Interpreting "vmstat -z" output

[Garrett Wollman]

On a server that's been experiencing some issues, I note the following
in "vmstat -z":

ITEM                   SIZE  LIMIT     USED     FREE      REQ FAIL SLEEP

UMA Kegs:               208,      0,     188,      16,     188,   0,   0
UMA Zones:             3456,      0,     188,       0,     188,   0,   0
UMA Slabs:              568,      0, 1209668,    6211,50929964,   0,   0
UMA RCntSlabs:          568,      0,   50791,       1,   50791,   0,   0
UMA Hash:               256,      0,      78,      12,      80,   0,   0
16 Bucket:              152,      0,     227,      23,     227,   0,   0
32 Bucket:              280,      0,     522,      10,     522,   0,   0
64 Bucket:              536,      0,     783,       1,     783, 156,   0
128 Bucket:            1048,      0,   33513,       0,   33513,134423,   0
[...]

How should I interpret the failure count for "64 Bucket" and "128
Bucket"?  Does it represent a problem, or something that needs to be
tuned?  There are no obvious tunables, but the code is not exactly
transparent.  No other zones show failures.

-GAWollman