Companies / Snowflake / incidents
Snowflake: INC20000277
Affected components: AWS - US East (Ohio) - Data Loading and UnloadingData Loading and Unloading
Timeline of updates
- investigating
Current status: We're investigating an issue with Snowflake Data Cloud. We'll provide an update within 1 hour. Customer experience: Customers hosted in the specified regions may be unable to execute scheduled or manual tasks. Tasks may be stuck in a pending state or fail to initiate, and affected users may see an "Unexpected exception during task dispatch" message or similar error messages.
- investigating
Current status: Our investigation has identified that tasks appear to be failing due to an a potential issue with a backend record lookup on the metadata database layer, and we're working to determine what's causing these records to return null and result in task failures. We'll provide another update within 1 hour. Customer experience: Customers hosted in the specified region may experience failures when executing scheduled or manual tasks. Tasks may be marked as failed even though the underlying query completed successfully, or may fail to initiate. Affected users may see an "Unexpected exception during task completion" message or similar error messages. Incident start time: 10:58 UTC October 13, 2026
- investigating
Current status: We've determined that a directory issue within the components of the metadata store responsible for facilitating the impacted tasks is preventing some task records from being read consistently, which is causing the task failures. We're working to identify the origin of the directory issue to assist us in restoring normal task execution. We'll provide another update within 1 hour. Customer experience: Customers hosted in the specified region may experience intermittent issues with scheduled or manual tasks. Most tasks complete as expected, but some may be marked as failed even though they ran successfully. A small number of tasks may remain in a pending state or fail to start, and affected users may see an "Unexpected exception during task dispatch" message or similar error messages. Incident start time: 10:58 UTC October 13, 2026
- investigating
Current status: Our investigation indicates that the directory issue is limited to the records the metadata store keeps for task execution. We're developing a fix to restore the impacted metadata and resolve the impact; however, due to the complexity of the issue, we expect this will take an extended amount of time. Once complete, we'll proceed with deploying the fix as a potential mitigation. We'll provide another update within 2 hours. Customer experience: Customers hosted in the specified region may experience intermittent issues with scheduled or manual tasks. Most tasks complete as expected, but some may be marked as failed even though they ran successfully. A small number of tasks may remain in a pending state or fail to start, and affected users may see an "Unexpected exception during task dispatch" message or similar error messages. Incident start time: 10:58 UTC October 13, 2026
- investigating
Current status: We're continuing our development of the fix to restore the impacted metadata. We'll provide another update within 2 hours. While our remediation efforts are ongoing, before rerunning a task that's reported as failed, we recommend checking query history to confirm whether the task's query completed successfully. Customer experience: Customers hosted in the specified region may experience intermittent issues with scheduled or manual tasks. Most tasks complete as expected, but some may be marked as failed even though they ran successfully. A small number of tasks may remain in a pending state or fail to start, and affected users may see an "Unexpected exception during task dispatch" message or similar error messages. Incident start time: 10:58 UTC October 13, 2026
- investigating
Current status: We're validating our fix to restore the impacted metadata and preparing to begin deployment to the affected region. We'll provide another update within 2 hours. While our remediation efforts are ongoing, before rerunning a task that's reported as failed, we recommend checking query history to confirm whether the task's query completed successfully. Customer experience: Customers hosted in the specified region may experience intermittent issues with scheduled or manual tasks. Most tasks complete as expected, but some may be marked as failed even though they ran successfully. A small number of tasks may remain in a pending state or fail to start, and affected users may see an "Unexpected exception during task dispatch" message or similar error messages. Incident start time: 10:58 UTC October 13, 2026
- investigating
Current status: We're continuing to develop the fix to restore the impacted task metadata. Once it's fully developed and validated, we'll begin deploying it to the affected region. While our remediation efforts are ongoing, before rerunning a task that's reported as failed, we recommend checking query history to confirm whether the task's query completed successfully. We'll provide another update within 4 hours. Customer experience: Customers hosted in the specified region may experience intermittent issues with scheduled or manual tasks. Most tasks complete as expected, but some may be marked as failed even though they ran successfully. A small number of tasks may remain in a pending state or fail to start, and affected users may see an "Unexpected exception during task dispatch" message or similar error messages. ETA: An ETA is not yet available. We'll provide one as soon as possible. Incident start time: 10:58 UTC October 13, 2026 Preliminary root cause: A subset of infrastructure in the region experienced an issue that left the task metadata inconsistent. As a result, some tasks were reported as failed even though they completed successfully, while others remained pending or failed to start.
