%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/talandel/TOOLS/monitor-sedna/notebook/core/monitor.py'>
If you submit the job with job scheduler; below are list of enviroment variable one can pass
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 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. For DELTA experiment, year corresponds to really 'year'
%env month=
for monitoring this corresponds to file path path-XIOS.{month}/
For DELTA experiment, year corresponds to really 'month'
proceed saving? True or False , Default is setted as True
proceed plotting? True or False , Default is setted as True
proceed computation? or just load computed result? True or False , Default is setted as True
save output file used for plotting
using kerchunked file -> False, not using kerhcunk -> True
name of control file to be used for computation/plots/save/ We have number of M_xxx.csv
Monitor.sh calls M_MLD_2D
and AWTD.sh, Fluxnet.sh, Siconc.sh, IceClim.sh, FWC_SSH.sh, Integrals.sh , Sections.sh
M_AWTMD
M_Fluxnet
M_Ice_quantities
M_IceClim M_IceConce M_IceThick
M_FWC_2D M_FWC_integrals M_FWC_SSH M_SSH_anomaly
M_Mean_temp_velo M_Mooring
M_Sectionx M_Sectiony
%%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= irene8004.c-irene.tgcc.ccc.cea.fr starting dask cluster on local= True workers 16 10000000000 False tgcc local cluster starting This code is running on irene8004.c-irene.tgcc.ccc.cea.fr using SEDNA_DELTA_MONITOR file experiment, read from ../lib/SEDNA_DELTA_MONITOR.yaml on year= 2015 on month= 12 outputpath= ../results/SEDNA_DELTA_MONITOR/ daskreport= ../results/dask/6610542irene8004.c-irene.tgcc.ccc.cea.fr_SEDNA_DELTA_MONITOR_12M_FWC_integrals/ CPU times: user 516 ms, sys: 147 ms, total: 663 ms Wall time: 14.6 s
Client-60ad5715-2aef-11ed-b847-080038b5adab
Connection method: Cluster object | Cluster type: distributed.LocalCluster |
Dashboard: http://127.0.0.1:8787/status |
c6cc21b1
Dashboard: http://127.0.0.1:8787/status | Workers: 12 |
Total threads: 48 | Total memory: 2.86 TiB |
Status: running | Using processes: True |
Scheduler-82f9f7e6-cac1-4056-9153-a30f21390cf9
Comm: tcp://127.0.0.1:38427 | Workers: 12 |
Dashboard: http://127.0.0.1:8787/status | Total threads: 48 |
Started: Just now | Total memory: 2.86 TiB |
Comm: tcp://127.0.0.1:40185 | Total threads: 4 |
Dashboard: http://127.0.0.1:34591/status | Memory: 244.27 GiB |
Nanny: tcp://127.0.0.1:33871 | |
Local directory: /tmp/dask-worker-space/worker-tc4k5g8c |
Comm: tcp://127.0.0.1:41037 | Total threads: 4 |
Dashboard: http://127.0.0.1:45791/status | Memory: 244.27 GiB |
Nanny: tcp://127.0.0.1:36051 | |
Local directory: /tmp/dask-worker-space/worker-dzhny8wy |
Comm: tcp://127.0.0.1:46077 | Total threads: 4 |
Dashboard: http://127.0.0.1:40061/status | Memory: 244.27 GiB |
Nanny: tcp://127.0.0.1:34711 | |
Local directory: /tmp/dask-worker-space/worker-cbxvkq9b |
Comm: tcp://127.0.0.1:39457 | Total threads: 4 |
Dashboard: http://127.0.0.1:35161/status | Memory: 244.27 GiB |
Nanny: tcp://127.0.0.1:41452 | |
Local directory: /tmp/dask-worker-space/worker-c70pg6_s |
Comm: tcp://127.0.0.1:41892 | Total threads: 4 |
Dashboard: http://127.0.0.1:38560/status | Memory: 244.27 GiB |
Nanny: tcp://127.0.0.1:37324 | |
Local directory: /tmp/dask-worker-space/worker-8yejs5t9 |
Comm: tcp://127.0.0.1:37385 | Total threads: 4 |
Dashboard: http://127.0.0.1:39383/status | Memory: 244.27 GiB |
