bobjoy

42 积分	0 好友	1 主题

发消息

ORA-00480: LCK* process terminated 实例crash

1^#

发表于 2011-12-20 14:53:47 | 查看: 9093| 回复: 4

hi,
环境:10.2.0.5 two-node rac for aix 6.1
现象：1号节点instance crash，详见最后一次启动日志
之前开发人员杀过用户进程，过程没有看到。

日志：
Mon Dec 19 15:49:10 GMT+08:00 2011Errors in file /oracle/admin/epmsc/bdump/epmsc1_pmon_20120358.trc:
ORA-00480: LCK* process terminated with error
Mon Dec 19 15:49:10 GMT+08:00 2011PMON: terminating instance due to error 480
Mon Dec 19 15:49:10 GMT+08:00 2011System state dump is made for local instance
System State dumped to trace file /oracle/admin/epmsc/bdump/epmsc1_diag_20972238.trc
Mon Dec 19 15:49:12 GMT+08:00 2011Shutting down instance (abort)

metalink对ORA-00480描述也较少，各位大侠有啥发现？

复件 ora_log.rar

1.87 MB, 下载次数: 1342

log and trace

分享0

收藏0 回复只看该作者道具举报

Maclean Liu(刘相兵

2^#

发表于 2011-12-20 15:16:29

cd /oracle/admin/epmsc/bdump/
ls -ltr *22479184*
ls -ltr *lck*

将相关的lck trace上传一下

回复只看该作者道具举报

Maclean Liu(刘相兵

3^#

发表于 2011-12-20 16:02:03

分析：

LCK0进程的近期等待事件是ksxr poll remote instances，进程号 PID= 65 SPID 22479184  状态是DEAD

   SO: 700001ae7555ae0, type: 2, owner: 0, flag: INIT/-/-/0x00
  (process) Oracle pid=65, calls cur/top: 7000019d7f0e428/7000019d7f0e428, flag: (6) SYSTEM
         int error: 0, call error: 0, sess error: 0, txn error 0
  (post info) last post received: 0 0 21
            last post received-location: ksbria
            last process to post me: 700001ae7555ae0 1 6
            last post sent: 0 0 167
            last post sent-location: kqrbtm
            last process posted by me: 700001ae6581b50 31 0
(latch info) wait_event=0 bits=0
Process Group: DEFAULT, pseudo proc: 700001ae561ca38
O/S info: user: oracle, term: UNKNOWN, ospid: 22479184 (DEAD)
OSD pid info: Unix process pid: 22479184, image: oracle@pmscpdba (LCK0)
SO: 700001ae48bac70, type: 4, owner: 700001ae7555ae0, flag: INIT/-/-/0x00
(session) sid: 3211 trans: 0, creator: 700001ae7555ae0, flag: (51) USR/- BSY/-/-/-/-/-
            DID: 0000-0041-00000002, short-term DID: 0000-0000-00000000
            txn branch: 0
            oct: 0, prv: 0, sql: 0, psql: 0, user: 0/SYS
service name: SYS$BACKGROUND
waiting for 'rdbms ipc message' wait_time=0, seconds since wait started=3
            timeout=129, =0, =0
            blocking sess=0x0 seq=17047
Dumping Session Wait History
   for 'ksxr poll remote instances' count=1 wait_time=0.000005 sec
            =0, =0, =0
   for 'rdbms ipc message' count=1 wait_time=0.000358 sec
            timeout=129, =0, =0
   for 'ksxr poll remote instances' count=1 wait_time=0.000006 sec
            =0, =0, =0
   for 'rdbms ipc message' count=1 wait_time=0.000137 sec
            timeout=129, =0, =0
   for 'ksxr poll remote instances' count=1 wait_time=0.000007 sec
            =0, =0, =0
   for 'rdbms ipc message' count=1 wait_time=0.000260 sec
            timeout=129, =0, =0
   for 'ksxr poll remote instances' count=1 wait_time=0.000007 sec
            =0, =0, =0
   for 'rdbms ipc message' count=1 wait_time=0.000323 sec
            timeout=129, =0, =0
   for 'ksxr poll remote instances' count=1 wait_time=0.000005 sec
            =0, =0, =0
   for 'rdbms ipc message' count=1 wait_time=0.000145 sec

