May 29, 2007
安装了Oracle10g 10.2.0.3 感受众多BUG
作者:eygle
出处:http://blog.eygle.com
今天，终于还是在笔记本上装上了Oracle，打过Patch之后的版本是10.2.0.3SQL> select * from v$version;
BANNER
----------------------------------------------------------------
Oracle Database 10g Enterprise Edition Release 10.2.0.3.0 - Prod
PL/SQL Release 10.2.0.3.0 - Production
CORE 10.2.0.3.0 Production
TNS for 32-bit Windows: Version 10.2.0.3.0 - Production
NLSRTL Version 10.2.0.3.0 - Production
看一下Bug修正列表，在关心的几个环节浏览了一下。
首先是ASSM的一堆Bug：
| Fixed in Release | Bug Number | Description |
|---|---|---|
| 10.2.0.3 | 3279497 | Interupted (Ctrl-C) TRUNCATE can corrupt ASSM table |
| 10.2.0.3 | 3610777 | Excessive CPU from INDEX insert into ASSM index segment |
| 10.2.0.3 | 4055634 | TRUNCATE table has to much overhead when operating on empty segments |
| 10.2.0.3 | 4188324 | Wasted space possible in ASSM segments |
| 10.2.0.3 | 4475314 | DELETE then INSERT without commit on ASSM segment can be slow |
| 10.2.0.3 | 4660718 | Allow choice between space / performance for ASSM segments |
| 10.2.0.3 | 4694312 | Index inserts to ASSM segments can be slow |
| 10.2.0.3 | 4887955 | Unnecessary extent allocated in locally managed ASSM tablespace in RAC |
| 10.2.0.3 | 5034679 | Hang / errors making ASSM tablespace read only with concurrent DML |
| 10.2.0.3 | 5349632 | Poor performance for ASSM insert after large DELETE |
| 10.2.0.2 | 4288876 | Excessive redo / undo generation from insert/update with ASSM indexes |
| 10.2.0.2 | 4486753 | DMLS contention with ASSM segments |
| 10.2.0.2 | 4517470 | Table corruption with RAC when ASSM managed and truncate used |
不看不知道，一看吓一跳吧，ASSM的Bug还真不少，而且还都是比较恐怖的Bug。
再看看10g的一大亮点ASM的BUG：
| Fixed in Release | Bug Number | Description |
|---|---|---|
| 10.2.0.3 | 3811904 | Rebalance can fail if there is one full disk |
| 10.2.0.3 | 3885499 | ASM hang possible |
| 10.2.0.3 | 4229751 | CONVERT TABLESPACE fails to create file with DB_FILE_NAME_CONVERT on ASM |
| 10.2.0.3 | 4450268 | ASM disk mounts are serialized |
| 10.2.0.3 | 4466433 | Restoring current controlfile on ASM creates a file with "backup" string |
| 10.2.0.3 | 4561867 | ASMB may exit prematurely causing OERI in client |
| 10.2.0.3 | 4671721 | OERI [kfcInitDlocn02] possible in ASM instance |
| 10.2.0.3 | 4691191 | Failure during file grow may leak ASM disk space |
| 10.2.0.3 | 4709210 | ASM LMS OERI[504] on [kfcl le freelist] latch |
| 10.2.0.3 | 4709214 | ASM may crash with OERI[kfclGetLock40] |
| 10.2.0.3 | 4752481 | Skewed ASM disk allocation when single file resized many times |
| 10.2.0.3 | 4865736 | ASM resource should not have dependency on VIP |
| 10.2.0.3 | 4872617 | Dump (kfclDumpLe) in ASM |
| 10.2.0.3 | 4883032 | ASM fails with ORA-15096 during crash or instance recovery |
| 10.2.0.3 | 4886660 | ASMB can fail with OERI:kffmAllocate_1 |
| 10.2.0.3 | 4907496 | Processes hang after adding disk back to ASM diskgroup using force option |
| 10.2.0.3 | 4943406 | LMON needs to message SMON to start recovery in ASM instance |
