def get_dfia()

in cluster-trace-gpu-v2020/analysis/utils.py [0:0]

• 11 lines of code
• 2 McCabe index (conditional complexity)

def get_dfia(dfi):
    dfi_s = dfi[dfi.start_time > 0][['job_name','task_name','start_time']].groupby(['job_name','task_name']).min()  # start_time
    dfi_e = dfi[dfi.end_time > 0][['job_name','task_name','end_time']].groupby(['job_name','task_name']).max()  # end_time
    dfi_m = dfi[(dfi.start_time > 0) & (dfi.end_time > 0)][['job_name','task_name','end_time','start_time']]
    dfi_m['runtime'] = dfi_m.end_time-dfi_m.start_time
    dfi_m = dfi_m.groupby(['job_name','task_name']).mean()[['runtime']].reset_index() # runtime
    dfi_u = dfi[['job_name','task_name','status']].drop_duplicates().groupby(['job_name','task_name']).max() # status
    dfia = dfi_u
    for df in [dfi_s, dfi_e, dfi_m]:
        dfia = dfia.merge(df, on=['job_name','task_name'], how='left')
    return dfia