最后post LCK0的是 700001ae7555ae0  其自身 LCK0 post LCK0
LCK0最后post 的是 700001ae6581b50

LCK0当时持有大量的 enqueue lock 资源，几百个，其中有十几个类型为CI Cross Instance Enqueue CI:Cross Instance Call Invocation

猜测LCK 进程一直在高负载工作，占用大量的CPU

------------process 0x700001ae9aff608--------------------
proc version    : 0
Local node       : 0
pid             : 22479184
lkp_node       : 0
svr_mode       : 0
proc state       : KJP_NORMAL
Last drm hb acked : 0
Total accesses : 7592925
Imm.  accesses : 7592506
Locks on ASTQ    : 0
Locks Pending AST : 0
Granted locks    : 59683
AST_Q:
PENDING_Q:
GRANTED_Q:
lp 700001ae9e902a0 gl KJUSERPR rp 7000019bb8b9f78 [0x6][0x2],[CI]

0x6 => Test call
   master 0 pid 22479184 bast 1 rseq 20 mseq 0 history 0x9a5
   open opt  KJUSERPROCESS_OWNED
lp 700001ae9e90540 gl KJUSERPR rp 7000019bb8b9d68 [0x1a][0x2],[CI]

1a=> 26  Purge dictionary Object number Cache

   master 0 pid 22479184 bast 1 rseq 19 mseq 0 history 0x9a5
   open opt  KJUSERPROCESS_OWNED
lp 700001ae9e90a80 gl KJUSERPR rp 7000019bb8b9b58 [0x1e][0x2],[CI]

这里 1e => 30 process waiters after row cache requeue
      0x2 => Used to invoke the function in backgroud process

   master 0 pid 22479184 bast 1 rseq 19 mseq 0 history 0x9a5
   open opt  KJUSERPROCESS_OWNED
lp 700001ae9e90fd8 gl KJUSERPR rp 7000019b96217e8 [0x31][0x2],[CI]
   master 1 pid 22479184 bast 1 rseq 16 mseq 0 history 0x95
   open opt  KJUSERPROCESS_OWNED
lp 700001ae9e91518 gl KJUSERPR rp 7000019b6357810 [0x35][0x2],[CI]
   master 1 pid 22479184 bast 1 rseq 17 mseq 0 history 0x95
   open opt  KJUSERPROCESS_OWNED
lp 700001ae9e917d0 gl KJUSERPR rp 7000019cacf75f8 [0x39][0x2],[CI]
   master 1 pid 22479184 bast 1 rseq 22 mseq 0 history 0x95
   open opt  KJUSERPROCESS_OWNED
lp 700001ae9e91a70 gl KJUSERPR rp 7000019becd45b8 [0x41][0x2],[CI]
......................

SO: 700001ae58922c0, type: 4, owner: 700001ae75552f0, flag: INIT/-/-/0x00
(session) sid: 3245 trans: 0, creator: 700001ae75552f0, flag: (100051) USR/- BSY/-/-/-/-/-
            DID: 0001-003A-00000002, short-term DID: 0001-003A-00000003
            txn branch: 0
            oct: 0, prv: 0, sql: 0, psql: 0, user: 0/SYS
service name: SYS$BACKGROUND
waiting for 'smon timer' wait_time=0, seconds since wait started=1
            sleep time=12c, failed=0, =0
            blocking sess=0x0 seq=64966
Dumping Session Wait History
   for 'DFS lock handle' count=1 wait_time=0.000950 sec
            type|mode=43490005, id1=1, id2=2
   for 'DFS lock handle' count=1 wait_time=0.000376 sec
            type|mode=43490005, id1=1, id2=1
   for 'enq: TT - contention' count=1 wait_time=0.000296 sec
            name|mode=54540004, tablespace ID=1, operation=10
   for 'enq: US - contention' count=1 wait_time=0.000428 sec
            name|mode=55530006, undo segment #=29, 0=0
   for 'smon timer' count=1 wait_time=3.119856 sec
            sleep time=12c, failed=0, =0
   for 'smon timer' count=1 wait_time=4.882208 sec
            sleep time=12c, failed=0, =0
   for 'smon timer' count=1 wait_time=2.254375 sec
            sleep time=12c, failed=0, =0
   for 'DFS lock handle' count=1 wait_time=0.001051 sec
            type|mode=43490005, id1=1, id2=2
   for 'DFS lock handle' count=1 wait_time=0.000274 sec
            type|mode=43490005, id1=1, id2=1
   for 'enq: US - contention' count=1 wait_time=0.000496 sec
            name|mode=55530006, undo segment #=3, 0=0
Sampled Session History of session 3245 serial 1

