Commit 1a55e8c9 authored by Ryan Randles Jones's avatar Ryan Randles Jones
Browse files

Delete Jobs-and-Users-ReqMemCPU.ipynb

parent 26c21463
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Notebook Setup"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"import sqlite3\n",
"import slurm2sql\n",
"import pandas as pd\n",
"import matplotlib.pyplot as plt\n",
"%matplotlib inline\n",
"import seaborn as sns\n",
"import plotly.express as px\n",
"import matplotlib.ticker as ticker\n",
"import numpy as np"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from RC_styles import rc_styles as style"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from sklearn.cluster import KMeans"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# creates database of info from March 2020 using sqlite 3\n",
"db = sqlite3.connect('/data/rc/rc-team/slurm-since-March.sqlite3')"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# df is starting database\n",
"df = pd.read_sql('SELECT * FROM slurm', db)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# voluntary\n",
"\n",
"# for displaying all available column options\n",
"pd.set_option('display.max_columns', None)\n",
"df.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# converts units in ReqMemCPU column from bytes to gigs\n",
"df['ReqMemCPU'] = df['ReqMemCPU'].div(1024**3)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# df_completed is dataframe of all completed jobs\n",
"df_completed = df[df.State.str.contains('COMPLETED')]\n",
"#df_completed.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# df_batch is df with only batch jobs\n",
"df_batch = df[df.JobName.str.contains('batch')]\n",
"#df_batch.head(5)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Average RAM per CPU Requested by User"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# df_2 is database of completed jobs with only User and ReqMemCpu\n",
"# it is used for the user dataframes\n",
"\n",
"df_2 = df_completed.loc[:,['User','ReqMemCPU']]\n",
"#df_2.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"df_2['ReqMemCPU'] = df_2['ReqMemCPU'].apply(np.ceil)\n",
"#df_2.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# fills empty strings in User column with NaN and then filters them out to give a dataset of users with no empty strings \n",
"\n",
"nan_value = float(\"NaN\")\n",
"\n",
"df_2.replace(\"\", nan_value, inplace=True)\n",
"\n",
"df_2.dropna(subset = [\"User\"], inplace=True)\n",
"#df_2.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# count = count of jobs per user\n",
"# mean,std,min,25%,50%,75%, and max refers to the gigs of memory per cpu requested by that user for all their jobs\n",
"df_user = df_2.groupby('User')['ReqMemCPU'].describe().reset_index()\n",
"#df_user.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# voluntary\n",
"\n",
"# description of number of jobs run per user - can be used to choose the Upper Limit Job Count\n",
"df_user['count'].describe()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# variable for to be used in names of plots to describe the max job count per user\n",
"\n",
"# max = 367257\n",
"UpperlimitJobCount = 100"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# creates database from df_user that returns all jobs per user up to the UpperlimitJobCount defined above\n",
"jobscount_cutoff = df_user[(df_user['count'] <= UpperlimitJobCount)]\n",
"#jobscount_cutoff.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# df_user_graph is df_user sorted in ascending order by count for easy readibility of graph\n",
"df_user_graph_full = jobscount_cutoff.sort_values(by='count', ascending=True)\n",
"df_user_graph_full.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"df_user_graph = df_user_graph_full.loc[:,['User','count','mean']]\n",
"df_user_graph.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"style.default_axes_and_ticks()\n",
"style.figsize()\n",
"\n",
"user_graph1 = sns.scatterplot(x=\"count\", y=\"mean\",data=df_user_graph)\n",
"\n",
"plt.title('Average Requested RAM per CPU by User for all Users Running %i Jobs or less'%UpperlimitJobCount)\n",
"\n",
"plt.xlabel('Job Count Per User')\n",
"plt.ylabel('Average Requested RAM per CPU (Gigs)')\n",
"\n",
"plt.show()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"kmeans = KMeans(n_clusters=3)\n",
"model = kmeans.fit(df_user_graph[['count', 'mean']])\n",
"# Now, we can get the predicted model labels, or Centroids, in the form of an array:\n",
"model.cluster_centers_"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# attach predicted cluster to original points\n",
"df_user_graph['predicted'] = model.labels_\n",
"df_user_graph.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Create a dataframe for cluster_centers (centroids)\n",
"centroids = pd.DataFrame(model.cluster_centers_, columns=[\"count\", \"mean\"])"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"style.default_axes_and_ticks()\n",
"style.figsize()\n",
"\n",
"## Plot scatter by cluster / color, and centroids\n",
"colors = [\"red\", \"green\", \"blue\"]\n",
"df_user_graph['color'] = df_user_graph['predicted'].map(lambda p: colors[p])\n",
"ax = df_user_graph.plot( \n",
" kind=\"scatter\", \n",
" x=\"count\", y=\"mean\",\n",
" c = df_user_graph['color']\n",
")\n",
"centroids.plot(\n",
" kind=\"scatter\", \n",
" x=\"count\", y=\"mean\", \n",
" marker=\"*\", c=[\"r\", \"g\", \"b\"], s=550,\n",
