search

snowplow / snowplow-elasticsearch-loader

Writes Snowplow enriched events from Kinesis to Elasticsearch
http://snowplowanalytics.com/
11 stars 18 forks source link

2.0.5 still prints the rejected event's payload in the err log? #234

Closed arthur78 closed 2 years ago

arthur78 commented 2 years ago

This is an example of what we can see in the error log for the 2.0.5 version of the loader:

{"schema":"iglu:com.snowplowanalytics.snowplow.badrows/generic_error/jsonschema/1-0-0","data":{"processor":{"artifact":"snowplow-elastics earch-loader-core","version":"2.0.5"},"failure":{"timestamp":"2022-06-07T09:51:48.046Z","errors":["Elasticsearch rejected record with message failed to parse field [] of type [long] in document with id 'b8aae16c-8bf3-4ea7-8189-93ebeb0b289c' . Preview of field's value: 'bf1d0d3c'"]},"payload":"{}"}}

Does it look good to you? Maybe I misunderstood the release note, but I would not expect to see such type of messages in the error logs since 2.0.5.

istreeter commented 2 years ago

HI @arthur78 thanks for getting in touch!

From looking at the output you shared, I guess have you have enabled the config option output.bad.type = stderr on this line of the hocon file. Is that right?

If you are using the stderr bad sink then bad rows will indeed still appear in the logs. You're right that our release notes did not make that clear, sorry about that.

For production pipelines, we always use the kinesis sink for failed events, i.e. set output.bad.type = kinesis. From my experience using this config setting we never see event payloads in the logs any more.

If you don't want to use the kinesis sink for failed events, but you also don't want to see failed events in your logs, then I think you can set output.bad.type = none.

arthur78 commented 2 years ago

Hi @istreeter thank you for the information! (I've changed to the "none" option – no more failed events in the error log.)