- investigating
Current status: We've modified our remediation approach. After further investigation, we've determined that restarting infrastructure in the affected region is the most effective way to restore normal task execution. The infrastructure restart effort is in progress, and we've observed a decrease in task failure rates on infrastructure that has been restarted. We're monitoring to confirm normal task execution is fully restored across all the infrastructure. As a side effect of the restarts, task graph run history from before the incident start time (10:58 UTC October 13, 2026) may currently be unavailable, and history from after that time may be incomplete. We're investigating options to restore this history and we'll provide another update by 22:00 UTC on October 14. Customer experience: Prior to the mitigation: Customers hosted in the affected region may have experienced intermittent issues with scheduled or manual tasks. Most tasks completed as expected, but some may have been marked as failed even though they ran successfully. A small number of tasks may have remained in a pending state or failed to start. Affected users may have seen an error message, such as "Unexpected exception during task dispatch." After the mitigation: The next time each task runs, it may still be marked as failed with an internal error message even though its query ran successfully. This may happen up to one additional time per task. Customers may also see incomplete task graph run history. Specifically: - When customers open or refresh the Task Graphs tab on the Tasks page in Snowsight (Transformation » Tasks), or the Run History tab for individual tasks, they may not see task runs from before 10:58 UTC October 13, 2026. Some task runs from after that time may not appear. - The INFORMATION_SCHEMA.COMPLETE_TASK_GRAPHS function may not include some recently completed runs. - The INFORMATION_SCHEMA.CURRENT_TASK_GRAPHS function may show extra rows for previously scheduled runs that are no longer running. - The SNOWFLAKE.ACCOUNT_USAGE.COMPLETE_TASK_GRAPHS view will continue to show history from before 10:58 UTC October 13, 2026, but some runs after that time may not appear. ETA: An ETA is not yet available. We'll provide one as soon as possible. Workaround: Before rerunning a task that was reported as failed, we recommend using the QUERY_HISTORY function or the SNOWFLAKE.ACCOUNT_USAGE.QUERY_HISTORY view to confirm whether the queries run by the task actually succeeded or failed. Incident start time: 10:58 UTC October 13, 2026 Preliminary root cause: A subset of infrastructure in the region experienced an issue that left the task metadata inconsistent. As a result, some tasks were reported as failed even though they completed successfully, while others remained pending or failed to start.
- identified
Current status: While the infrastructure restarts in the affected region are still in progress, our monitoring indicates that the service is gradually recovering and the majority of failed tasks are now succeeding after being restarted. We're continuing to monitor the affected infrastructure to confirm that the service fully recovers as expected, and we're working on restoring the task graph run history affected by the restarts to resolve the remaining impact. We'll provide another update within 8 hours. Customer experience: Prior to the mitigation: Customers hosted in the affected region may have experienced intermittent issues with scheduled or manual tasks. Most tasks completed as expected, but some may have been marked as failed even though they ran successfully. A small number of tasks may have remained in a pending state or failed to start. Affected users may have seen an error message, such as "Unexpected exception during task dispatch." After the mitigation: The next time each task runs, it may still be marked as failed with an internal error message even though its query ran successfully. This may happen up to one additional time per task. Customers may also see incomplete task graph run history. Specifically: - When customers open or refresh the Task Graphs tab on the Tasks page in Snowsight (Transformation » Tasks), or the Run History tab for individual tasks, they may not see task runs from before 10:58 UTC October 13, 2026. Some task runs from after that time may not appear. - The INFORMATION_SCHEMA.COMPLETE_TASK_GRAPHS function may not include some recently completed runs. - The INFORMATION_SCHEMA.CURRENT_TASK_GRAPHS function may show extra rows for previously scheduled runs that are no longer running. - The SNOWFLAKE.ACCOUNT_USAGE.COMPLETE_TASK_GRAPHS view will continue to show history from before 10:58 UTC October 13, 2026, but some runs after that time may not appear. Workaround: Before rerunning a task that was reported as failed, we recommend using the QUERY_HISTORY function or the SNOWFLAKE.ACCOUNT_USAGE.QUERY_HISTORY view to confirm whether the queries run by the task actually succeeded or failed. If the queries failed, we recommend rerunning the task to bypass the impact. Incident start time: 10:58 UTC October 13, 2026 Preliminary root cause: A subset of infrastructure in the region experienced an issue that left the task metadata inconsistent. As a result, some tasks were reported as failed even though they completed successfully, while others remained pending or failed to start.
