Control file parallel write
当数据库中有多个控制文件的拷贝时,Oracle 需要保证信息同步地写到各个控制文件当中,这是一个并行的物理操作过程,因为称为控制文件并行写,当发生这样的操作时,就会产生control file parallel write等待事件。
控制文件频繁写入的原因很多,比如:
· 日志切换太过频繁,导致控制文件信息相应地需要频繁更新。
· 系统I/O 出现瓶颈,导致所有I/O出现等待。
当系统出现日志切换过于频繁的情形时,可以考虑适当地增大日志文件的大小来降低日志切换频率。
当系统出现大量的control file parallel write 等待事件时,可以通过比如降低控制文件的拷贝数量,将控制文件的拷贝存放在不同的物理磁盘上的方式来缓解I/O 争用。
这个等待事件包含三个参数:
Files: Oracle 要写入的控制文件个数。
Blocks: 写入控制文件的数据块数目。
Requests: 写入控制请求的I/O 次数。
Control file sequential read
当数据库需要读取控制文件上的信息时,会出现这个等待事件,因为控制文件的信息是顺序写的,所以读取的时候也是顺序的,因此称为控制文件顺序读,它经常发生在以下情况:
备份控制文件
RAC 环境下不同实例之间控制文件的信息共享
读取控制文件的文件头信息
读取控制文件其他信息
这个等待事件有三个参数:
File#: 要读取信息的控制文件的文件号。
Block#: 读取控制文件信息的起始数据块号。
Blocks: 需要读取的控制文件数据块数目。
模拟案例:
1、模拟事务处理
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
|
16
:
59
:
48
SYS@ test1>select username,sid,serial#
from
v$session
where
username is
not
null
;
USERNAME SID SERIAL#
------------------------------ ---------- ----------
SCOTT
1
7
SYS
34
4
TOM
41
3
16
:
59
:
57
SCOTT@ test1>begin
17
:
04
:
05
2
for
i
in
1.
.1000000
loop
17
:
04
:
05
3
execute immediate
'insert into t1(id) values ('
||i||
')'
;
17
:
04
:
05
4
end loop;
17
:
04
:
05
5
end;
17
:
04
:
05
6
/
17
:
00
:
00
TOM@ test1> begin
17
:
04
:
15
2
for
i
in
1.
.1000000
loop
17
:
04
:
15
3
execute immediate
'insert into scott.t1(id) values ('
||i||
')'
;
17
:
04
:
15
4
end loop;
17
:
04
:
15
5
end;
17
:
04
:
15
6
/
|
2、查看等待事件:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
|
17
:
05
:
40
SYS@ test1>select sid,event,p1,p2,p3,WAIT_TIME
from
v$session_wait
where
sid
in
(
1
,
41
);
SID EVENT P1 P2 P3 WAIT_TIME
---------- -------------------------------------------------- ---------- ---------- -----
1
latch: redo allocation
1993106152
182
0
-1
41
latch: cache buffers chains
2011026752
150
0
-1
Elapsed:
00
:
00
:
00.01
17
:
06
:
11
SYS@ test1>r
1
* select sid,event,p1,p2,p3,WAIT_TIME
from
v$session_wait
where
sid
in
(
1
,
41
)
SID EVENT P1 P2 P3 WAIT_TIME
---------- -------------------------------------------------- ---------- ---------- -------
1
log file switch (checkpoint incomplete)
0
0
0
3
41
buffer busy waits
4
4048
1
94
Elapsed:
00
:
00
:
00.01
17
:
06
:
16
SYS@ test1>r
1
* select sid,event,p1,p2,p3,WAIT_TIME
from
v$session_wait
where
sid
in
(
1
,
41
)
SID EVENT P1 P2 P3 WAIT_TIME
---------- -------------------------------------------------- ---------- ---------- -----
1
log file switch (checkpoint incomplete)
0
0
0
0
41
log file switch (checkpoint incomplete)
0
0
0
0
Elapsed:
00
:
00
:
00.02
17
:
06
:
18
SYS@ test1>r
1
* select sid,event,p1,p2,p3,WAIT_TIME
from
v$session_wait
where
sid
in
(
1
,
41
)
SID EVENT P1 P2 P3 WAIT_TIME
---------- -------------------------------------------------- ---------- ---------- ---------- ----------
1
log buffer space
0
0
0
0
41
log buffer space
0
0
0
0
Elapsed:
00
:
00
:
00.01
17
:
06
:
20
SYS@ test1>r
1
* select sid,event,p1,p2,p3,WAIT_TIME
from
v$session_wait
where
sid
in
(
1
,
41
)
SID EVENT P1 P2 P3 WAIT_TIME
---------- -------------------------------------------------- ---------- ---------- -----
1
log file switch (checkpoint incomplete)
0
0
0
134
41
log file switch (checkpoint incomplete)
0
0
0
3
Elapsed:
00
:
00
:
00.01
17
:
06
:
22
SYS@ test1>r
1
* select sid,event,p1,p2,p3,WAIT_TIME
