Resolving Event Failures

You can view the list of failed Event Types, the failure reason, and the number of Events failing for that reason for an object, in the Pipeline Overview page. The resolution of failed Events in Hevo is decided by the type of failure, the configuration of the Pipeline, and the underlying failure reason. You may choose to resolve and replay only a partial list or just the critical failed Events for a failed Event Type. The resolved Events are processed and the remaining are again listed as failed Events. You can also permanently discard the failed Events.

The failures can be classified as:

  • Failures that you must resolve and replay manually. For example, in case of bad transformations code, you must fix the code, click DONE next to the failure message, and then, click the vertical ellipse icon, and Replay.

  • Failures that you must resolve, but which are auto-replayed by Hevo. For example, as soon as you correct the schema mapping error for an Event, Hevo immediately queues it up for auto-replay. For some of these errors, either you or Hevo may replay the Event. For example, if an Event fails due to insufficient disk space, Hevo automatically replays this Event every three hours assuming that you have fixed the issue. However, you can manually increase the disk space allocation and immediately replay the Event without waiting for Hevo.

  • Transient failures that are automatically resolved by Hevo. For example, Hevo may park some Events as Failed Events to reduce the data transfer load, or if some internal thresholds are reached. Similarly, an Event may fail and an error code may be generated for it, but if Hevo finds Auto Mapping enabled, it automatically fixes the issue, and the error is dismissed. These failures are not displayed in the Pipeline Activity page.

  • Unanticipated failures, which are individually investigated and resolved by Hevo.

Hevo provides you three ways to resolve Event failures:

  • Use Transformations

  • Manually fix the issue and replay the Events

  • Skip the failed Events


Resolving Event Failures through Transformations

You can apply Transformations on failed Events based on the failure reason, for example, in case of transformation script timed out error, you must optimize the transformation code to load the failed Events successfully to the Destination. Read Transformations for steps to do this. You can also test the Transformations on sample failed Events.


Manually Resolving Event Failures

In the Pipeline Overview page, perform the following steps to manually resolve Event failures:

Fix the Underlying Issues

  1. In the Pipeline Objects section, click on the summary count of the failed Events for an object and select View Details to view the failure reason.

    View Details

  2. Click the action button and make the required changes. For example, in the image below, click MAP SCHEMA to update the schema mapping. When finished, click DONE.

    Manually resolve error


Replay the Events

After fixing the error, you can replay the Events related to a specific failure reason individually for each Event Type or replay failed Events for all failure reasons across multiple Event Types and objects at once as a bulk action.

  • To replay fixed Events related to a specific failure reason for each Event Type individually:

  • To replay fixed Events for all failure reasons across multiple Event Types in an object:

  • To replay fixed Events across multiple Event Types and objects:

    1. In the Pipeline Objects section, click on the summary count of failed Events for all objects to to view the object-wise breakup.

      Summary count of failed Events

    2. Choose the objects you want to resolve or include all the objects by selecting the Objects checkbox. The Bulk Action toolbar is displayed.

      Objects checkbox

    3. In the Bulk Actions toolbar, click the MORE drop-down and select Replay Failed Events to replay all the Events.

      Bulk Replay Failed Events


Skipping Failed Events

You can skip the failed Events and not load them to the Destination. These Events are discarded or dropped by Hevo.

To discard failed Events, do one of the following:

  • Skip failed Events related to a specific failure reason for each Event Type individually:

    1. In the Pipeline Objects section, click the More (More) icon in the Event Type row, and then, Skip. A warning message is displayed.

      Skip failed Events related to a specific failure

    2. Click YES, GO AHEAD to purge the failed Events for the selected Event Type.

      Warning message for skipping Events

  • Skip all failed Events in an object at once

    1. In the Pipeline Objects section, click the More (More) icon on the object, and select Skip Failed Events. A warning message is displayed.

      Skip all Failed Events in an object

    2. Click YES, GO AHEAD to purge the selected set of failed Events.

      Warning message for skipping Events

  • Skip failed Events across multiple Event Types and objects.

    To do this:

    1. In the Pipeline Objects section, click on the summary count of failed Events for all objects to view the object-wise breakup.

      Summary count of Failed Events

    2. Select the objects whose failed Events you want to skip or select the Objects check box to skip the failed Events for all the objects.

      Objects checkbox

    3. In the Bulk Actions toolbar, click the MORE drop-down, and select Skip Failed Events. A warning message is displayed.

      Bulk skip failed Events

    4. Click YES, GO AHEAD to purge the selected set of failed Events.

      Warning message for skipping Events

Note:

  • When replayed, Events are fed back into the Transformations stage of the Pipeline. In case the Events were created through the Transformations code, they are fed back to the Schema Mapper stage instead of the Transformations stage.

  • If ingestion is log-based, being done via a webhook, or is from a DynamoDB Source, it may not be possible to re-ingest the skipped Events, as the Source system itself would have purged the data.


Auto-replaying Failed Events

Hevo auto-replays failed Events as frequently as five minutes. The total time taken for the Event to be processed, however, may be determined by other factors too, such as, the amount of time taken by the job to complete.



Revision History

Refer to the following table for the list of key updates made to this page:

Date Release Description of Change
Feb-21-2022 1.82 Updated section Manually Resolving Event Failures to add information about bulk options to Replay or Skip failed Events.
Mar-23-2021 1.59 Added the section, Resolving Event Failures through Transformations.
Last updated on 08 Mar 2022