SMON 进程之前在等待'DFS lock handle

id1 => 43490005 CI
id2= >  1    Reuse (checkpoint and invalidate) block range
id3 =>  2,1 Used to invoke the function in backgroud process

回复只看该作者道具举报

Maclean Liu(刘相兵

4^#

发表于 2011-12-20 16:07:22

相关的Bug 信息 <table class="t_table" style="background: " width="98%"><tbody><tr><td>Type</td><td>B - Defect</td><td>Fixed in Product Version</td><td>-</td></tr><tr><td>Severity</td><td>2 - Severe Loss of Service</td><td>Product Version</td><td>10.2.0.4</td></tr><tr><td>Status</td><td>33 - Suspended, Req'd Info not Avail</td><td>Platform</td><td>212 - IBM AIX on POWER Systems (64-bit)</td></tr><tr><td>Created</td><td>17-May-2011</td><td>Platform Version</td><td>6.1</td></tr><tr><td>Updated</td><td>16-Nov-2011</td><td>Base Bug</td><td>-</td></tr><tr><td>Database Version</td><td>10.2.0.4.7</td><td> </td><td> </td></tr><tr><td>Affects Platforms</td><td>Generic</td><td> </td><td> </td></tr><tr><td>Product Source</td><td>Oracle</td><td> </td><td> </td></tr></tbody></table>               <img src="https://support.oracle.com/CSP/ui/images/collapse_white.png?MOS_5.4.0.2.0_GENERIC_111212" alt="" border="0">       Related Products                                      <table class="t_table" style="background: " width="98%"><tbody><tr><td>Line</td><td>Oracle Database Products</td><td>Family</td><td>Oracle Database</td></tr><tr><td>Area</td><td>Oracle Database</td><td>Product</td><td>5 - Oracle Server - Enterprise Edition</td></tr></tbody></table>       Hdr: 12564483 10.2.0.4.7 RDBMS 10.2.0.4 RAC PRODID-5 PORTID-212 Abstract: DATABASE HANG WITH CI-0X1E-0X2 LOCK *** 05/17/11 06:35 pm *** (ADD: Impact/Symptom->DATABASE HANG ) *** 05/17/11 06:35 pm ***         BUG TYPE CHOSEN   ===============   Code      SubComponent: Real Application Clusters   =======================================   DETAILED PROBLEM DESCRIPTION   ============================   2 node RAC 10.2.0.4.7, Database has been hanging for few times. Alert log   report many:   WARNING: inbound connection timed out (ORA-3136)      Latest occurrence is at: May 12 starting from 8:15, till instance shutdown   abort around 9:04.      Proper system state dump collected for the last occurrence.      DIAGNOSTIC ANALYSIS   ===================   per system state dump cvalle1_ora_897314.trc:      PROCESS 16      OSD pid info: Unix process pid: 483334, image: oracle@sdbsora01 (MMON)         waiting for 'DFS lock handle' blocking sess=0x0 seq=4637 wait_time=0   seconds since wait started=213                  type|mode=43490005, id1=1e, id2=2           ----------resource 0x700000413c74e78----------------------        resname       : [0x1e][0x2],[CI]        Local node    : 0        dir_node      : 0        master_node   : 0           GRANTED_Q :        lp 7000004350d38f0 gl KJUSERPR rp 700000413c74e78 [0x1e][0x2],[CI]               ==>同样的CI          master 0 owner 1  bast 0 rseq 3 mseq 0x2 history 0xd4977d8d          open opt  KJUSERNO_XID        CONVERT_Q:        lp 7000004340e3340 gl KJUSERNL rl KJUSEREX rp 700000413c74e78   [0x1e][0x2],[CI]          master 0 pid 483334 bast 0 rseq 4 mseq 0 history 0x49ab549a          convert opt KJUSERNODEADLOCKWAIT KJUSERNODEADLOCKBLOCK      MMON is waiting for CI-0x1e-0x2 lock. This lock is held on inst 2. There is   no resource dump in inst 2 system state cvalle2_ora_8700252.trc. But LCK0   process dump shows it is holding CI-0x1e-0x2:      PROCESS 20:      OSD pid info: Unix process pid: 328170, image: oracle@sdbsora02 (LCK0)     waiting for 'rdbms ipc message' blocking sess=0x0 seq=3266 