Hi Jesry, Regarding the gauge metrics, I have verified that it's a bug introduced when bumping the Beam version. I have opened a ticket [1] to track it.
Regarding the warning message, I think you could just ignore it. Per my understanding, it's saying that it's using a deprecated API of pyarrow. However, it should still work. Regards, Dian [1] https://issues.apache.org/jira/browse/FLINK-26846 On Wed, Mar 23, 2022 at 1:13 AM Jesry Pandawa <jesry.p.pand...@gmail.com> wrote: > Hello, > > I try to run my flink java job using Python UDF, but i have some > doubt using gauge metrics and vectorized udf on it. > > 1. i try to create a simple udf like this example on the docs > > https://nightlies.apache.org/flink/flink-docs-release-1.14/docs/dev/python/table/metrics/#gauge. > i also have set the metric reporter. but still cannot get the gauge metric. > when i try to use counter, i can see the metrics. > 2. to use vectorized udf, you just need to add this parameter > *func_type="pandas"* in decorator udf. when i try it using simple > scalar udf like the example on the documentation and adding that parameter, > there is an error log on my application even though the job still running > fine. > > This is the error message: > > 10:35:09,378 ERROR > > /Users/jpandawa/.pyenv/versions/3.8.5/lib/python3.8/site-packages/pyflink/fn_execution/beam/beam_sdk_worker_main.py:35 > - > > /Users/jpandawa/.pyenv/versions/3.8.5/lib/python3.8/site-packages/pyflink/table/utils.py:55: > FutureWarning: Schema passed to names= option, please pass schema= > explicitly. Will raise exception in future > > return pa.RecordBatch.from_arrays(arrays, schema) is this expected? > > > Regards, > > Jesry >
