No, your sleeps are not distributed evenly. It is not possible to attribute the
sleeps to individual blocks. You should run the second query repeatedly to find
out which of those blocks persist in the cache. Those are the most likely to be
the hot blocks causing the latch contention (but not necessarily so).

The block from file number 0 is from the SYSTEM tablespace. File numbers in the
X$ tables often fail to correspond directly to the file numbers shown in the
data dictionary. In general, these file numbers just need to be adjusted by 1
before joining to a data dictionary table. However, there is an exception in 8.1
which I'll treat in a separate mail to answers.


-----Original Message-----
Sent: Thursday, 31 August 2000 4:48


I ran your script the hot_hash_latches i got this output.

   CHILD#     SLEEPS CUR_BUFFERS CR_BUFFERS
--------- ---------- ----------- ----------
    12721       7203           7          0
    10052       6630           5          6
    10254       4850           5          0
    12840       4676           6          1
    10248       2117           7          0
    10170       1642           3          0
    15594       1410           3          0
    12722       1407           8          0
     5981       1118           5          0
     9154        799           3          0
     9159        634           2          0
     9153        557           3          0
     9155        465           3          0
     8343        239           8          0

Then i went down further and find out the the buffers mapping to the high sleeps
latches.

HLADDR       CHILD#      FILE#     DBABLK     SLEEPS
-------- ---------- ---------- ---------- ----------
02C19A38      12721          0      12720       7280
02C19A38      12721          6     159400       7280
02C19A38      12721         11      53436       7280
02C19A38      12721          5      17354       7280
02C19A38      12721          6     140193       7280
02C19A38      12721          7      17773       7280
02C19A38      12721          7      36980       7280

7 rows selected.


from the above output is sleeps are distributed evenly ? and also i see the
file# 0  i don't have any data file with file_id 0 ,can you please explain how
you interpret that.