsBase]:
+        return ExecutorSettings(
+            efficiency_report=True,
+            init_seconds_before_status_checks=5,
+            # seconds_between_status_checks=5,
+        )
+
+    def test_simple_workflow(self, tmp_path):
+        self.run_workflow("simple", tmp_path)
+
+        # The efficiency report is created in the
+        # current working directory
+        pattern = re.compile(r"efficiency_report_[\w-]+\.csv")
+        report_found = False
+
+        for fname in os.listdir():
+            if pattern.match(fname):
+                report_found = True
+                report_path = tmp_path / fname
+                # Verify it's not empty
+                assert (
+                    report_path.stat().st_size > 0
+                ), f"Efficiency report {report_path} is empty"
+                break
+        assert report_found, "Efficiency report file not found"
 
 
 class TestWorkflowsRequeue(TestWorkflows):