Currently, the code stores the error logs in the DB like the other endpoints. This is a temporary (and unwise) choice for error logging because of the big amount of data it is likely to generate, and the fact we will want to only keep a few weeks of it.
So we should send those logs elsewhere. After a sync with olblak, we need to prototype sending it instead to a FluentD socket to integrate with the existing error logging setup in the K8S cluster used in jenkins.io