| 10.2.0.3 | 4960705 | ASM waits longer than necessary during reconfiguration |
| 10.2.0.3 | 4966352 | OERI[unsorted_pins_4] possible on ASM |
| 10.2.0.3 | 4967266 | DB recovery blocked by ASM recovery |
| 10.2.0.3 | 5012099 | ASM rebalance problems |
| 10.2.0.3 | 5024639 | ASM hangs if private interface down on a surviving node |
| 10.2.0.3 | 5039964 | ASM disks show as provisioned although kfed shows valid disk header |
| 10.2.0.3 | 5089630 | Dump (kfncInitSlavePool) / ORA-15012 from Log Miner / Streams against LOGs on ASM |
| 10.2.0.3 | 5090822 | OERI[17092] / ORA-10382 when ASM IO completes concurrently with DISKGROUP dismount |
| 10.2.0.3 | 5134663 | OERI[2103] with ASM |
| 10.2.0.3 | 5226903 | OERI:[kfcDeadlockAvoid01] when running "check all repair" |
| 10.2.0.3 | 5360719 | ASM and database instances hang when disk array gets disconnected |
| 10.2.0.3 | 5515492 | Data corruption when ASM failgroup removed |
| 10.2.0.2 | 4747535 * | A diskgroup cannot be mounted on both nodes at the same time. This bug is alerted in Note:353065.1 |
| 10.2.0.2 | 4237629 | OERI[kfdjoin5] on ASM instance with multiple diskgroups |
| 10.2.0.2 | 4362919 | CRS ASM failure recovery does not recover database instances |
| 10.2.0.2 | 4413010 | OERI[kffdmove01] selecting from V$ASM_ALIAS in ASM instance |
| 10.2.0.2 | 4455512 | ASM instance startup can be slow in RAC environments |
| 10.2.0.2 | 4478323 | OERI[kfcDeadlockAvoid01] on SQL against V$ASM_ALIAS |
| 10.2.0.2 | 4490933 | RMAN backup fails when backupsize > free on ASM but reclaimable space available |
| 10.2.0.2 | 4503419 | Cannot drop disk normal with less than 3 failgroups |
| 10.2.0.2 | 4550821 | OERI[kfklufsscannext] in ASM during disk discovery |
| 10.2.0.2 | 4666938 | OERI[kjbdowncvt:__l] can occur in ASM |
| 10.2.0.2 | 4671216 | ASM operations on a file are blocked while it is resized |
| 10.2.0.2 | 4693247 | Dump (kfcgetbuffer) while resizing a tablespace using ASM storage |
| 10.2.0.2 | 4700681 | ASM disk shrinkage can corrupt ASM metadata |
| 10.2.0.2 | 4772979 | OERI[kfdskAlloc0] in RDBMS instance when adding back a dropped disk |
建议大家看看Bug列表，还是有好处的，这就可以知道在自己的产品环境上可以去避免哪些问题。
这些Bug遇到的不多，不错CRS的Bug却是遇到了不少，BUG列表里关于CRS的BUG太多，不列了。
-The End-
Posted by eygle at 4:27 PM | Comments (13)
DBA警示录:Move数据表与索引重建
作者:eygle
出处:http://blog.eygle.com
昨天在ITPUB上再次看到一个由Move操作引发的故障。
导致故障的原因是:
对表做了move后,没有rebuild index,然后就关闭数据库了,导致数据库不能重新启动
此时Alert文件中记录了如下信息：
Errors in file /space/oracle/admin/test/udump/test_ora_22836.trc:
ORA-00704: bootstrap process failure
ORA-00604: error occurred at recursive SQL level 1
ORA-01502: index 'SYS.I_ACCESS1' or partition of such index is in unusable state
Mon May 28 16:31:10 2007
Error 704 happened during db open, shutting down database
USER: terminating instance due to error 704
Instance terminated by USER, pid = 22836
ORA-1092 signalled during: ALTER DATABASE OPEN...
上一次见到类似的事故是在2005-11-18，已经快2年的时间，所以怎么说呢，应该说事故并非多发。
不过还是有必要提醒一下大家：
SYS的对象最好不要轻易Move，出于谨慎以及对于DBA的基本要求，如果真要操作SYS对象，一定要做好完备的测试。当然数据库的备份是必不可少的。
-The End-
Posted by eygle at 9:05 AM | Comments (0)