Wait Event

[elmaazouz kamal]

Hello Every,
Does any one have a documentation about the folowing wait system events:
 
PX Deq: Test for msg
virtual circuit status                                                 
dispatcher timer

Best regards.

[ddf]

What problems are you attempting to solve?

The PX Deq wait is a parallel query coordinator wait for messages to
dequeue from the slaves

Oracle's shared server configuration uses a common queue for incoming
requests (calls) to shared servers and other queues for responses
going back through a process known as a dispatcher. These queues and
their clients constitute the virtual circuit. When a process is ready
to take a request or response from its queue, but finds that there is
no request or response to take, then it waits for the 'virtual circuit
status' to change

The dispatcher timer is started when the shared server dispatcher or
dispatchers are literally waiting for something to do.

Apparently you're working a problem using Shared Server network
configuration. What might that issue be?



David Fitzjarrell

[elmaazouz kamal]

I'm trying to optimize the redo apply process in a physical standby dataguard configuration using the best practices document from oracle.
The three wait events recently posted was the top wait system events, for the dispatcher timer, I disabled dispatchers on the Both environnements.
But the event "PX Deq: Test for msg" is still on the top events, especially I'm using parallel recovery on a 2 processor dual core standby Server using the following query:

ALTER DATABASE RECOVER MANAGED STANDBY DATABASE USING CURRENT LOGFILE DISCONNECT FROM SESSION NODELAY PARALLEL 2;
 
Here  is the new result :
select event, total_waits,
round(time_waited/100) "TIME(s)",
average_wait*10 "AVG(ms)",
TO_CHAR(SYSDATE, 'DD-MON-YYYY HH:MI:SS') time
from v$system_event where time_waited > 100
order by time_waited;
SQL> SQL>   2    3    4    5    6
EVENT                                                            TOTAL_WAITS    TIME(s)    AVG(ms) TIME
---------------------------------------------------------------- ----------- ---------- ---------- --------------------
log file single write                                                     58          1       17.4 31-DEC-2008 07:33:33
RFS dispatch                                                            1505          2          1 31-DEC-2008 07:33:33
enq: PS - contention                                                       3          2      544.3 31-DEC-2008 07:33:33
MRP wait on state change                                                   2          2        977 31-DEC-2008 07:33:33
rdbms ipc reply                                                          160          4       22.1 31-DEC-2008 07:33:33
control file heartbeat                                                     1          4     3907.3 31-DEC-2008 07:33:33
ASM background starting                                                    4          6     1481.7 31-DEC-2008 07:33:33
KSV master wait                                                          873         10       11.1 31-DEC-2008 07:33:33
MRP wait on process start                                                  2         10     4882.8 31-DEC-2008 07:33:33
RFS close                                                                 14         10      712.3 31-DEC-2008 07:33:33
control file sequential read                                           24911         10         .4 31-DEC-2008 07:33:33

EVENT                                                            TOTAL_WAITS    TIME(s)    AVG(ms) TIME
---------------------------------------------------------------- ----------- ---------- ---------- --------------------
MRP wait on startup clear                                                 12         12      977.4 31-DEC-2008 07:33:33
RFS create                                                                17         17     1000.4 31-DEC-2008 07:33:33
PX Idle Wait                                                              14         22     1595.3 31-DEC-2008 07:33:33
ARCH wait on c/f tx acquire 2                                             23         22      977.6 31-DEC-2008 07:33:33
enq: CF - contention                                                      52         23      449.3 31-DEC-2008 07:33:33
PX Deq Credit: need buffer                                              1453         28       19.5 31-DEC-2008 07:33:33
RFS ping                                                                 290         48      164.4 31-DEC-2008 07:33:33
log file sequential read                                                5973         79       13.2 31-DEC-2008 07:33:33
control file parallel write                                             3850         82       21.3 31-DEC-2008 07:33:33
Log archive I/O                                                        17476        294       16.8 31-DEC-2008 07:33:33
PX Deq Credit: send blkd                                               13667        445       32.5 31-DEC-2008 07:33:33

