This was an analysis input/parameter issue, not a tool problem. MACS failed to call peaks due to the parameter settings versus the data content - likely the depth of coverage. Lowering the MFOLD as reported by MACS is just one option to test - if you want to use this older version of the tool.
I can see that you have had better results with MACS2. See the MACS2 documentation about how to tune those parameters. Most parameters included on a tool form are the same as in a 3rd party tool's manual. Any set will impact analysis results the same as if used line-command.
Thanks, Jen, Galaxy team