I found several things that were causing jobs to fail excluded them with advanced search syntax.  My jobs are completing successfully again.

1. Records Management: cutoff documents cannot be OCRed.  Exclude cutoff documents using advanced search syntax: - {LFRM:ActualCutoffDate= "*"} 
2. Giant pages take too long to OCR, DCC considers that a failure.  Exclude documents with pages larger than 100MB using advanced search syntax: - {LF:imagesize > 105911401}
3. Corrupt documents cannot be OCRed.  I found two documents (so far) with corrupt images.  I  excluded those by entry ID, for example: - {LF:ID=123456}

My search string looks like this:

({LF:AssociatedPages="Y"} & {LF:OCR=none})
 & {LF:pagecount > 0}
 - {LFRM:ActualCutoffDate= "*"}
 - {LF:ID=123456}
 - {LF:imagesize > 105911401}