%matplotlib inline
import pandas as pd
import socket
host = socket.getfqdn()
from core import load, zoom, calc, save,plots,monitor
#reload funcs after updating ./core/*.py
import importlib
importlib.reload(load)
importlib.reload(zoom)
importlib.reload(calc)
importlib.reload(save)
importlib.reload(plots)
importlib.reload(monitor)
<module 'core.monitor' from '/ccc/work/cont003/gen7420/odakatin/monitor-sedna/notebook/core/monitor.py'>
# 'month': = 'JOBID' almost month but not really,
# If you submit the job with job scheduler, above
#below are list of enviroment variable one can pass
#%env local='2"
# local : if True run dask local cluster, if not true, put number of workers
# setted in the 'local'
# if no 'local ' given, local will be setted automatically to 'True'
#%env ychunk='2'
#%env tchunk='2'
# controls chunk. 'False' sets no modification from original netcdf file's chunk.
# ychunk=10 will group the original netcdf file to 10 by 10
# tchunk=1 will chunk the time coordinate one by one
#%env control=FWC_SSH
# name of control file to be used for computation/plots/save/
#%env file_exp=
# 'file_exp': Which 'experiment' name is it?
#. this corresopnds to intake catalog name without path and .yaml
#%env year=
# for Validation, this correspoinds to path/year/month 's year
# for monitoring, this corresponids to 'date' having * means do all files in the monitoring directory
# setting it as *0[0-9] &*1[0-9]& *[2-3][0-9], the job can be separated in three lots.
#%env month=
# for monitoring this corresponds to file path path-XIOS.{month}/
#
#%env save= proceed saving? True or False , Default is setted as True
#%env plot= proceed plotting? True or False , Default is setted as True
#%env calc= proceed computation? or just load computed result? True or False , Default is setted as True
#%env save=False
%%time
# 'savefig': Do we save output in html? or not. keep it true.
savefig=True
client,cluster,control,catalog_url,month,year,daskreport,outputpath = load.set_control(host)
!mkdir -p $outputpath
!mkdir -p $daskreport
client
local True using host= irene4180.c-irene.mg1.tgcc.ccc.cea.fr starting dask cluster on local= True workers 16 10000000000 False not local in tgcc rome local cluster starting This code is running on irene4180.c-irene.mg1.tgcc.ccc.cea.fr using SEDNA_ALPHA_MONITOR file experiment, read from ../lib/SEDNA_ALPHA_MONITOR.yaml on year= * on month= 25 outputpath= ../results/SEDNA_ALPHA_MONITOR/25/ daskreport= ../results/dask/2585135irene4180.c-irene.mg1.tgcc.ccc.cea.fr_SEDNA_ALPHA_MONITOR_25AWTMD/ CPU times: user 375 ms, sys: 258 ms, total: 633 ms Wall time: 11.6 s
Client
|
Cluster
|
df=load.controlfile(control)
#Take out 'later' tagged computations
#df=df[~df['Value'].str.contains('later')]
df
Value | Inputs | Equation | Zone | Plot | Colourmap | MinMax | Unit | Oldname | Unnamed: 10 | |
---|---|---|---|---|---|---|---|---|---|---|
AW_maxtemp_depth | gridT.votemper,gridS.vosaline,param.mask,param... | calc.AWTD4(data) | ALL | AWTD_map | jet | (0,800) | m | M-5 |
Each computation consists of
%%time
import os
calcswitch=os.environ.get('calc', 'True')
loaddata=((df.Inputs != '').any())
print('calcswitch=',calcswitch,'df.Inputs != nothing',loaddata)
data = load.datas(catalog_url,df.Inputs,month,year,daskreport) if ((calcswitch=='True' )*loaddata) else 0
data
calcswitch= False df.Inputs != nothing True CPU times: user 445 µs, sys: 0 ns, total: 445 µs Wall time: 418 µs
0
%%time
monitor.auto(df,data,savefig,daskreport,outputpath,file_exp='SEDNA'
)
#calc= False #save= False #plot= True Zone='ALL' Value='AW_maxtemp_depth' cmap='jet' clabel='m' clim= (0, 800) outputpath='../results/SEDNA_ALPHA_MONITOR/25/' nc_outputpath='../nc_results/SEDNA_ALPHA_MONITOR/25/' filename='SEDNA_AWTD_map_ALL_AW_maxtemp_depth' #3 no computing , loading starts dtaa=save.load_data(plot=Plot,path=nc_outputpath,filename=filename) start saving data filename= ../nc_results/SEDNA_ALPHA_MONITOR/25/SEDNA_AWTD_map_ALL_AW_maxtemp_depth/t_*/x_*/y_*.nc
--------------------------------------------------------------------------- OSError Traceback (most recent call last) <timed eval> in <module> /ccc/work/cont003/gen7420/odakatin/monitor-sedna/notebook/core/monitor.py in auto(df, val, savefig, daskreport, outputpath, file_exp) 68 print('dtaa=save.load_data(plot=Plot,path=nc_outputpath,filename=filename)' ) 69 with performance_report(filename=daskreport+"_calc_"+step.Value+".html"): ---> 70 data=save.load_data(plot=Plot,path=nc_outputpath,filename=filename) 71 #saveswitch=False 72 display(data) /ccc/work/cont003/gen7420/odakatin/monitor-sedna/notebook/core/save.py in load_data(plot, path, filename) 36 data=load_twoD(path,filename,nested=False) 37 else: ---> 38 data=load_twoD(path,filename) 39 print('load computed data completed') 40 return data /ccc/work/cont003/gen7420/odakatin/monitor-sedna/notebook/core/save.py in load_twoD(path, filename, nested) 50 ,concat_dim=('x','y','t') 51 ,combine='by_coords' #param_xios ---> 52 ,coords='minimal') 53 54 def twoD(data,path='../result',filename='toto',nested=True): ~/monitor/lib/python3.7/site-packages/xarray/backends/api.py in open_mfdataset(paths, chunks, concat_dim, compat, preprocess, engine, lock, data_vars, coords, combine, autoclose, parallel, join, attrs_file, **kwargs) 919 920 if not paths: --> 921 raise OSError("no files to open") 922 923 # If combine='by_coords' then this is unnecessary, but quick. OSError: no files to open