1.oracle的dump理解一 BH buffer header
欢迎转载，转载请标明出处：http://blog.csdn.net/notbaron/article/details/51228444
Oracle的oradebug之蛤蟆不归路
蛤蟆本和ORACLE无关，没想到一个半年的项目，硬是一步一步走上”不归路”，
路越走越黑，不是路变黑了，而是走的人少了，你再也不能随手找到一个他人留下的火把来照明前方的路。偶尔会在道路两旁看到倒下的前人,心中开始彷徨~这是不是一条不归路？不走出去鬼知道~
1 查看dump路径：
SQL>Show parameters dump
其中包含的文件都是[ORACLE_SID]_ora_[PID].trc 格式的。
2 跟踪LGWR
查看LGWR进程PID
SQL> select prc.pid from v$bgprocess bgp,v$processprc where bgp.name='LGWR' and prc.addr=bgp.paddr;
PID
----------
12
进行跟踪：
SQL> oradebug setorapid 12;
Oracle pid: 12, Windows thread id: 11884,image: ORACLE.EXE (LGWR)
挂住LGWR
SQL> oradebug suspend
已处理的语句
继续LGWR
SQL> oradebug resume
已处理的语句
3 DUMP内存结构
获取当前会话的SID
SQL> select sid from v$mystat where rownum=1;
SID
----------
136
或者
SQL> select userenv('SID') from dual;
3.1 Dump当前进行进程
SQL> oradebug setmypid
已处理的语句
查看trace文件的名字
SQL> oradebug tracefile_name;
C:\APP \diag\rdbms\toaddb\toaddb\trace\toaddb_ora_10780.trc
为了便于查找，修改名字，如下
SQL> alter session set tracefile_identifier=test;
会话已更改。
SQL> oradebug tracefile_name;
C:\APP\diag\rdbms\toaddb\toaddb\trace\toaddb_ora_10780_TEST.trc
3.2 查看DUMP支持的命令
SQL> oradebug dumplist
TRACE_BUFFER_ON
TRACE_BUFFER_OFF
LATCHES
PROCESSSTATE
SYSTEMSTATE
INSTANTIATIONSTATE
REFRESH_OS_STATS
SESSION_STATS_FREELIST
PDBSTATS
SQLNET_SERVER_TRACE
CONTEXTAREA
HANGDIAG_HEADER
HEAPDUMP
HEAPDUMP_ADDR
POKE_ADDRESS
POKE_LENGTH
POKE_VALUE
POKE_VALUE0
GLOBAL_AREA
REALFREEDUMP
DUMP_SGA_METADATA
FLUSH_JAVA_POOL
POOL_SIMULATOR
PGA_DETAIL_GET
PGA_DETAIL_DUMP
PGA_DETAIL_CANCEL
PGA_SUMMARY
MODIFIED_PARAMETERS
CROSSIC
EVENT_TSM_TEST
ERRORSTACK
CALLSTACK
TEST_STACK_DUMP
TEST_GET_CALLER
RECORD_CALLSTACK
EXCEPTION_DUMP
BG_MESSAGES
ENQUEUES
KSTDUMPCURPROC
KSTDUMPALLPROCS
KSTDUMPALLPROCS_CLUSTER
SIMULATE_EOV
KSFQP_LIMIT
KSKDUMPTRACE
DBSCHEDULER
LDAP_USER_DUMP
LDAP_KERNEL_DUMP
DUMP_ALL_OBJSTATS
DUMPGLOBALDATA
HANGANALYZE
HANGANALYZE_PROC
HNGDET_MEM_USAGE_DUMP
DEAD_CLEANUP_STATE
HANGANALYZE_GLOBAL
CBDB_ENTRIES
GES_STATE
RACDUMP
CGS
OCR
CSS
CRS
SYSTEMSTATE_GLOBAL
GIPC
MMAN_ALLOC_MEMORY
MMAN_CREATE_DEF_REQUEST
MMAN_CREATE_IMM_REQUEST
MMAN_IMM_REQUEST
DUMP_ALL_COMP_GRANULE_ADDRS
DUMP_ALL_COMP_GRANULES
DUMP_ALL_REQS
DUMP_TRANSFER_OPS
DUMP_ADV_SNAPSHOTS
