Goldengate无法读取裸设备上的日志

风三

环境：AIX5.3 + oracle10204 + goldengate112101
1、源端是一套两个节点的RAC，redo log位于裸设备上，归档日志分别放在两个节点的/backup/arch目录下
2、goldengate放在节点2上，节点1的归档目录通过nfs，在节点2上只读挂载到/backup/arch1目录下。


源端抓取进程的配置文件
GGSCI>add ext extbs, tranlog, threads 2, begin now
GGSCI>add exttrail /ggs/dirdat/u1, ext extbs, megabytes 100
GGSCI>edit param extbs
extract extbs
userid goldengate, password goldengate
tranlogoptions altarchivelogdest threadid 1 /backup/arch1
tranlogoptions altarchivelogdest threadid 2 /backup/arch
tranlogoptions altarchivedlogformat %t_%s_%r.arc
tranlogoptions rawdeviceoffset 0
exttrail /ggs/dirdat/u1
discardfile  /ggs/dirrpt/extbs.dsc,append
numfiles 3000
dynamicresolution
table lxz.tb_latency;

extract进程启动之后短时间内状态为running，但实际上队列文件大小没有变化(从trace文件也能看出，其实这段时间也没在读取日志)。一旦当前日志变为归档，抓取进程立刻异常终止，报错信息有如下两种：
2013-07-31 10:26:19  ERROR   OGG-01028  Oracle GoldenGate Capture for Oracle, extboss.prm:  Partial block (511/512) encountered getting file info for seqno 17330.
2013-07-31 10:26:19  INFO    OGG-00991  Oracle GoldenGate Capture for Oracle, extboss.prm:  EXTRACT EXTBOSS stopped normally.

或者

2013-07-31 22:06:05  ERROR   OGG-01028  Oracle GoldenGate Capture for Oracle, extboss.prm:  Error validating file info for sequence 1960 because of an invalid thread ID (2).
2013-07-31 22:06:05  INFO    OGG-00991  Oracle GoldenGate Capture for Oracle, extboss.prm:  EXTRACT EXTBOSS stopped normally.
2013-04-10 22:06:05  WARNING OGG-01077  Oracle GoldenGate Capture for Oracle, extboss.prm:  Error closing file, handle: 0×00000000, err: 9 – Bad file number.




为了排除NFS的问题，手工把两个节点的归档文件copy到同一个本地目录下，修改参数文件之后发现抓取进程也无法读取日志文件，所以应该跟NFS也无关。


能否帮忙指点下，问题大概出在什么地方？谢谢了。

风三

换了个11.2.1.0.6的版本搞定了