I'm dissolving a line feature class with over 400k rows of data based on two fields: 1) Key 2) TLGID.
However, when I run the dissolve with single feature activated, some smaller lines disappear altogether. When I change the setting to multi-feature and run the dissolve again, I get shape_lengths of 0 for those same lines that disappeared before. I've attached photos of the before and after the multi-feature dissolve.
BEFORE multi-feature dissolve:

AFTER multi-feature dissolve:

Is there a cluster tolerance that is set by default that deletes segments that don't meet a minimum length? In my case 2.8 meters...