Azure / azure-stream-analytics

Azure Stream Analytics
MIT License
224 stars 935 forks source link

Output data error handling set to Drop does not seem to work #179

Open Protheon opened 11 months ago

Protheon commented 11 months ago

Hi, We have a simple setup in place, where an azure devops webhook pushes data into an eventhub topic, which then pushes to the input of a stream analytics job. Stream analytics job then sends the output to an azure function http trigger.

We have set up the azure function trigger to return a 422 when it fails to process the event and in SOJ error policy, we have configured SOJ to Drop in Ouput data error.

We expected this drop action to skip the event that is being sent when a 422 occurs, however what we see is that soj keeps spamming the azure function with the same event, only to receive another 422.

Is this scenario unusual or is there something going wrong in the soj drop configuration? In the documentation, in drop section, it reads:

"All transient errors (for example, network errors) are retried regardless of the output error handling policy configuration." Does this mean that http error codes are also ignored?

If so, is there any way to drop outputs that do not succeed, such as a returning a different error code?

Steps to reproduce:

Create a webhook that pushes data to an eventhub top. configure a soj to receive the event and push to a http trigger. In http trigger set it to return 422 in case of fail.

Here is the soj arm template (redacted):

"resources": [ { "type": "Microsoft.StreamAnalytics/StreamingJobs", "apiVersion": "2021-10-01-preview", "name": "", "location": "", "identity": { }, "properties": { "sku": { "Name": "StandardV2", "StreamingUnits": "1/3" }, "outputStartMode": "JobStartTime", "outputErrorPolicy": "drop", "eventsOutOfOrderPolicy": "adjust", "eventsOutOfOrderMaxDelayInSeconds": 0, "eventsLateArrivalMaxDelayInSeconds": 5, "dataLocale": "en-US", "inputs": [ { "name": "", "properties": { "type": "Stream", "datasource": { "type": "Microsoft.ServiceBus/EventHub", "properties": { "serviceBusNamespace": "", "eventHubName": "", "consumerGroupName": "", "authenticationMode": "" } }, "serialization": { "type": "Json", "properties": { "encoding": "UTF8" ], "transformation": { "name": "", "properties": { "streamingUnits": 3, "query": "" } }, "outputs": [ { "name": "", "properties": { "datasource": { "type": "Microsoft.AzureFunction", "properties": { "FunctionAppName": "", "FunctionName": "", "ApiKey": "", "MaxBatchSize": 262144, "MaxBatchCount": 100 ]

Any help is appreciated!

curt-w commented 8 months ago

I'm also experiencing this unexpected behaviour. Function is returning a 500 but continues to be spammed.