oracle 9i RAC 问题集合
oracle 9i RAC 问题集合
1。三台IBM小机做RAC过程中,发现个问题。
其中有一张表约有500万条记录,如果其中的一个Instance去Select后,另外2个Instance在去Select这个表时候非常慢,只有将第一个Instance shutdown后,另外2个才能正常访问。
或者说那个表只允许一个Instance进行Select等访问?
解决方法:
是BUG,我升级到9204后问题解决了。
2。解决 PRKR-1023 : 在两个节点都做
$ ln -sf /etc/srvConfig.loc /oracle/srvm/config/srvConfig.loc
3。解决 PRKR-1064 :在两个节点都做
$ su
root's Password:
# mkdir /var/opt/oracle
# chown oracle:dba /var/opt/oracle
# chmod -R 777 /var/opt/oracle
# ln -sf /etc/srvConfig.loc /var/opt/oracle/srvConfig.loc
# ls -l /var/opt/oracle/srvConfig.loc
lrwxrwxrwx 1 oracle dba 18 Nov 27 15:18 /var/opt/oracle/srvConfig.loc -> /etc/srvConfig.loc
#
# more /var/opt/oracle/srvConfig.loc
srvconfig_loc=/dev/rsrvconfig
# ls -l /dev/rsrvconfig
crw-rw-r-- 1 oracle dba 45, 21 Nov 25 14:52 /dev/rsrvconfig
#
# dd if=/dev/rsrvconfig of=/dev/null bs=8192
16384+0 records in
16384+0 records out
#
4。创建数据库的时候报错(发生在5%,启动实例的时候)
*
ERROR at line 1:
ORA-03113: end-of-file on communication channel
调整aio,AIX 5L 的默认最小AIO值是10,调的时候,可以以10为单位累加,
注意:
4/1。两个节点要一致
4/2。调整aio之后需要reboot,然后起HA,直到不报错为止。
Change / Show Characteristics of Asynchronous I/O
Type or select values in entry fields.
Press Enter AFTER making all desired changes.
[Entry Fields]
MINIMUM number of servers [20] #
MAXIMUM number of servers [100] #
Maximum number of REQUESTS [8192] #
Server PRIORITY [39] #
STATE to be configured at system restart available +
State of fast path enable
5.创建enhanced concurrent vg 时 系统报错:
this system does not support Enhanced Concurrent Capable volume groups
这时,要检查一下系统是否安装了 bos.clvm 的包
6。 创建数据库时没有权限访问裸设备
解决方法:
在安装oracle rac时oracle无法正常访问并行裸设备。
export PGSD_SUBSYS=grpsvcs 在oracle用户下的.profile里加入即可解决oracle无法并行访问raw lv
或者不安装VSD用的pssp rsct.basic.sp &rsct.compat.basic.sp和所有的VSD安装包
7。 HACMP同步失败
报错信息如下:
ERROR: Verification of Cluster Topology for RSCT failed.
See "/var/ha/log/topsvcs.default" for detailed information.
WARNING: File 'netmon.cf' is missing or empty on node dtwtdb01. This file is needed for a cluster with the single-adapter network net_rs232_01. Please create 'netmon.cf' file on node dtwtdb01 as described in 'HACMP Planning and Installation Guide'.
WARNING: File 'netmon.cf' is missing or empty on node dtwtdb02. This file is needed for a cluster with the single-adapter network net_rs232_01. Please create 'netmon.cf' file on node dtwtdb02 as described in 'HACMP Planning and Installation Guide'.
解决方法:
到两个节点的/usr/sbin/cluster目录下检查是否有netmon.cf文件,如果没有,请增加该文件,并增加内容;如果有,请检查内容,内容如下:
192.168.129.1
192.168.129.2
172.168.129.1
172.168.129.2
10.66.8.50
10.66.8.51
8.用netca来建立listener。启动netca,发现java报错。
错误如下:
Error loading native library:libnjni9.so.
解决办法:
$ORACLE_HOME/lib/libnjni9.so这个文件改名
9。 不能两个节点同时mount问题需要考虑的问题
问题现象
在启动数据库实例的时候,在第一个节点启动Oracle没有问题,在启动第二个节点时,Oracle数据库不能到open状态;如果这时将第一个节点shutdown,第二个节点会自动open;也就是说:哪个节点先启动,就能到open状态;后启动的实例需要等第一个节点释放资源后才能启动成功。
问题解决,HACMP配置问题:
从日志文件分析,List of nodes: 0,和List of nodes: 1,可以看出,数据库启动的时候并不能认出两个节点,说明两台主机的通讯存在问题,因此,问题应该出在HACMP的配置中。检查HACMP配置。
问题解决,Oracle初始化参数问题2:
在参数文件中添加cluster_interconnect参数
问题解决,网络配置问题3:
修改网络参数,
/usr/sbin/no -o udp_sendspace=65536
/usr/sbin/no -o udp_recvspace=655360
/usr/sbin/no -o tcp_sendspace=65536
/usr/sbin/no -o tcp_recvspace=65536
/usr/sbin/no -o rfc1323=1
/usr/sbin/no -o sb_max=1310720
/usr/sbin/no -o ipqmaxlen=512
The information in this article applies to:
Oracle Server - Enterprise Edition - Version: 9.2.0 to 9.2.0
This problem can occur on any platform.
Oracle Database
Symptoms
When starting the second instance of a RAC Cluster, The second node will hang indefinitely until the first node is stopped. Starting the second node first will result in the same behaviour for the first node. In essence, only one node can be brought up.
Changes
This would mostly occur in 9i so change that could trigger the problem is
1) Upgrade to 920
2) Applying the 9202 or 9203 patchset
Cause
If udp_sendspace and udp_recvspace is less than db_block_size * db_file_multiblock_read_count then there is a chance that some cr blocks are not received on the second instance which results in this hang
Fix
Check the udp send and recv parameters
For example on AIX we can get this values using
no -o udp_sendspace
no -o udp_recvspace
On increasing these values the problem does not reproduce.
udp_recvspace = 65536
udp_sendspace = 65536
解决办法:
加大udp_sendspace(用no –o udp_sendspace更改,并加入到rc.net,5L也可以使用¬-p参数使重启后还生效),如果还有问题,请把db_file_multiblock_read_count改小
阅读(1127) | 评论(0) | 转发(0) |