Single-task message wating for 30 seconds

Question

Hi guys,
I encountered some production issues: in some moments some random sql are taking a lot of time to excute.
I enabled the sql_trace on the system and managed to catch an sql that raises issues. this is the trace parsed with tk prof:
SELECT NVL(RUNNING_TOTALS_AMOUNT, 0)
FROM
ACT_TOTALS_TO_DATE WHERE ACCOUNT_NO = :B1
call     count       cpu    elapsed       disk      query    current        rows
Parse        1      0.00      29.35          0          0          1           0
Execute      2      0.00       0.00          0          0          1           0
Fetch        2      0.00       0.00          0          4          0           0
total        5      0.00      29.35          0          4          2           0
Misses in library cache during parse: 0
Optimizer mode: RULE
Parsing user id: 158     (recursive depth: 2)
Rows     Row Source Operation
      0  SORT GROUP BY (cr=4 pr=0 pw=0 time=62 us)
      0   MERGE JOIN OUTER (cr=4 pr=0 pw=0 time=38 us)
      0    TABLE ACCESS BY INDEX ROWID ACT_TOTALS_TO_DATE_MRC (cr=4 pr=0 pw=0 time=33 us)
      0     INDEX UNIQUE SCAN IDX_ATTDM_ACC_NO (cr=4 pr=0 pw=0 time=29 us)(object id 233866)
      0    REMOTE  ACT_TOTALS_TO_DATE_CALLS (cr=0 pr=0 pw=0 time=0 us)
Elapsed times include waiting on following events:
  Event waited on                             Times   Max. Wait  Total Waited
  ----------------------------------------   Waited  ----------  ------------
  single-task message                             1       29.34         29.34
  SQL*Net message to dblink                       9        0.00          0.00
  SQL*Net message from dblink                     9        0.00          0.00
********************************************************************************usually this SQLs executes very good. the execution plan is what I expect. if I execute it on that machine it works fine.
this SQL is executed by a pl/sql procedure. it ussualy works fine, but at that moment the entire procedure took around 30 seconds.
ACT_TOTALS_TO_DATE is a view that uses table from another instance. the view uses db links.
this problem appears on random SQLs and the parse time is huge.
can you please tell me what is this wait?:
  Event waited on                             Times   Max. Wait  Total Waited
  ----------------------------------------   Waited  ----------  ------------
  single-task message                             1       29.34         29.34this is the wait line from raw trace file:
WAIT #18: nam='single-task message' ela= 29347212 p1=0 p2=0 p3=0 obj#=-1 tim=1272918022974421
I can admit that our code is not well build and it is not using BIND very good. I encountered a lot of hard parses and with this I have latches on shared pool.
but can this explain this behavior?
Thanks for your time.
Edited by: alinux on Apr 22, 2011 6:03 PM

Answer 1

Guys,
how can a SQL wait 30 seconds in PARSE phase of its execution for:
'single-task message' wait?
or
In my trace file I have other SQL that have long parsing time but no wait related to these:
SELECT y, aG,b
FROM
xxx WHERE CONTRACT_ID = :B1
call     count       cpu    elapsed       disk      query    current        rows
Parse        1      5.14       5.02          0          0          0           0
Execute      2      1.41       1.37          0          0          0           0
Fetch        2      0.00       0.00          0          6          0           2
total        5      6.55       6.40          0          6          0           2
Misses in library cache during parse: 0
Optimizer mode: RULE
Parsing user id: 158     (recursive depth: 2)
Rows     Row Source Operation
      2  TABLE ACCESS BY INDEX ROWID xxx (cr=6 pr=0 pw=0 time=109 us)
      2   INDEX UNIQUE SCAN yyy (cr=4 pr=0 pw=0 time=63 us)(object id 234057)
********************************************************************************Thanks a lot
Edited by: alinux on Apr 29, 2011 5:51 PM