Nanny: tcp://127.0.0.1:41302 | |
Local directory: /tmp/dask-worker-space/worker-a8dfnolr |
Comm: tcp://127.0.0.1:44576 | Total threads: 4 |
Dashboard: http://127.0.0.1:43663/status | Memory: 244.27 GiB |
Nanny: tcp://127.0.0.1:42762 | |
Local directory: /tmp/dask-worker-space/worker-e23pgfbj |
Comm: tcp://127.0.0.1:41751 | Total threads: 4 |
Dashboard: http://127.0.0.1:44917/status | Memory: 244.27 GiB |
Nanny: tcp://127.0.0.1:36955 | |
Local directory: /tmp/dask-worker-space/worker-0a8m0r51 |
Comm: tcp://127.0.0.1:33880 | Total threads: 4 |
Dashboard: http://127.0.0.1:44090/status | Memory: 244.27 GiB |
Nanny: tcp://127.0.0.1:44952 | |
Local directory: /tmp/dask-worker-space/worker-5a68__i8 |
Comm: tcp://127.0.0.1:42644 | Total threads: 4 |
Dashboard: http://127.0.0.1:44887/status | Memory: 244.27 GiB |
Nanny: tcp://127.0.0.1:34690 | |
Local directory: /tmp/dask-worker-space/worker-mor5v4j3 |
Comm: tcp://127.0.0.1:40613 | Total threads: 4 |
Dashboard: http://127.0.0.1:40774/status | Memory: 244.27 GiB |
Nanny: tcp://127.0.0.1:41871 | |
Local directory: /tmp/dask-worker-space/worker-vmj32a0v |
Comm: tcp://127.0.0.1:45162 | Total threads: 4 |
Dashboard: http://127.0.0.1:46348/status | Memory: 244.27 GiB |
Nanny: tcp://127.0.0.1:38643 | |
Local directory: /tmp/dask-worker-space/worker-ik4wpq7n |
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 | |
---|---|---|---|---|---|---|---|---|---|---|
FWC_integrals | calc.FWC_load_integrals(data,nc_outputpath) | BBFG | FWC_integrals | (12000,24000) | Km^3 | I-1 |
Each computation consists of
%%time
import os
calcswitch=os.environ.get('calc', 'True')
lazy=os.environ.get('lazy','False' )
loaddata=((df.Inputs != '').any())
print('calcswitch=',calcswitch,'df.Inputs != nothing',loaddata, 'lazy=',lazy)
data = load.datas(catalog_url,df.Inputs,month,year,daskreport,lazy=lazy) if ((calcswitch=='True' )*loaddata) else 0
data
calcswitch= True df.Inputs != nothing False lazy= False CPU times: user 424 µs, sys: 0 ns, total: 424 µs Wall time: 420 µs
0
%%time
monitor.auto(df,data,savefig,daskreport,outputpath,file_exp='SEDNA'
)
#calc= True #save= True #plot= False Value='FWC_integrals' Zone='BBFG' Plot='FWC_integrals' cmap='' clabel='Km^3' clim= (12000, 24000) outputpath='../results/SEDNA_DELTA_MONITOR/' nc_outputpath='../nc_results/SEDNA_DELTA_MONITOR/' filename='SEDNA_FWC_integrals_BBFG_FWC_integrals' #3 Start computing data= calc.FWC_load_integrals(data,nc_outputpath) monitor.optimize_dataset(data) start loading data filename= ../nc_results/SEDNA_DELTA_MONITOR/SEDNA_maps_BBFG_FWC_2D/t_*/y_*/x_*.nc dim ('x', 'y', 't') load computed data completed add optimise here once otimise can recognise
<xarray.Dataset> Dimensions: (t: 365) Coordinates: * t (t) object 2015-01-01 12:00:00 ... 2015-12-31 12:00:00 Data variables: FWC_Arctic (t) float64 dask.array<chunksize=(1,), meta=np.ndarray> FWC_CRF (t) float64 dask.array<chunksize=(1,), meta=np.ndarray>
#4 Saving SEDNA_FWC_integrals_BBFG_FWC_integrals data=save.datas(data,plot=Plot,path=nc_outputpath,filename=filename) start saving data saving data in a csv file ../nc_results/SEDNA_DELTA_MONITOR/SEDNA_FWC_integrals_BBFG_FWC_integrals2015-01-01_2015-12-31.nc save computed data at ../nc_results/SEDNA_DELTA_MONITOR/SEDNA_FWC_integrals_BBFG_FWC_integrals2015-01-01_2015-12-31.nc completed CPU times: user 3min 49s, sys: 18.6 s, total: 4min 8s Wall time: 5min 54s