At the moment, workers will buffer any amount of log data (until they run out of memory and crash). We should abort any job that produces more than some limit, as it's probably stuck in a loop.
ocluster copied to clipboard
ocurrent
