What determines pipeline success vs failure status? Can pipeline be successful even if a job fails?

Summary:

What exactly determines the pipeline status? Do all jobs within the pipeline have to complete successfully for a pipeline to show status as successful? Is there any way we can show status as successful even if a job within the pipeline fails?

I created a pipeline, utilizing a Groovy script, that will run a data load, determine if there are any data load errors, then import the data load errors as new metadata. When there is a data load error, this causes this data load step to have a failure status. On failure, the pipeline stage 1 then goes to the next stage, which will refresh the database (to commit the new metadata), and will then progress to the next stage, which re-runs the same Groovy script in stage1 and assuming all data load errors have been resolved, the expectation is