CODEWITHSUNDEEP
pythongoogle-cloud-dataflowapache-beam
Yohei Onishi
Yohei Onishi

Reputation: 1524

Apache Beam Python SDK - Inaccurate session window interval

I am trying to process data with 60 minutes session interval using Apache Beam Python SDK. But the actual session interval was inaccurate such as 3:00:00 or 1:01:00 or 1:50:00 when I run my application.

Would you help me find a solution to fix this issue and process data with 60 minutes session?

I built my pipeline as bellow.

    with Pipeline(options=pipeline_options) as pipeline:
        (
            pipeline
            | "Read" >> ReadFromText(known_args.input, skip_header_lines=1)
            | "Convert" >> ParDo(Convert())
            | "Add Timestamp" >> Map(lambda x: window.TimestampedValue(x, get_timestamp_from_element(x).timestamp()))
            | "Use User ID As Key" >> Map(lambda x: (x["user_id"], x))
            | "Apply Session Window" >> WindowInto(window.Sessions(known_args.session_interval))
            | "Group" >> GroupByKey()
            | "Write To CSV" >> ParDo(WriteToCSV(known_args.output))
        )
        result = pipeline.run()
        result.wait_until_finish()

session_interval (60 minutes) is provided as bellow. 

    parser.add_argument(
        "--session_interval",
        help="Interval of each session",
        default=60*60) # 60 mins

WriteToCSV function process data per session. I logged the session duration but it was not accurate.

class WriteToCSV(DoFn):
    def __init__(self, output_path):
        self.output_path = output_path

    def process(self, element, window=DoFn.WindowParam):
        window_start = window.start.to_utc_datetime()
        window_end = window.end.to_utc_datetime()
        duration = window_end - window_start
        logging.info(">>> new %s record(s) in %s session (start %s end %s)", len(click_records), duration, window_start, window_end)
        ....

Then I got this log messages when I run this application locally with DirectRunner.

new 5 records in 3:00:00 session (start 2018-10-19 02:00:00 end 2018-10-19 05:00:00)
new 2 records in 1:01:00 session (start 2018-10-19 02:02:00 end 2018-10-19 03:03:00)
new 2 records in 1:50:00 session (start 2018-10-19 03:10:00 end 2018-10-19 05:00:00)

I also deployed the pipeline to Dataflow then got the same result.

new 2 record(s) in 1:50:00 session (start 2018-10-19 11:10:00 end 2018-10-19 13:00:00)
new 2 record(s) in 1:01:00 session (start 2018-10-19 10:02:00 end 2018-10-19 11:03:00)
new 5 record(s) in 3:00:00 session (start 2018-10-19 10:00:00 end 2018-10-19 13:00:00)

Upvotes: 1

Views: 203

Answers (1)

Jayadeep Jayaraman
Jayadeep Jayaraman

Reputation: 2825

In your beam pipeline, the variable ``known_args.session_intervalinwindow.Sessions` defined the gap duration i.e. the duration in which if no further events come for a specific key the window is closed. Each session can be of different start and end duration based on the number of events that are processed by the pipeline for a given key. This is explained pictorially here

For example

Key 1 - 10:00 AM ----|
Key 1 - 10:45 AM     |
Key 1 - 11:30 AM     |====> One Session Window for Key 1 of Duration 4hours 30 minutes
Key 1 - 12:15 PM     |
Key 1 - 01:00 PM ----|
Key 1 - 02:30 PM =========> Start of new session window for Key 1

Key 2 - 10:00 AM-----|
Key 2 - 10:30 AM     |====> One Session window for key 2 of Duration 1:00 hour
Key 2 - 11:00 PM-----|
Key 2 - 12:30 PM =========> Start of new session window for Key 2

If you are interested in grouping and processing events every 60 minutes then you need to use FixedWindows.

Upvotes: 1

Related Questions