wait_time=0   seconds since wait started=0                  timeout=fd, =0, =0      Dumping Session Wait History       for 'ksxr poll remote instances' count=1 wait_time=0              => 同样的等待事件                  =0, =0, =0       for 'rdbms ipc message' count=1 wait_time=0                  timeout=106, =0, =0       for 'ksxr poll remote instances' count=1 wait_time=0                  =0, =0, =0         lp 70000043741bb08 gl KJUSERPR rp 7000004131d6588 [0x1e][0x2],[CI]        master 0 pid 328170 bast 1 rseq 51 mseq 0 history 0x495149a5      Its waiting status seems normal.      There are also many user sessions waiting for 'reliable messages' and the   => mmon wait for it   number of sessions waiting for this is increasing.      But ct is on 10.2.0.4.7 already and bug 6148054 - supercede by Bug:7801939,   the fix has already been applied.      Need to find out why LCK0 not releasing the lock and why database hang.      WORKAROUND?   ===========   No      TECHNICAL IMPACT   ================   Causing downtime.      RELATED ISSUES (bugs, forums, RFAs)   ===================================   None. Most bug related with CI-0x1e-0x2 where LCK0 is holder, LCK0 is   waiting for latch etc, not like this case.      HOW OFTEN DOES THE ISSUE REPRODUCE AT CUSTOMER SITE?   ====================================================   Intermittent      DOES THE ISSUE REPRODUCE INTERNALLY?   ====================================   Not attempted      EXPLAIN WHY THE ISSUE WAS NOT TESTED INTERNALLY.   ================================================   not feasible      IS A TESTCASE AVAILABLE?   ========================   No 官方认为Bug 7801939在 10.2.0.5 中已被修复 Bug 7801939  Contention for "channel operations parent latch" child latch This note gives a brief overview bug 7801939. The content was last updated on: 18-NOV-2010 Click <a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1" target="_blank">here</a> for details of each of the sections below. Affects:<blockquote><table class="t_table" style="background: "><tbody><tr><td>Product (Component)</td><td>Oracle Server  (Rdbms)</td></tr><tr><td>Range of versions believed to be affected</td><td>Versions BELOW 11.2</td></tr><tr><td>Versions confirmed as being affected</td><td><ul><li><a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1#AFFECTS_10.2.0.4" target="_blank">10.2.0.4</a></li></ul></td></tr><tr><td>Platforms affected</td><td>Generic (all / most platforms affected)</td></tr></tbody></table></blockquote>Fixed:<blockquote><table class="t_table" style="background: "><tbody><tr><td>This issue is fixed in</td><td><ul><li><a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1#FIXED_11.2.0.1" target="_blank">11.2.0.1 (Base Release)</a></li><li><a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1#FIXED_10.2.0.5" target="_blank">10.2.0.5 (Server Patch Set)</a></li><li><a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1#FIXED_10.2.0.4.4" target="_blank">10.2.0.4.4 (Patch Set Update)</a></li><li><a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=342443.1" target="_blank">10.2.0.4 Patch 34 on Windows Platforms</a></li></ul></td></tr></tbody></table></blockquote><table class="t_table" style="background: " width="90%"><tbody><tr><td>Symptoms:</td><td>Related To:</td></tr><tr><td><ul><li><a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1#TAGS_LATCHC" target="_blank">Latch Contention</a></li></ul></td><td><ul><li><a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1#TAGS_AQ" target="_blank">Advanced Queuing</a></li></ul></td></tr></tbody></table>Description