[Tool] Collecting failed test details

[schaffung]

The purpose of running the test cases is to find if gluster is failing any corner case or if any feature is broken after certain changes. The focus is hence on what is failing ( in case of functional test cases ). Hence, it is important that we can quickly get to the traceback thrown and collected by the test case. For further analysis, it is good to have all the traceback ( after some level of parsing ) documented.

Rather than doing this step manually after every run, a tool can go through the existing logs of a run and collect the failure records for faster debugging.

[BarakSason]

Why not have the framework handle this?
Through the course of the framework's execution, the framework have all the necessary information to export this data.
IMO, instead of creating a separate too for this functionality, a preferable solution is to enable this functionality from within the framework itself, using a CLI flag.
Since the framework already have the necessary information, generating such output would require much less compute power (and time) than going through all the generated logs.
Output should probably be in a single file format, containing all the TBs that occurred.