Currently the agent post chain collects only a limited amount of logs for troubleshooting a failed job.

It's then necessary to create a new step dedicated to gather agent specific artifacts useful to determine the issue of a failed deployment, n particular if the problem happens during the bootstrap: in such case all the journals for the various agent services should be collected and published.

Note: the logs may contain relevant data (such as password, keys or tokens) so that they must be properly obfuscated if required before publishg them as job artifacts