The interprocess messaging mechanism may exhibithigh latch contention under certain workloads.This shows as contention for a particular"channel operations parent latch" In particular this can affect AQ. Note: This fix also addresses the issue in <a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=6148054.8" target="_blank">bug 6148054</a>       <table class="t_table" style="background: " width="98%"><tbody><tr><td colspan="6" rowspan="1">Bug 6148054 - RAC hang waiting for "reliable message" [ID 6148054.8]</td></tr><tr><td colspan="6" rowspan="1"></td></tr><tr><td colspan="2" rowspan="1" width="25%"> </td><td colspan="3" rowspan="1" width="50%">Modified 30-MAR-2011     Type PATCH     Status PUBLISHED</td><td> </td></tr></tbody></table> Bug 6148054  RAC hang waiting for "reliable message" This note gives a brief overview of bug 6148054. The content was last updated on: 30-MAR-2011 Click <a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1" target="_blank">here</a> for details of each of the sections below. Affects:<blockquote><table class="t_table" style="background: "><tbody><tr><td>Product (Component)</td><td>Oracle Server  (Rdbms)</td></tr><tr><td>Range of versions believed to be affected</td><td>Versions >= 10.2.0.4 but BELOW 10.2.0.5</td></tr><tr><td>Versions confirmed as being affected</td><td><ul><li><a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1#AFFECTS_10.2.0.4" target="_blank">10.2.0.4</a></li></ul></td></tr><tr><td>Platforms affected</td><td>Generic (all / most platforms affected)</td></tr></tbody></table> It is believed to be a <a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1#TAGS_REGRESSION" target="_blank">regression</a> in default behaviour thus:    Regression introduced in 10.2.0.4 Note that this fix has been <a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1#TAGS_SUPERCEDED" target="_blank">superceded</a> by the fix in <a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=7801939.8" target="_blank">Bug:7801939</a> </blockquote>Fixed:<blockquote><table class="t_table" style="background: "><tbody><tr><td>This issue is fixed in</td><td><ul><li><a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1#FIXED_11.1.0.6" target="_blank">11.1.0.6 (Base Release)</a></li></ul></td></tr></tbody></table></blockquote><table class="t_table" style="background: " width="90%"><tbody><tr><td>Symptoms:</td><td>Related To:</td></tr><tr><td><ul><li><a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1#TAGS_HANG" target="_blank">Hang (Process Hang)</a></li><li> Waits for "reliable message"</li><li><a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=170464.1" target="_blank">Waits for "wait for unread message on broadcast channel"</a></li></ul></td><td><ul><li><a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=245840.1#TAGS_OPS" target="_blank">RAC (Real Application Clusters) / OPS</a></li></ul></td></tr></tbody></table>Description<blockquote> In a RAC environment which has the fix for <a href="https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=4605569.8" target="_blank">bug 4605569</a>present some processes may periodically hangwaiting on "reliable message" and "wait for unread message on broadcast channel" </blockquote>

[[i] 本帖最后由 maclean 于 2011-12-20 16:08 编辑 [/i]]

回复只看该作者道具举报

Maclean Liu(刘相兵

5^#

发表于 2011-12-20 16:14:26

具体建议：

目前metalink对该bug没有明确的定位，原本认为该bug在现有的patch中修复了，实际发现可能是伪修复；所以目前官方没有提供可用的one-off patch 补丁或 workaround 的方式

若该问题反复出现，那么建议：

1. 升级到最新的11.2.0.3

2. 调优数据库降低工作负载

回复只看该作者道具举报

返回列表

		自动登录	找回密码
密码			注册