Actual Behaviour

Currently, when marking pod status, we will simply mark the TaskRun that has any of its steps terminated with an non-zero code failed with simply a reason “Failed”. This is covering up the actual Failure reason and inject the failure messages, which are hardly debuggable, into the failure message to be the only place . We should better surface the actual failure reasons based on the conditions at getting failure messages from pod phase and container status.

Expected behaviour

related: #7013
/kind feature