Oracle in ora-3136,ora-609

Fatal NI Connect error 12170.

VERSION Information:
TNS for 64-bit windows:version 11.2.0.1.0-production
Oracle bequeath NT Protocol Adapter for 64-bit windows:version 11.2.0.1.0-production
Windows NT TCP/IP NT Protocol Adapter for 64-bit windows:version 11.2.0.1.0-production
Time:01-11 Month-2011 16:46:57
Tracing not turned on.
Tns error struct:
NS Main ERR code:12535

Tns-12535:tns: Operation timeout
NS Secondary Err code:12606
NT Main ERR code:0
NT Secondary Err code:0
NT OS Err code:0
Client Address: (address= (PROTOCOL=TCP) (host=192.168.3.53) (port=2950))
Warning:inbound Connection timed out (ORA-3136)

Fatal NI Connect error 12560, connecting to:
(Local=no)

  VERSION information:
TNS for 64-bit windows:version 11.2.0.1.0-production
Oracle bequeath NT Pro Tocol Adapter for 64-bit windows:version 11.2.0.1.0-production
Windows NT TCP/IP NT Protocol Adapter for 64-bit Win Dows:version 11.2.0.1.0-production
  time:09-11 month -2011 11:22:32
  tracing not turned on.
  T NS error struct:
    ns main err code:12560
   
Tns-12560:tns: Protocol Adapter Error
&N bsp;   NS Secondary err code:0
    NT main err code:0
    NT Secondary E RR code:0
    nt OS err code:0
Opiodr aborting process unknown ospid (21388) as a result of ORA-6
Wed Nov 11:22:49

ora-3136 this error, in most cases, we can be ignored, because this error is generally due to the client due to the correct password of Meiyu, connection timeout caused. To give a very simple example, we use Sqlplus user/password@tnsname, but the password entered is wrong, Oracle hint: Ora-01017:invalid Username/password; Logon denied, after, do not do anything, the connection hangs there, wait a minute, can see this error in Alertlog.

Therefore, one possibility for a ora-3136 error is that the client uses the wrong password to log on, but then does not exit the connection.

But ora-3136 's error is not only a possibility, but also when received from a malicious client connection, such as Dos attacks, in addition, when the database load is heavier, there will be such an error. It can be seen that there are 3 possibilities in Metalink "Troubleshooting ora–3136 WARNING Inbound Connection out [ID 465043.1]":

There can is three main reasons for this error-

1. Server Gets a connection request from a malicious client which are not supposed to connect to the database, in which CA Se the error thrown is the correct behavior. You can get the client address for which the error is thrown via sqlnet log file.
2. The server receives a valid client connection request but the client takes a long time to authenticate more than the DE Fault seconds.
3. The DB server is heavily loaded due to which it cannot finish the client logon within the timeout specified.
According to my understanding, in short, in Oracle listening to receive a request from the client, when fork to the server process, if the process of discovering an unexpected, such as a password error, such as the database load is too heavy, will be the parameter ora-3136 error.

There may be a database server pressure, but also may be the application of server pressure, as long as the 60 seconds in this certification process is not completed, will report this error.

Run AWR, found that shared pool support is relatively large, the library cache latch and shared pool latch and swap, page in increased CPU wio increased, the phenomenon and our system several times exactly the same, after several tuning, We analyze the main process: 1, the analysis of business connections, mainly through the monitoring log and switch firewall, determine the number of connections per second, determine which business connection over, after the identification of the business connection pool has problems, in other sites crawler crawl, the pressure is very large, limited connection pool, business connectivity problems to solve; Analyze whether there are no binding variables, after analysis, the system code is good, all using binding variables; 3, analysis shared pool configuration, after checking, using automatic memory configuration, 10g can not solve the problem is that when automatically configured, there will be bugs caused by the increase in the pool, modified for manual configuration, 4, Analysis of the machine memory parameter configuration, found that the remaining memory is small, set memory retention, to ensure that when the system pressure, the system can still allocate memory, instead of using memory as cache,oracle instead to take the address from swap to cause the system does not respond to the error ORA-01017, Analysis of the OS memory allocation mainly look at the shared memory and the process memory, found that memory consumption is the main or shared pool support is too large, reduce, the OS cache is too expensive, set to Directio, lower cache use memory, after this situation has not been seen. As performance said, small lotus you have to see in the machine memory is what is consumed by what, pin live SGA is not the most critical, if enough memory, the SGA is transferred to swap the probability is very low, and will soon come back, so I think still have to analyze the end of memory used where.

Swap serious

Replace with swap, page in elevation->hdisk0 busy 100%-> Overall system load high->fork server process failure->ora-3136 error
With swap substitution, page in elevation->hdisk0 busy 100%-> Overall system load high->library cache latch and shared pool latch

But this case doesn't find the reason why swap is high.