from
v$session_wait
where
sid
in
(
1
,
41
)
SID EVENT P1 P2 P3 WAIT_TIME
---------- -------------------------------------------------- ---------- ---------- ---------- ----------
1
latch: cache buffers chains
2011026752
150
0
-1
41
latch: cache buffers chains
2011026752
150
0
-1
Elapsed:
00
:
00
:
00.01
|
3、查看关于controlfile的事件
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
|
17
:
06
:
24
SYS@ test1>select EVENT ,TOTAL_WAITS,TIME_WAITED
from
v$system_event
where
event like
'%control%'
;
EVENT TOTAL_WAITS TIME_WAITED
-------------------------------------------------- ----------- -----------
control file sequential read
51623
326
control file parallel write
6327
5102
control file heartbeat
1
400
Elapsed:
00
:
00
:
00.02
17
:
07
:
31
SYS@ test1>r
1
* select EVENT ,TOTAL_WAITS,TIME_WAITED
from
v$system_event
where
event like
'%control%'
EVENT TOTAL_WAITS TIME_WAITED
-------------------------------------------------- ----------- -----------
control file sequential read
52314
329
control file parallel write
6410
5164
control file heartbeat
1
400
Elapsed:
00
:
00
:
00.03
17
:
07
:
41
SYS@ test1>r
1
* select EVENT ,TOTAL_WAITS,TIME_WAITED
from
v$system_event
where
event like
'%control%'
EVENT TOTAL_WAITS TIME_WAITED
-------------------------------------------------- ----------- -----------
control file sequential read
52597
330
control file parallel write
6433
5177
control file heartbeat
1
400
Elapsed:
00
:
00
:
00.02
|
4、查看告警日志:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
|
Beginning log switch checkpoint up to RBA [
0x438.
2.10
], SCN:
1762466
Thread
1
advanced to log sequence
1080
(LGWR switch)
Current log#
3
seq#
1080
mem#
0
: /u01/app/oracle/oradata/redo03a.log
Fri Jul
25
17
:
10
:
26
2014
Archived Log entry
1065
added
for
thread
1
sequence
1079
ID
0x4906548b
dest
1
:
Thread
1
cannot allocate
new
log, sequence
1081
Checkpoint
not
complete
Current log#
3
seq#
1080
mem#
0
: /u01/app/oracle/oradata/redo03a.log
Completed checkpoint up to RBA [
0x438.
2.10
], SCN:
1762466
Beginning log switch checkpoint up to RBA [
0x439.
2.10
], SCN:
1762522
Thread
1
advanced to log sequence
1081
(LGWR switch)
Current log#
4
seq#
1081
mem#
0
: /u01/app/oracle/oradata/redo04a.log
Fri Jul
25
17
:
10
:
29
2014
Archived Log entry
1066
added
for
thread
1
sequence
1080
ID
0x4906548b
dest
1
:
Thread
1
cannot allocate
new
log, sequence
1082
Checkpoint
not
complete
Current log#
4
seq#
1081
mem#
0
: /u01/app/oracle/oradata/redo04a.log
Completed checkpoint up to RBA [
0x439.
2.10
], SCN:
1762522
Fri Jul
25
17
:
10
:
32
2014
Beginning log switch checkpoint up to RBA [
0x43a.
2.10
], SCN:
1762575
Thread
1
advanced to log sequence
1082
(LGWR switch)
Current log#
3
seq#
1082
mem#
0
: /u01/app/oracle/oradata/redo03a.log
Archived Log entry
1067
added
for
thread
1
sequence
1081
ID
0x4906548b
dest
1
:
|
5、查看日志组信息
1
2
3
4
5
6
7
|
17
:
16
:
22
SYS@ test1>select group#,sequence#,status,bytes/
1024
/
1024
from
v$log;
GROUP# SEQUENCE# STATUS BYTES/
1024
/
1024
---------- ---------- ---------------- ---------------
3
1082
CURRENT
4
4
1081
INACTIVE
4
Elapsed:
00
:
00
:
00.03
日志组size太小,日志组数量较少,日志切换太频繁,导致产生大量的控制文件读和写,调整日志组的size,并增加日志组的个数。
|
本文转自 客居天涯 51CTO博客,原文链接:http://blog.51cto.com/tiany/1530190,如需转载请自行联系原作者