ADJUST_SCN
NEXT_SCN_WRAP
CONTROLF
FLUSH_CACHE
FULL_DUMPS
BUFFERS
RECOVERY
SET_TSN_P1
GLOBAL_BUFFER_DUMP
BUFFER
PIN_BLOCKS
BC_SANITY_CHECK
PIN_RANDOM_BLOCKS
SET_NBLOCKS
CHECK_ROREUSE_SANITY
DUMP_PINNED_BUFFER_HISTORY
KCBO_OBJ_CHECK_DUMP
KCB_WORKING_SET_DUMP
KCBS_ADV_INT_DUMP
KCBI_DUMP_FREELIST
SET_AFN
SET_ISTEMPFILE
FLUSH_BUFFER
FLUSH_OBJECT
SCN_AUTO_ROLLOVER_TS_OVERRIDE
REDOLOGS
ARCHIVE_ERROR
LOGHIST
REDOHDR
LOGERROR
OPEN_FILES
DATA_ERR_ON
DATA_READ_ERR_ON
DATA_ERR_OFF
BLK0_FMTCHG
UPDATE_BLOCK0_FORMAT
TR_SET_BLOCK
TR_SET_ALL_BLOCKS
TR_SET_SIDE
TR_CRASH_AFTER_WRITE
TR_READ_ONE_SIDE
TR_CORRUPT_ONE_SIDE
TR_RESET_NORMAL
TEST_DB_ROBUSTNESS
LOCKS
GC_ELEMENTS
FILE_HDRS
KRB_CORRUPT_INTERVAL
KRB_CORRUPT_SIZE
KRB_CORRUPT_REPEAT
KRB_CORRUPT_OFFSET
KRB_PIECE_FAIL
KRB_OPTIONS
KRB_FAIL_INPUT_FILENO
KRB_SIMULATE_NODE_AFFINITY
KRB_BSET_DAYS
KRB_SET_TIME_SWITCH
KRB_OVERWRITE_ACTION
KRB_CORRUPT_SPHEADER_INTERVAL
KRB_CORRUPT_SPHEADER_REPEAT
KRB_CORRUPT_SPBITMAP_INTERVAL
KRB_CORRUPT_SPBITMAP_REPEAT
KRB_CORRUPT_SPBAD_INTERVAL
KRB_CORRUPT_SPBAD_SIGNAL
KRB_UNUSED_OPTION
KRBMRSR_LIMIT
KRBMROR_LIMIT
KBRS_TRACE
KRDRSBF
KRA_OPTIONS
FBTAIL
FBINC
FBHDR
FLASHBACK_GEN
KTPR_DEBUG
IMDB_PINNED_BUFFER_HISTORY
DUMP_TEMP
DROP_SEGMENTS
TEST_SPACEBG
TREEDUMP
LONGF_CREATE
KDLIDMP
ROW_CACHE
LIBRARY_CACHE
LIBRARY_CACHE_OBJECT
CURSORDUMP
CURSORTRACE
CURSOR_STATS
XS_SESSION_STATE
SHARED_SERVER_STATE
LISTENER_REGISTRATION
JAVAINFO
KXFPCLEARSTATS
KXFPDUMPTRACE
KXFPBLATCHTEST
KXFXSLAVESTATE
KXFXCURSORSTATE
KXFRHASHMAP
WORKAREATAB_DUMP
KUPPLATCHTEST
OBJECT_CACHE
SAVEPOINTS
RULESETDUMP
RULESETDUMP_ADDR
FAILOVER
OLAP_DUMP
ASMDISK_ERR_ON
ASMDISK_READ_ERR_ON
ASMDISK_ERR_OFF
IOERREMUL
IOERREMULRNG
SELFTESTASM
ALRT_TEST
AWR_TEST
AWR_FLUSH_TABLE_ON
AWR_FLUSH_TABLE_OFF
ASHDUMP
ASHDUMPSECONDS
MMON_TEST
ATSK_TEST
HM_FW_TRACE
HM_FDG_VERS
IR_FW_TRACE
KSDTRADV_TEST
GWM_TRACE
GWM_TEST
一般DUMP命令式两个参数，一个是DUMP的名字，另一个是DUMP的等级。
等级的话越高DUMP的越多。
3.3 Dump buffers
可以dump 缓存，缓存头，变量连接链。
只有4种：
1＝缓存头
2＝1 +裸块＋交易头
３＝２ +完成的块
４＝工作数据集，缓存头，hashchain上连接的裸块
SQL> oradebug dump buffers 1
已处理的语句
SQL> oradebug tracefile_name
C:\APP\diag\rdbms\toaddb\toaddb\trace\toaddb_ora_10780_TEST.trc
等级1，蛤蟆的DUMP出来大小是8MB。
