Hi community,
Please check out this couple of tools I wrote for visualizing logs. The part that could be particularly interesting to this community is the “gallery” for the second tool, “splot” (http://jkff.info/software/timeplotters/splot.html#gallery). It shows a few postmortems of various performance issues of a large cluster (5000 workers) as depicted by a moire pattern in the cluster activity graph.