EVENT                                                            TOTAL_WAITS    TIME(s)    AVG(ms) TIME
---------------------------------------------------------------- ----------- ---------- ---------- --------------------
PX Deq: Par Recov Reply                                                  531        717       1351 31-DEC-2008 07:33:33
SQL*Net more data from client                                            751        768       1023 31-DEC-2008 07:33:33
checkpoint completed                                                     268       1268     4731.5 31-DEC-2008 07:33:33
RFS write                                                               1169       1770     1514.4 31-DEC-2008 07:33:33
smon timer                                                                21       5624   267786.9 31-DEC-2008 07:33:33
ASM background timer                                                    3041       6067     1995.1 31-DEC-2008 07:33:33
pmon timer                                                              2084       6071     2913.1 31-DEC-2008 07:33:33
class slave wait                                                         856       6997     8173.5 31-DEC-2008 07:33:33
PX Deq: Test for msg                                                   16333      10080      617.2 31-DEC-2008 07:33:33
SQL*Net message from client                                             2217      24901    11232.1 31-DEC-2008 07:33:33
rdbms ipc message                                                      16793      93292     5555.4 31-DEC-2008 07:33:33


the event SQL*Net message from client   seems to be a network bandwidth issue.
how about the PX Deq: Test for msg   and rdbms ipc message ??
Best regards.

--
Amicalement
Kamal ELMAAZOUZ

[ddf]

Comments embedded.

On Dec 31, 2:06 pm, "elmaazouz kamal" <elmaaz...@gmail.com> wrote:
> I'm trying to optimize the redo apply process in a physical standby
> dataguard configuration using the best practices document from oracle.
> The three wait events recently posted was the top wait system events, for
> the dispatcher timer, I disabled dispatchers on the Both environnements.
> But the event "PX Deq: Test for msg" is still on the top events, especially
> I'm using parallel recovery on a 2 processor dual core standby Server using
> the following query:
>

> *ALTER DATABASE RECOVER MANAGED STANDBY DATABASE USING CURRENT LOGFILE

> the event **SQL*Net message from client   seems to be a network bandwidth
> issue.
> how about the **PX Deq: Test for msg  

I've explained this one in my last post. It's the parallel
coordinator waiting to dequeue a message from a slave.

> and **rdbms ipc message* *??

This is the cumulative wait time from idle LGWR, DBWR and LMS0
processes. So your log writer, database writer and cache fusion
(LMS0) processes (a component of RAC) are idle.

> Best regards.
> **
> *

> Kamal ELMAAZOUZ- Hide quoted text -
>
> - Show quoted text -


David Fitzjarrell

[jgar the jorrible]

On Dec 31 2008, 11:04 am, "elmaazouz kamal" <elmaaz...@gmail.com>
wrote:

Also see:

http://oracledoug.com/px8.html

http://www.rittmanmead.com/2005/04/27/tracing-parallel-execution/

Pay attention to which metric is reset when.

http://tonguc.wordpress.com/2008/05/05/data-guard-redo-apply-and-media-recovery-best-practices-10g/

http://www.juliandyke.com/Diagnostics/Events/EventReference.html#10390

Also, show parameter parallel.

I notice the PX Deq: Test for msg is (barely) documented in O9, but
not in 10. I have no idea if this is a doc bug or if what you've done
has set off some strange old code.

"...events which you can do nothing about so don't bother looking at
them". old metalink note.

Make of what you will not-a-bug 6892603 - doesn't really explain
anything, but looks like someone tried to set down parallelism for
some dg performance problem, and saw that wait.


jg
--
@home.com is bogus.
What happens when you depend on drive mirroring as a backup:
http://journalspace.com/this_is_the_way_the_world_ends/not_with_a_bang_but_a_whimper.html