等级2，DUMP出来有470MB了，和使用的内存大小相关
查看SGA中大小：
SQL> select component,current_size,min_size fromv$sga_dynamic_components where component like '%buffer cache%';
COMPONENT CURRENT_SIZE
----------------------------------------------------------------------------
MIN_SIZE
----------
DEFAULT buffer cache 520093696
423624704
基本和BUFFERCACHE大小一致。
3.3.1 Buffer Head dump
我们查看下等级1 的TRC文件如下：
显示了trc的路径，数据块的版本，数据块所在的操作系统，CPU型号，内存 大小
进程亲和度，数据块实例名字，进程数量，进程ID，
Dump所使用的命令，然后是SGA中所有的bufferhead了。
Trace fileC:\APP\diag\rdbms\toaddb\toaddb\trace\toaddb_ora_12328.trc
Oracle Database 12c Enterprise EditionRelease 12.1.0.2.0 - 64bit Production
With the Partitioning, OLAP, AdvancedAnalytics and Real Application Testing options
Windows NT Version V6.1 Service Pack 1
CPU : 4 - type 8664, 4 PhysicalCores
Process Affinity : 0x0x0000000000000000
Memory (Avail/Total): Ph:997M/6135M,Ph+PgF:6552M/12269M
Instance name: toaddb
Redo thread mounted by this instance: 1
Oracle process number: 50
Windows thread id: 12328, image: ORACLE.EXE(SHAD)
*** 2016-04-20 17:08:51.948
*** SESSION ID:(244.56563) 2016-04-2017:08:51.948
*** CLIENT ID:() 2016-04-20 17:08:51.948
*** SERVICE NAME:(SYS$USERS) 2016-04-2017:08:51.948
*** MODULE NAME:(sqlplus.exe) 2016-04-2017:08:51.948
*** CLIENT DRIVER:(SQL*PLUS) 2016-04-2017:08:51.948
*** ACTION NAME:() 2016-04-20 17:08:51.948
Processing Oradebug command 'setmypid'
*** 2016-04-20 17:08:51.948
Oradebug command 'setmypid' console output:<none>
*** 2016-04-20 17:09:03.755
Processing Oradebug command'tracefile_name'
*** 2016-04-20 17:09:03.755
Oradebug command 'tracefile_name' consoleoutput:
C:\APP\diag\rdbms\toaddb\toaddb\trace\toaddb_ora_12328.trc
*** 2016-04-20 17:09:31.208
Processing Oradebug command 'dump buffers1'
Dump of buffer cache at level 1 for pdb=0tsn=65535 rdba=0
BH (0x7ff183e76d8) file#: 1 rdba:0x0040403d (1/16445) class: 1 ba: 0x7ff181c0000
set: 10 pool: 3 bsz: 8192 bsi: 0 sflg: 2 pwc: 0,0
dbwrid: 0 obj: 18 objn: 18 tsn: [0/0] afn: 1 hint: f
hash: [0x7ff2cb62d98,0x7ff2cb62d98] lru: [0x7ff1afdc000,0x7ff1b3fab40]
ckptq: [NULL] fileq: [NULL]
objq: [0x7ff187d9d28,0x7ff17fdf368] objaq: [0x7ff187d9d38,0x7ff17fdf378]
