From 3d5a06458f87486079e01284d8382bcf4a88685a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Ma=C3=ABl=20Madon?= <mael.madon@irit.fr>
Date: Thu, 17 Nov 2022 11:22:34 +0100
Subject: [PATCH] add option --session_stat to get a separate summary JSON with
session info
---
src/user_session_builder.py | 13 ++++++++++++-
src/workload.py | 11 +++++++++++
swf2userSessions.py | 20 ++++++++++++++++++--
3 files changed, 41 insertions(+), 3 deletions(-)
diff --git a/src/user_session_builder.py b/src/user_session_builder.py
index 0d3a320..f649473 100644
--- a/src/user_session_builder.py
+++ b/src/user_session_builder.py
@@ -155,7 +155,7 @@ class User:
def to_dict(self, descr=None, cmd=None, version=1):
- """User object to dictionnary, for printing or json export."""
+ """User object to dictionnary, for printing or SABjson export."""
if descr is None:
descr = f"Session Annotated Batsim JSON for user{self.id}"
@@ -171,3 +171,14 @@ class User:
res["command"] = cmd
return res
+
+
+ def to_session_stat(self):
+ """Return a dictionnary with session stats"""
+
+ return {
+ "first_submit_time": min([s.first_submit for _, s in self.sessions.items()]),
+ "max_finish_time": min([s.max_finish_time for _, s in self.sessions.items()]),
+ "nb_sessions": len(self.sessions),
+ "sessions": [s.to_session_stat() for _, s in self.sessions.items()]
+ }
\ No newline at end of file
diff --git a/src/workload.py b/src/workload.py
index 5536aa7..0135d2d 100644
--- a/src/workload.py
+++ b/src/workload.py
@@ -97,3 +97,14 @@ class Session:
"jobs": [j.to_dict(session_start_offset = self.first_submit)
for j in self.jobs]
}
+
+
+ def to_session_stat(self):
+ """Return a dictionnary with session stats"""
+
+ return {
+ "first_submit_time": self.first_submit,
+ "last_submit_time": max([j.submit_time for j in self.jobs]),
+ "finish_time": self.max_finish_time,
+ "nb_jobs": len(self.jobs)
+ }
\ No newline at end of file
diff --git a/swf2userSessions.py b/swf2userSessions.py
index 087b235..b3ce842 100755
--- a/swf2userSessions.py
+++ b/swf2userSessions.py
@@ -6,8 +6,8 @@ from src.workload import SwfField, Job
def swf2sessions(input_swf, output_dir, delim_approach, delim_threshold,
- dynamic_reduction, build_graph_rep, job_walltime_factor,
- given_walltime_only, job_grain, quiet):
+ dynamic_reduction, build_graph_rep, session_stat,
+ job_walltime_factor, given_walltime_only, job_grain, quiet):
users = {}
if not os.path.exists(output_dir):
@@ -66,6 +66,14 @@ def swf2sessions(input_swf, output_dir, delim_approach, delim_threshold,
if build_graph_rep:
user.export_dependancy_graph(f"{output_dir}/graphs")
+ if session_stat:
+ stats = {}
+ for user_id, user in users.items():
+ stats[user_id] = user.to_session_stat()
+
+ with open(f"{output_dir}/session_stat.json", "w") as file:
+ json.dump(stats, file)
+
if not quiet:
print("\nSWF parsing done.")
print("Number of users: ", len(users))
@@ -120,6 +128,13 @@ if __name__ == "__main__":
action="store_true",
help="Build a graphical representation of each session graph and save "
"them in a subfolder as gml files")
+ parser.add_argument(
+ '--session_stat', action="store_true",
+ help="Output a separate file `session_stat.json` containing summary "
+ "information about the user sessions. In particular, it contains info "
+ "about the original session durations (finish times) that are NOT "
+ "embedded in the SABjsons."
+ )
parser.add_argument("-q",
"--quiet",
@@ -174,6 +189,7 @@ if __name__ == "__main__":
delim_threshold=threshold,
dynamic_reduction=not (args.no_dynamic_reduction),
build_graph_rep=args.graph,
+ session_stat=args.session_stat,
job_walltime_factor=args.job_walltime_factor,
given_walltime_only=args.given_walltime_only,
job_grain=args.job_grain,
--
GitLab