The following topics can help you troubleshoot issues that you encounter when using HealthOmics workflows and data stores.
Workflows
Topics
How do I troubleshoot a failed run?
Use the GetRun API operation to retrieve the failure reason. For more information, see Run failure reasons.
How do I troubleshoot a failed task?
Review the error code from the task failure message to understand the failure. Review the task logs in CloudWatch to see detailed logging messages for the task. If you aren’t getting detailed log messages, you can revise your workflow to output additional log statements. For more information, see Monitoring HealthOmics with CloudWatch Logs.
Where do I find the engine logs for successfully completed runs?
HealthOmics publishes logs to CloudWatch for failed runs only. If a run completes successfully, HealthOmics delivers the engine logs to your Amazon S3 bucket. For more information, see Logs in Amazon S3.
How can I reduce the input parameter size for a workflow?
You can specify up to 50 KB of input parameters for a workflow. You can use directory imports or sample sheets to remain within this size constraint. For more information, see Managing input parameter size.
Why is my run not completing?
If there are issues with your code and the processes have not exited properly, your run could become unresponsive or “stuck”. For more information on how to prevent and catch unresponsive runs, see Guidance for unresponsive runs.
Data stores
Topics
Why is S3 GetObject failing on my read set?
Most commonly, the failure is due to a missing permission. Sequence store S3 reading permission is a bi-directional configuration requiring both the sequence store S3 access policy to allow access and the IAM principal to have a policy attached allowing access. For more detail on the policy requirements see Permissions for data access using Amazon S3 URIs. Check that the following configurations are in place:
-
The sequence store S3 access policy has explicitly allowed access to the IAM principal or the root of the principal’s account.
-
Check that the IAM principal has a policy explicitly providing permission to the resource being accessed. Note that the IAM principal policy must use the Access Point ARN and not the Access point Alias based path when defining permissions and that the ARN is in the condition and not used to specify a resource.
-
If your store uses a customer managed key (CMK-KMS), ensure that the IAM principal has kms:decrypt permissions on the key. See the KMS cross-account access guide for configuring usage across accounts.
If you have a policy that's using tag based access controls, ensure the following:
-
Ensure that the sequence store has finished synchronizing the tags. For this, the store’s status needs to be active and not updating.
-
Ensure that there are no typos in the tag key or key value on the read set and the policy.
Why can't I see my annotation store or variant store in Athena?
In Lake Formation, be sure to create a resource link based on the store that was shared with you. Once you create a resource link that you have permission to access, the store should be visible in Athena. For more information, see Configuring Lake Formation to use HealthOmics.
Why can't I access my data store in Athena?
If your annotation or variant store is visible but you are receiving an error message saying that access is denied, check which query engine version you're using. Only queries run using engine version 3 are supported. To read more about Athena query engine versions, see the Amazon Athena documentation.
