Offspring Master shouldn't log errors for builders that are not expected to be running

[Kevin McDermott]

Currently, LexbuilderMaster.scanSlaves scans all slaves, irrespective of whether or not they're flagged as active.

This means that when scanning slaves, if a slave was marked as is_active=False it would still be scanned, and so the logs are constantly reporting things like

Wed, 02 Jan 2013 14:56:20 ERROR Scan of olive failed. Socket error: [Errno 113] No route to host
Wed, 02 Jan 2013 14:56:20 ERROR Scan of lexbuild7 failed. Socket error: [Errno -2] Name or service not known
Wed, 02 Jan 2013 14:56:20 ERROR Scan of lexbuild8 failed. Socket error: [Errno -2] Name or service not known
Wed, 02 Jan 2013 14:56:20 ERROR Scan of lexbuild4 failed. Socket error: [Errno -2] Name or service not known
Wed, 02 Jan 2013 14:56:20 ERROR Scan of lexbuild1 failed. Socket error: [Errno -2] Name or service not known
Wed, 02 Jan 2013 14:56:20 ERROR Scan of lexbuild5 failed. Socket error: [Errno -2] Name or service not known
Wed, 02 Jan 2013 14:56:20 ERROR Scan of lexbuild10 failed. Socket error: [Errno -2] Name or service not known
Wed, 02 Jan 2013 14:56:21 ERROR Scan of lexbuild3 failed. Socket error: [Errno -2] Name or service not known

scanSlaves (actually _get_builders) should only scan active slaves.

[Steve Magoun (smagoun)]

Scanning inactive slaves is a feature, it helps the admin know whether an inactive slave is available to be put back into service. For example, looking at the builders page I see that monkeyapple is inactive but online (healthy), and ready to be put back into service if necessary.

[David Murphy (schwuk)]

Setting as high priority as we want to reduce noise in the logs.