第二次在METALINK上面开SR，SR 7840455.992: Large SWAP use on RAC

这次是第二次开SR

处理的应该是个印度人Ahmed Atta，态度礼貌，回复及时，比那些西半球的人好多了

言归正传

10月26日早上10：30，仓储DB（RAC，two node）两个节点都出现了SWAP使用急升的情况，约10秒就有1m的增加。因为之前一直保持着2.6G的使用长时间没有释放（共4G的SWAP），所以很快就突破了3G大关。马上申请半个小时的停机重启。

作为新手的我，毫不犹豫又开SR了

metalink让我收集RDA，然后初步认为是SWAP太小，认为我们16G Physical memory应该设置12G的SWAP（16*0.75），然后让我再次出事的时候做：

1-Run the following commands from the OS prompt .

$ top
&
$ ps -eaf |grep <SID>

2-Get the PID for the process that consumes huge swap space.

3-
a) conn / as sysdba from another session

$ sqlplus /nolog
connect / as sysdba
oradebug setorapid <PID>
oradebug unlimit
oradebug dump systemstate 10

And for 10046 trace, the syntax is :

SQL> oradebug setorapid <PID>
SQL> oradebug unlimit
SQL> oradebug Event 10046 trace name context forever, level 12


b) Exit sqlplus

c) Wait 5 minutes and then repeat step #a to get a second systemstate dump.

d) Upload the 10046 trace file and Systemstate dmp located in the user_dump_dest
e) Please also upload the output for the following query:

select b.username, a.pid
from v$process a, v$session b
where a.addr=b.paddr
and a.pid= <> ; <<------------------Enter the PID for the process in question

 

唯有等问题再次出现了

 

######待续############

 

 

PS:

Note:314422.1 "Remote Diagnostic Agent (RDA) 4.0 - Overview"
Note:330363.1 "Remote Diagnostic Agent (RDA) 4 - FAQ"
Note:330344.1 "Remote Diagnostic Agent (RDA) 4.0 - Training"
Note:330362.1 "Remote Diagnostic Agent (RDA) 4 - Troubleshooting Guide"

Note.286388.1 Ext/Pub How to configure swap space