st:XCURRENT md: NULL fpin: 'kdswh11: kdst_fetch' fscn: 0x0.23b419 tch: 96
flags: only_sequential_access
LRBA: [0x0.0.0] LSCN: [0x0.0] HSCN: [0xffff.ffffffff] HSUB: [65535]
BH (0x7ff19fe6f58) file#: 1 rdba:0x0040daf0 (1/56048) class: 1 ba: 0x7ff19db4000
set: 9 pool: 3 bsz: 8192 bsi: 0 sflg: 2 pwc: 0,0
dbwrid: 0 obj: 37 objn: 37 tsn: [0/0] afn: 1 hint: f
hash: [0x7ff2cb62e08,0x7ff2cb62e08] lru: [0x7ff19fe7180,0x7ff19fe6f00]
ckptq: [NULL] fileq: [NULL]
objq: [0x7ff19fe71a8,0x7ff19fe6de8] objaq: [0x7ff19fe71b8,0x7ff19fe6df8]
st:XCURRENT md: NULL fpin: 'kdiwh100: kdircys' fscn: 0x0.23b281 tch: 6
flags: only_sequential_access
LRBA: [0x0.0.0] LSCN: [0x0.0] HSCN: [0xffff.ffffffff] HSUB: [65535]
…….(此处省略13W行字)
BH (0x7ff1bfe76d8) file#: 1 rdba:0x0040e508 (1/58632) class: 1 ba: 0x7ff1bdc0000
set: 12 pool: 3 bsz: 8192 bsi: 0 sflg: 2 pwc: 0,0
dbwrid: 0 obj: 36 objn: 36 tsn: [0/0] afn: 1 hint: f
hash: [0x7ff2c6567f8,0x7ff2c6567f8] lru: [0x7ff1bfe7900,0x7ff1bfe7680]
ckptq: [NULL] fileq: [NULL]
objq: [0x7ff1bfe7928,0x7ff1bfe76a8] objaq: [0x7ff1bfe7938,0x7ff1bfe76b8]
st:XCURRENT md: NULL fpin: 'kdiwh100: kdircys' fscn: 0x0.23b5ef tch: 17
flags: only_sequential_access
LRBA: [0x0.0.0] LSCN: [0x0.0] HSCN: [0xffff.ffffffff] HSUB: [65535]
*** 2016-04-20 17:09:33.074
Oradebug command 'dump buffers 1' consoleoutput: <none>
3.3.2 Buffer head 介绍
其中摘取一个BH如下：
BH (0x7ff183e76d8) file#: 1 rdba:0x0040403d (1/16445) class: 1 ba: 0x7ff181c0000
set: 10 pool: 3 bsz: 8192 bsi: 0 sflg: 2 pwc: 0,0
dbwrid: 0 obj: 18 objn: 18 tsn: [0/0] afn: 1 hint: f
hash: [0x7ff2cb62d98,0x7ff2cb62d98] lru: [0x7ff1afdc000,0x7ff1b3fab40]
ckptq: [NULL] fileq: [NULL]
objq: [0x7ff187d9d28,0x7ff17fdf368] objaq: [0x7ff187d9d38,0x7ff17fdf378]
st:XCURRENT md: NULL fpin: 'kdswh11: kdst_fetch' fscn: 0x0.23b419 tch: 96
flags: only_sequential_access
LRBA: [0x0.0.0] LSCN: [0x0.0] HSCN: [0xffff.ffffffff] HSUB: [65535]
逻辑结构如下图1:
DUMP出来的BH共9行
第一行：0x7ff183e76d8是BH的哈希值，file#: 1 是块所在的文件，通过v$dbfile可以查看。rdba:0x0040403d (1/16445)是rowid中的相对文件号，可以通过如下命令获取十进制数：
SQL> selectdbms_utility.make_data_block_address(1,16445) from dual;
DBMS_UTILITY.MAKE_DATA_BLOCK_ADDRESS(1,16445)
---------------------------------------------
4210749
4210749的16进制就是0x0040403d，前十位004是文件号，后二十位0403d是块号。class: 1 表示bufferheader对应的类型，
class:表示buffer header对应block的类型:
1=data block, 9=2nd level bmb,
2=sort block, 10=3rd level bmb,
3=save undo block, 11=bitmap block,
4=segment header, 12=bitmap index block,
5=save undo header, 13=unused,
6=free list, 14=undo header,
7=extent map, 15=undo block
ba: 0x7ff181c0000 是BUFFER中block address，是块在内存中的物理地址
第二行：set: 10
pool: 3 bsz: 8192 表示块大小，bsi: 0 sflg: 2 pwc: 0,0
第三行
dbwrid: 0 进程dbwr的id obj: 18块上数据对应在哪个对象里 objn: 18tsn: [0/0] 表空间编码 afn: 1 hint: f
第四行
hash:[0x7ff2cb62d98,0x7ff2cb62d98] 下一个和前一个BH的HASH值
lru:[0x7ff1afdc000,0x7ff1b3fab40]下一个及前一个BH的在LRU链表上的HASH值
第五行
ckptq: [NULL] 在检查点队列上的HASH值fileq: [NULL]在文件队列上的HASH值
第六行
objq:[0x7ff187d9d28,0x7ff17fdf368]是对象队列HASH值
objaq:[0x7ff187d9d38,0x7ff17fdf378]是辅助对象队列HASH值
第七行
st: XCURRENT 表示块状态，具体如下：
state:
0, FREE, no valid block image
1, XCUR, a current mode block, exclusive tothis instance 正在被当前的instance独占。
2, SCUR, a current mode block, shared withother instances正在被当前的instance共享
3, CR, a consistent read (stale) blockimage 一致读
4, READ, buffer is reserved for a blockbeing read from disk 正在从磁盘上读取块
5, MREC, a block in media recovery mode 处于介质恢复模式
6, IREC, a block in instance (crash)recovery mode处于实例恢复模式
md: NULL fpin: 'kdswh11: kdst_fetch' fscn:0x0.23b419
tch: 96 buffer 的访问次数
第八行
flags: only_sequential_access
flag中，每位代表如下含义：
0 buffer_dirty 14 stale
1 notify_after_change 15 deferred_ping
2 mod_started 16 direct_access
3 block_has_been_logged 17 hash_chain_dump
4 temp_data 18 ignore_redo
5 being_written 19 only_sequential_access
6 waiting_for_write 20 prefetched_block
7 multiple_waiters 21 block_written_once
8 recovery_reading 22 logically_flushed
9 unlink_from_lock 23 resilvered_already
10 down_grade_lock 25 redo_since_read
11 clone_being_written 29plugged_from_foreign_db
12 reading_as_CR 30 flush_after_writing
13 gotten_in_current_mode
第九行
LRBA: [0x0.0.0] L表示Low，也就是第一次被修改的时候的RBA
LSCN: [0x0.0] L表示Low，H表示High
HSCN: [0xffff.ffffffff] HSUB: [65535]
汇总总的BH数量，蛤蟆此处是 15468个，合计大小是15468 x 8192=126713856=120MB
说明BUFFER CACHE使用了120MB
DUMP一下，量太大，分个WORD继续。
再分享一下我老师大神的人工智能教程吧。零基础！通俗易懂！风趣幽默！还带黄段子！希望你也加入到我们人工智能的队伍中来！https://blog.csdn.net/jiangjunshow