pipelines/samples/core/caching/caching_sample.py

84 lines
3.0 KiB
Python

kfp_endpoint = None
import datetime
import time
import kfp
from kfp.components import create_component_from_func
@create_component_from_func
def do_work_op(seconds: float = 60) -> str:
import datetime
import time
print(f"Working for {seconds} seconds.")
for i in range(int(seconds)):
print(f"Working: {i}.")
time.sleep(1)
print("Done.")
return datetime.datetime.now().isoformat()
def caching_pipeline(seconds: float = 60):
# All outputs of successfull executions are cached
work_task = do_work_op(seconds)
# Test 1
# Running the pipeline for the first time.
# The pipeline performs work and the results are cached.
# The pipeline run time should be ~60 seconds.
print("Starting test 1")
start_time = datetime.datetime.now()
kfp.Client(host=kfp_endpoint).create_run_from_pipeline_func(
caching_pipeline,
arguments=dict(seconds=60),
).wait_for_run_completion(timeout=999)
elapsed_time = datetime.datetime.now() - start_time
print(f"Total run time: {int(elapsed_time.total_seconds())} seconds")
# Test 2
# Running the pipeline the second time.
# The pipeline should reuse the cached results and complete faster.
# The pipeline run time should be <60 seconds.
print("Starting test 2")
start_time = datetime.datetime.now()
kfp.Client(host=kfp_endpoint).create_run_from_pipeline_func(
caching_pipeline,
arguments=dict(seconds=60),
).wait_for_run_completion(timeout=999)
elapsed_time = datetime.datetime.now() - start_time
print(f"Total run time: {int(elapsed_time.total_seconds())} seconds")
if elapsed_time.total_seconds() > 60:
raise RuntimeError("The cached execution was not re-used or pipeline run took to long to complete.")
# Test 3
# For each task we can specify the maximum cached data staleness.
# For example: task.execution_options.caching_strategy.max_cache_staleness = "P7D" # (7 days)
# The `max_cache_staleness` attribute uses the [RFC3339 duration format](https://tools.ietf.org/html/rfc3339#appendix-A). For example: "P0D" (0 days), "PT5H" (5 hours; notice the "T")
# Cached results that are older than the specified time span, are not reused.
# In this case, the pipeline should not reuse the cached result, since they will be stale.
def caching_pipeline3(seconds: float = 60):
# All outputs of successfull executions are cached
work_task = do_work_op(seconds)
# TODO(Ark-kun): Fix handling non-zero periods in the backend
work_task.execution_options.caching_strategy.max_cache_staleness = 'P0D' # = Period: Time: 0 seconds
# Waiting for some time for the cached data to become stale:
time.sleep(10)
print("Starting test 3")
start_time = datetime.datetime.now()
kfp.Client(host=kfp_endpoint).create_run_from_pipeline_func(
caching_pipeline3,
arguments=dict(seconds=60),
).wait_for_run_completion(timeout=999)
elapsed_time = datetime.datetime.now() - start_time
print(f"Total run time: {int(elapsed_time.total_seconds())} seconds")
if elapsed_time.total_seconds() < 60:
raise RuntimeError("The cached execution was apparently re-used, but that should not happen.")