> ## Documentation Index > Fetch the complete documentation index at: https://docs.fjall.io/llms.txt > Use this file to discover all available pages before exploring further. # unclassified > Deployment failed The deployment failed, but Fjall could not match the failure to a known error code. The cause could not be attributed automatically. It may sit in your AWS account, in your application's code, or on Fjall's side. The code sometimes carries a suffix naming the pipeline phase where the failure was located, such as `unclassified.synth`, `unclassified.image_build`, or `unclassified.ecs_stabilise`. The phase vocabulary is `preflight`, `credentials`, `synth`, `bootstrap`, `image_build`, `image_push`, `cloudformation`, `ecs_stabilise`, and `platform`. ## Why this happens Fjall classifies failures by reading structured error sources, such as CloudFormation stack events and ECS service events. When no structured source survives to the point where the deployment terminates, there is nothing to classify against, so Fjall records the failure as unclassified. The record is not empty. It still carries the failure detail, a masked excerpt of the first error window from the logs, and the full event stream stays available on the deployment's detail page. Unclassified rates are tracked. When a failure shape recurs often enough, it earns its own code, and stored failures can be reclassified under it. ## How to fix it 1. Read the failure detail and the masked log excerpt. They quote the first error the pipeline hit, which is usually the root cause. 2. If the code carries a phase suffix, narrow your investigation to that phase. A failure in `image_build` points at your Dockerfile and build inputs, while `cloudformation` points at the stack update. 3. Open the deployment on [fjall.io](https://fjall.io) to read the full event stream and log output around the excerpt. 4. Use the **Explain** action on the deployment's detail page for an AI readout of the failure. 5. Fix what the readout points at, then run `fjall deploy` again. ## What Fjall shows you The failure surfaces on the deployment's detail page on fjall.io, in the CLI's Error Detail block, and in the deployments list row title. The detail page carries the full event stream alongside the detail sentence and log excerpt. ## Related How Fjall contains and reports failed deployments. Deploy an application and read the CLI's failure output. Check the state of your deployments from the CLI. The full deployment pipeline, phase by phase.