请各位高手指点，不胜感激！

主机配置：主机1：IBM P595

主机2：IBM P570

Oracle版本：9.2.0.8

RAC实例： s1,s2

问题描述：由于网络问题，导致实例s2自动关闭，网络恢复正常后，启动实例s2，输出显示：

SQL>startup;

ORACLE instance started.

Total system global area 2270136344 bytes

Fixed size 742448 bytes

Variable size 1191182336 bytes

Database buffers 1073741824 bytes

Redo buffers 4468736

显示到这里，就挂起了

后来，将实例s2 shutdown abort，将目前正常运行的实例s1 shutdown immediate；然后启动实例s2 startup，启动正常，接着启动实例s1，实例s1像最先启动实例s2一样被挂起。

检查Alert文件，文件中有如下的内容：

Starting ORACLE instance (normal)

Wed Mar 7 12:32:19 2007

Global Enqueue Service Resources = 5060, pool = 8

Wed Mar 7 12:32:19 2007

Global Enqueue Service Enqueues = 7482

LICENSE_MAX_SESSION = 0

LICENSE_SESSIONS_WARNING = 0

SCN scheme 3

Using log_archive_dest parameter def***t value

LICENSE_MAX_USERS = 0

SYS auditing is disabled

Starting up ORACLE RDBMS Version: 9.2.0.8.0.

System parameters with non-def***t values:

processes = 200

timed_statistics = TRUE

shared_pool_size = 1073741824

java_pool_size = 16777216

pre_page_sga = TRUE

spfile = /dev/rlvhydk_spfile

cluster_interconnects = 90.88.20.30

active_instance_count = 2

control_files = /dev/rlvhydk_cntrl1, /dev/rlvhydk_cntrl2, /dev/rlvhydk_cntrl3

db_block_size = 8192

db_cache_size = 1073741824

compatible = 9.2.0.0.0

log_archive_start = TRUE

log_archive_dest_1 = LOCATION=/hydk/hydkarch1

log_buffer = 4194304

log_checkpoint_interval = 10000

db_file_multiblock_read_count= 32

cluster_database = TRUE

cluster_database_instances= 2

thread = 1

fast_start_mttr_target = 300

instance_number = 1

undo_management = AUTO

undo_tablespace = UNDOTBS1

remote_login_passwordfile= EXCLUSIVE

db_domain =

instance_name = hydk1

cursor_space_for_time = TRUE

job_queue_processes = 10

background_dump_dest = /oracle/admin/hydk/bdump

user_dump_dest = /oracle/admin/hydk/udump

max_dump_file_size = 10240

core_dump_dest = /oracle/admin/hydk/cdump

db_name = hydk

open_cursors = 300

partition_view_enabled = TRUE

Wed Mar 7 12:32:20 2007

cluster interconnect IPC version:Oracle UDP/IP

IPC Vendor 1 proto 2 Version 1.0

PMON started with pid=2, OS id=831572

DIAG started with pid=3, OS id=913520

LMON started with pid=4, OS id=917526

LMD0 started with pid=5, OS id=888880

LMS0 started with pid=6, OS id=979112

LMS1 started with pid=7, OS id=864330

DBW0 started with pid=8, OS id=548962

LGWR started with pid=9, OS id=503970

CKPT started with pid=10, OS id=782500

SMON started with pid=11, OS id=413766

RECO started with pid=12, OS id=794828

CJQ0 started with pid=13, OS id=1130744

Wed Mar 7 12:32:27 2007

ARCH: STARTING ARCH PROCESSES

ARC0 started with pid=14, OS id=712796

ARC0: Archival started

ARC1 started with pid=15, OS id=868590

ARC1: Archival started

Wed Mar 7 12:32:27 2007

ARCH: STARTING ARCH PROCESSES COMPLETE

Wed Mar 7 12:32:27 2007

ARC1: Becoming the 'no FAL' ARCH

ARC1: Becoming the 'no FAL' ARCHARC1: Thread not mounted

Wed Mar 7 12:32:27 2007

ARC0: Becoming the heartbeat ARCH

ARC0: Becoming the heartbeat ARCHARC0: Thread not mounted

Wed Mar 7 12:32:27 2007

ALTER DATABASE MOUNT

Wed Mar 7 12:32:28 2007

lmon registered with NM - instance id 1 (internal mem no 0)

Wed Mar 7 12:33:27 2007

ARC0: Thread not mounted

Wed Mar 7 12:34:27 2007

ARC0: Thread not mounted

Wed Mar 7 12:35:27 2007

ARC0: Thread not mounted

Wed Mar 7 12:36:27 2007

ARC0: Thread not mounted

Wed Mar 7 12:37:27 2007

ARC0: Thread not mounted

Wed Mar 7 12:37:27 2007

ARC1: Thread not mounted

Wed Mar 7 12:37:35 2007

IPC Send timeout detected. Sender ospid 917526

Wed Mar 7 12:38:27 2007

先确认两台机器之间的通讯是否都没问题,看情况象两机器之间通讯有问题.

再则当UNDO出现问题的时候也有可能出现此类问题,重新创建新的UNDO表空间实验一下.当然前提是并非特别重要的系统

谢谢sosdb，UNDO表空间估计是不能重建，风险太大，应用系统不能出问题，我先去确认两台主机的通信是否正常。

又要请教各位高手：我测试两台主机的心挑线是否连通时：

#stty >/dev/tty0

系统提示：The requested resource is busy.

ksh:/dev/tty0:0403-005 cannot create the specified file.

#cat /etc/hosts >/dev/tt0

系统提示：ksh:/dev/tty0:0403-005 cannot create the specified file.

请问是什么原因，该如何测试心跳线呢？

刚才用cluster的命令测试了，心跳线是正常的，主机之间的网络也是通的，可以相互ping通，说明通信没有问题，现在该怎么办呢？

把RAC的LOG贴出来看看,那里面也许有有用的信息.也有可能隐含参数"_skgxp_udp_ach_reaping_time" 需要修改,尝试把"_skgxp_udp_ach_reaping_time"修改为0.看以上的信息与ach有关与网络有关.

谢谢各位高手，今天终于找到原因，原来是做RAC的其中一个网卡因为交换机的原因没有连通，那天测网络的时候把这个网卡刚好漏了！

晕！