- identified
Current status: The infrastructure restarts in the affected region are complete, and our monitoring indicates that the service is continuing to recover. Tasks that were scheduled before the restarts and haven't yet run afterward may be marked as failed the next time they run, even though they actually run successfully. We're preparing a fix to restore the task graph run history and resolve the remaining impact. We'll provide another update within 15 hours. Customer experience: Prior to the mitigation: Customers hosted in the affected region may have experienced intermittent issues with scheduled or manual tasks. Most tasks completed as expected, but some may have been marked as failed even though they ran successfully. A small number of tasks may have remained in a pending state or failed to start. Affected users may have seen an error message, such as "Unexpected exception during task dispatch." After the mitigation: Tasks that haven't run since the infrastructure restarts may be marked as failed with an internal error message the next time they run, even though their queries run successfully. This may happen up to one additional time per task. Customers may also see incomplete task graph run history. Specifically: - When customers open or refresh the Task Graphs tab on the Tasks page in Snowsight (Transformation » Tasks), or the Run History tab for individual tasks, they may not see task runs from before 10:58 UTC October 13, 2026. Some task runs from after that time may not appear. - The INFORMATION_SCHEMA.COMPLETE_TASK_GRAPHS function may not include some recently completed runs. - The INFORMATION_SCHEMA.CURRENT_TASK_GRAPHS function may show extra rows for previously scheduled runs that are no longer running. - The SNOWFLAKE.ACCOUNT_USAGE.COMPLETE_TASK_GRAPHS view will continue to show history from before 10:58 UTC October 13, 2026, but some runs after that time may not appear. ETA: An ETA is not yet available. We'll provide one as soon as possible. Workaround: Before rerunning a task that was reported as failed, we recommend using the QUERY_HISTORY function or the SNOWFLAKE.ACCOUNT_USAGE.QUERY_HISTORY view to confirm whether the queries run by the task actually succeeded or failed. If the queries failed, we recommend rerunning the task to bypass the impact. Incident start time: 10:58 UTC October 13, 2026 Preliminary root cause: A subset of infrastructure in the region experienced an issue that left the task metadata inconsistent. As a result, some tasks were reported as failed even though they completed successfully, while others remained pending or failed to start.
- monitoring
Current status: The infrastructure restarts in the affected region are complete, and our monitoring indicates that the error leading to task failure has been resolved and previously affected tasks are now consistently running successfully. While the functional impact of the issue is resolved, we're deploying a secondary fix to restore the remaining task graph run history. Based on the nature of this activity, we expect this work will take an extended amount of time. Customer experience: While the functional impact regarding task failures or tasks silently succeeding is resolved, affected customers may still see incomplete task graph run history. Specifically: - When customers open or refresh the Task Graphs tab on the Tasks page in Snowsight (Transformation » Tasks), or the Run History tab for individual tasks, they may not see task runs from before 10:58 UTC October 13, 2026. Some task runs from after that time may not appear. - The INFORMATION_SCHEMA.COMPLETE_TASK_GRAPHS function may not include some recently completed runs. - The INFORMATION_SCHEMA.CURRENT_TASK_GRAPHS function may show extra rows for previously scheduled runs that are no longer running. - The SNOWFLAKE.ACCOUNT_USAGE.COMPLETE_TASK_GRAPHS view will continue to show history from before 10:58 UTC October 13, 2026, but some runs after that time may not appear. Incident start time: 10:58 UTC October 13, 2026 Incident end time: 17:35 UTC October 15, 2026 Preliminary root cause: Our investigation identified a metadata mismatch within the metadata database layer. As a result, some tasks were reported as failed even though they completed successfully, while others remained pending or failed to start.
- resolved
Current status: We've implemented the fix for this issue and monitored the environment to confirm that service was restored. We're continuing our deployment of a secondary fix to restore the remaining task graph run history for the affected time period. If you experience additional issues or have questions, please open a support case via Snowflake Community or the Support page in Snowsight. Customer experience: Customers hosted in the affected region may have experienced intermittent issues with scheduled or manual tasks. Most tasks completed as expected, but some may have been marked as failed even though they ran successfully. A small number of tasks may have remained in a pending state or failed to start. Affected users may have seen an error message, such as "Unexpected exception during task dispatch." After the primary mitigation, the next time each task ran, it may still have been marked as failed with an internal error message even though its query ran successfully. This may have happened up to one additional time per task. While the functional impact regarding task failures or tasks silently succeeding is resolved, affected customers may still see incomplete task graph run history. Specifically: - When customers open or refresh the Task Graphs tab on the Tasks page in Snowsight (Transformation » Tasks), or the Run History tab for individual tasks, they may not see task runs from before 10:58 UTC October 13, 2026. Some task runs from after that time may not appear. - The INFORMATION_SCHEMA.COMPLETE_TASK_GRAPHS function may not include some recently completed runs. - The INFORMATION_SCHEMA.CURRENT_TASK_GRAPHS function may show extra rows for previously scheduled runs that are no longer running. - The SNOWFLAKE.ACCOUNT_USAGE.COMPLETE_TASK_GRAPHS view will continue to show history from before 10:58 UTC October 13, 2026, but some runs after that time may not appear. Incident start time: 10:58 UTC October 13, 2026 Incident end time: 17:35 UTC October 15, 2026 Preliminary root cause: Our investigation identified a metadata mismatch within the metadata database layer. As a result, some tasks were reported as failed even though they completed successfully, while others remained pending or failed to start. A root cause analysis (RCA) will be conducted and a summary will be made available within 5 business days.
Impact, status and times are the vendor’s own, as last archived. View on stspg.io (opens in new tab). Snowflake profile · Recent Snowflake incidents · Markdown · JSON