" ax=ax\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# trying the same above graph using diffrerent syntax"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"df_user_graph_cluster = df_user_graph_full.loc[:,['count','mean']]\n",
"#df_user_graph_cluster.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"kmeans = KMeans(n_clusters=3, random_state=111)\n",
"kmeans.fit(df_user_graph_cluster)\n",
"print(kmeans.cluster_centers_)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"plt.scatter(df_user_graph_cluster['count'],df_user_graph_cluster['mean'], c=kmeans.labels_, cmap='rainbow')\n",
"plt.scatter(kmeans.cluster_centers_[:,0] ,kmeans.cluster_centers_[:,1], color='grey')\n",
"#plt.yscale(\"log\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Average RAM per CPU by Job"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# df_3 is database with only JobStep, User, JobName, ReqMemCpu, ArrayJob, and ArrayTaskID\n",
"# it is used to pull out needed information and create separate datasets to compare\n",
"df_3 = df_batch.loc[:,['ReqMemCPU','JobID']]\n",
"#df_3.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"df_3['ReqMemCPU'] = df_3['ReqMemCPU'].apply(np.ceil)\n",
"#df_3.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# variable for to be used in names of plots to describe the max gigs measured\n",
"UpperlimitGB = 50"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# creates database from df_3 that returns all RAM per CPU requested up to the UpperRAMlimit defined above\n",
"gig_cutoff = df_3[(df_3.ReqMemCPU <= UpperlimitGB)]\n",
"#gig_cutoff.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# renames JobID column to JobCount since that's what it is now\n",
"df_cpu_per_job = gig_cutoff.groupby('ReqMemCPU').count().rename(columns={'JobID': 'JobCount'}).reset_index()\n",
"#df_cpu_per_job.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"df_cpu_per_job['ReqMemCPU'].describe()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"style.default_axes_and_ticks()\n",
"style.figsize()\n",
"\n",
"cpu_per_job = sns.scatterplot(x=\"ReqMemCPU\", y=\"JobCount\",data=df_cpu_per_job)\n",
"\n",
"cpu_per_job.set_yscale('log')\n",
"\n",
"#cpu_per_job.yaxis.set_major_locator(ticker.MultipleLocator(100000))\n",
"#cpu_per_job.yaxis.set_major_formatter(ticker.ScalarFormatter())\n",
"\n",
"plt.title('Number of Jobs Requesting RAM per CPU for all Jobs %i gigs or less'%UpperlimitGB)\n",
"\n",
"plt.xlabel('Requested RAM per CPU (Gigs) per Job')\n",
"plt.ylabel('Job Count')\n",
"\n",
"plt.show()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"df_cpu_per_job_cluster = gig_cutoff.groupby('ReqMemCPU').count().rename(columns={'JobID': 'JobCount'}).reset_index()\n",
"df_cpu_per_job_cluster.head(30)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"kmeans = KMeans(n_clusters=3, random_state=111)\n",
"kmeans.fit(df_cpu_per_job_cluster)\n",
"print(kmeans.cluster_centers_)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"print(kmeans.labels_)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"plt.scatter(df_cpu_per_job_cluster['ReqMemCPU'],df_cpu_per_job_cluster['JobCount'], c=kmeans.labels_, cmap='rainbow')\n",
"plt.scatter(kmeans.cluster_centers_[:,0] ,kmeans.cluster_centers_[:,1], color='grey')\n",
"plt.yscale(\"log\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# renames JobID column to JobCount since that's what it is now\n",
"job_count = df_3.groupby('ReqMemCPU').count().rename(columns={'JobID': 'JobCount'}).reset_index()\n",
"job_count.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"UpperlimitJobCount2 = 20"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# creates database from df_3 that returns all Jobs up to the UpperlimitJobCount2 defined above\n",
"df_job_count = job_count[(job_count.JobCount <= UpperlimitJobCount2)]\n",
"df_job_count.head(5)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"style.default_axes_and_ticks()\n",
"style.figsize()\n",
"\n",
"job_count_graph = sns.scatterplot(x=\"JobCount\", y=\"ReqMemCPU\",data=df_job_count)\n",
"\n",
"#job_count_graph.set_yscale('log')\n",
"\n",
"#job_count_graph.yaxis.set_major_locator(ticker.MultipleLocator(100000))\n",
"#job_count_graph.yaxis.set_major_formatter(ticker.ScalarFormatter())\n",
"\n",
"plt.title('Number of Jobs Requesting RAM per CPU for all Jobs counts of %i or less'%UpperlimitJobCount2)\n",
"\n",
"plt.xlabel('Job Count')\n",
"plt.ylabel('Average Requested RAM per CPU (Gigs) per Job')\n",
"\n",
"plt.show()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# must run\n",
"\n",
"# creates database from df_3 that returns all Jobs up to the UpperlimitJobCount2 defined above\n",
"df_job_count_cluster = job_count[(job_count.JobCount <= UpperlimitJobCount2)]\n",
"df_job_count_cluster.head(50)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"kmeans2 = KMeans(n_clusters=4, random_state=111)\n",
"kmeans2.fit(df_job_count)\n",
"print(kmeans2.cluster_centers_)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"print(kmeans2.labels_)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"plt.scatter(df_job_count['JobCount'],df_job_count['ReqMemCPU'], c=kmeans2.labels_, cmap='rainbow')\n",
"plt.scatter(kmeans2.cluster_centers_[:,1] ,kmeans2.cluster_centers_[:,0], color='grey')\n",
"#plt.yscale(\"log\")"
]
}
],
"metadata": {
"language_info": {
"name": "python",
"pygments_lexer": "ipython3"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment