Link to home
Start Free TrialLog in
Avatar of swetha
swetha

asked on

ORA-12571, ORA-01001, ORA-03114

Hi,
We use a software called Docent 4.7 that is connected to Orace 7.3 We are having problems with the connectivity
to the database server. When the Docent server is restarted the database connectivity is fine, but in
a 10-15 minutes the connection goes down with one or more of the following error entries in Docent log
files:

1) Database Error: error number: 12571, server: docentdb, severity: 54, message: [SERVERERROR] ORA-12571:
TNS:packet writer failure

2) Database Error: error number: 1001, server: docentdb, severity: 0, message: [SERVERERROR] ORA-01001:
invalid cursor

3) Database Error: error number: 3114, server: docentdb, severity: 0, message: [SERVERERROR] ORA-03114:
not connected to ORACLE

Could someone please help us out with this?

Thanks,
Swetha
Avatar of fva
fva

Any errors at Oracle side (ALRT or others)?

Enable tracing at client (CLIENT_TRACE_LEVEL=ADMIN in sqlnet.ora) and see what client logs say after the error.

F.
Check the listener.ora file on the server.
Make sure automatic_ipc = off
For some reason this makes the listener unstable.  Ours would stay up for about a day to a week.
Give it a try and tell us what happens.

banic
Man, I'm not awake yet...
After the change you have to cycle the listener.
from a prompt...
>lsnrctl
LSNRCT> stop
LSNRCT> start
LSNRCT> exit
or
LSNRCT> reload

Also, Have a look in your Listener.log
This file might have some juice...
Terry
ORA-01001 is "invalid cursor"  you can try to debug the code, or even trace it.

If you need more info on tracing, things get a little more wordy.  

Hi Swetha,

Looks like your connection is getting expired after certain time of inactivity.  There may be several reasons for this.

> Verify in your SQLNET.ORA if you have SQLNET.EXPIRE_TIME variable set.  If so, comment that out.

> Verify if you have resource_limit set to true (from svrmgrl: show parameter resource_limit.
Then check the default profile for idle_time setting (select * from dba_profiles where profile = 'DEFAULT' and RESOURCE_NAME = 'IDLE_TIME';). If your resource limit is set to TRUE and you have a value other than UNLIMITED for idle_time, then this is what is disconnecting the users.

> Or else may be you are experiencing a problem with your network.  Typically, these errors (12571, 1001, 3114) occur due to a loss of network connectivity.  However if you are on a large network (WAN), you may need to increase the settings for your TCP/IP registry entry to 15 retries.

If you are using Windows operating system as a client to connect to Oracle, set the following variable in your registry.

For Windows NT:
Go to HKEY_LOCAL_MACHINE/SYSTEM/CurrentControlSet/Services/Tcpip/Parameters in the registry, add TcpMaxDataRetransmissions and set it to 15.

For Windows 95:
Go to
HKEY_LOCAL_MACHINE/SYSTEM/CurrentControlSet/Services/vxd/mstcp in the registry, add MaxDataRetires and set it to 15.

The reason is SQL*Net tries to transmit the packet five times before it detected the network was down. This is a default of TCP/IP on Windows NT and Windows 95.

Possibly the last packet sent is not being received by the server which returns the error 54, connection reset/network is busy.  SQL*Net assumes the server is down and closes the connection.

Cheers,
Ramesh
Hi Swetha,

You can also think of the following points,

TCP/IP keepalive on NT
   KEEPALIVE is an extension to TCP/IP which enables the closing of
   dead connections which are no longer being used.

   Problems can occur when the server does not close a connection
   after a PC client process has disappeared. This typically happens
   when a PC user switches off or reboots their machine while still
   connected to Oracle. Note that this is not an Oracle problem, but
   a limitation of TCP/IP, which has no way of knowing for sure
   whether a remote connection has disappeared.

   This feature is enabled by default on NT. Problem can occur however
   if the timeout period is too fast for some heavily used or slow
   network. In those conditions, the KEEPALIVE registry value can be
   used to specify a KEEPALIVE value before a connection gets cut.
 
   HKEY_LOCAL_MACHINE
     System
       CurrentControlSet
         Services
           TCP/IP
             Parameters
               KeepAlive  REG_DWORD  "number"

   A value of '10' is a common value specified for this variable.

   Again, this parameter can be useful on both client and server.
   Start with the machine generating the error, and if needed, also add
   it on the machine on the other side.

 TCP/IP timeouts on Windows 95
   The same parameter can also be specified on Windows 95. It has the
   same functionality, only the location of the parameter in the
   registry is different.

   HKEY_LOCAL_MACHINE
     System
       CurrentControlSet
         Services
           Winsock
             Parameters
               TcpMaxDataRetransmissions  REG_DWORD  "number"

   This parameter is not present in the registry by default. This
   means that the first time, this parameter will need to added to
   this registry key.

   The purpose and behavior of the parameter is the same on the Windows 95
   and Windows 98, as on the Windows NT platform.

 SDU & TDU parameters
   Part of this problem is the sequence of information that gets transmitted.
   If there are disruptions in the sequence, the errors ORA-12151 and
   ORA-12571 can also appear, alerting the application that not all information
   has been send across the network succesfully.

   The sequence of information is determined by the amount of data the program
   is sending, and the actual size the protocol can send across the network
   at a time.

   The more data the program wants to send in one 'go', the more sequences and
   transport packet split-ups will have to be made.

   By default, SQL*Net is using an SDU (Session Data Unit) of 2048 bytes (2Kb)
   and a TDU (Transport Data Unit) of 32768 (32Kb) bytes.
   On standard Ethernet connections, with no special modifications made, the
   SDU is 1500 bytes, while the TDU is 8760 bytes.

   With these values, each data request made by SQL*Net will have to be split
   up in several smaller packets to actually be able to transmit.
   
   Therefore, to minize the creation the additional packets, it is advised, in
   case of these errors, to synchronize the SDU and TDU parameters at the
   SQL*Net level with those of the actual network topology used.

   These SDU and TDU parameters have to be specified at both the client and
   the server configuration files:

   TNSNAMES.ORA:
   -------------
      ORCL.WORLD =
        (DESCRIPTION =
          (SDU=1500)
          (TDU=8760)
          (ADDRESS_LIST =
            (ADDRESS =(PROTOCOL=TCP)(Host=foobar)(Port=1521))
          )  
          (CONNECT_DATA =
            (SID = ORCL)
          )
        )

   LISTENER.ORA:
   -------------
      SID_DESC_LISTENER =
        (SID_LIST =
          (SID_DESC =
            (SDU = 1500)
            (TDU = 8760)
            (SID_NAME = ORCL)
          )
        )

   For more information about the SDU and TDU parameter, see Note 44694.1,
   Note 44694.1: SQL*Net Packet Sizes (SDU & TDU Parameters)

-- Setting a new TDU size on Windows NT
   You can modify the TDU size on NT, via the TcpWindowSize parameter:

   HKEY_LOCAL_MACHINE
     System
       CurrentControlSet
         Services
           Tcpip
             Parameters
               TcpWindowSize  REG_DWORD  "number"
   

Hope this helps you to solve your problem
Sameer
Please update and finalize this old, open question. Please:

1) Award points ... if you need Moderator assistance to split points, comment here with details please or advise us in Community Support with a zero point question and this question link.
2) Ask us to delete it if it has no value to you or others
3) Ask for a refund so that we can move it to our PAQ at zero points if it did not help you but may help others.

EXPERT INPUT WITH CLOSING RECOMMENDATIONS IS APPRECIATED IF ASKER DOES NOT RESPOND.

Thanks,

** Mindphaser - Community Support Moderator **

P.S.  Click your Member Profile, choose View Question History to go through all your open and locked questions to update them.
Avatar of DanRollins
No way to know the correct solution.  Lacking timely response from swetha or contributing experts, but seeing that there is some useful info here, I recommend:

    Refund points and save as a 0-pt PAQ.

DanRollins -- EE database cleanup volunteer
ASKER CERTIFIED SOLUTION
Avatar of Moondancer
Moondancer

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial