Currently in any multi-peer grid, there is a grid.log and processes.log for each peer. Since any given set of workflows can run across any of these peers, it can be difficult to follow end-to-end processes; this is particularly problematic when trying to diagnose problems on the grid that didn't appear in Dev Studio.
To resolve this, provide a centralised logging capability that consolidates all peer grid grids into one place, all peer processes.logs into one place, and potentially the logs for other components such as AM and repo.
There already ways to do this using LOG4J configurations that are not tested or supported by BMC, e.g. using syslog or an external database [Customer and Partner space members only]. Perhaps these could be used, or a dedicated logging facility created.
This should be made easy to enable or disable without restarting peers so that, where desired, it can be left disabled until needed for debugging. The logs generated should be in a format that can be easily zipped up and sent to BMC Support.