OCCI: ORA-21500 [17099] seg faults on 10.2.0.4.1 client

Question

Hi,
I've been getting this seg fault using OCCI, my environment:
Oracle client 10.2.0.4.1 x86_64 (jul 2009) on rhel5.4
Oracle database 9.2.0.8 x86_64 on rhel4.8
I've written a g++ multi thread app that is making use of Connection Pooling, within 5 minutes of my app starting it seg faults with:
ORA-21500: internal error code, arguments: [], [], [], [], [], [], [], []
ORA-21500: internal error code, arguments: [17099], [], [], [], [], [], [], []
Errors in file :
ORA-21500: internal error code, arguments: [], [], [], [], [], [], [], []
ORA-21500: internal error code, arguments: [17099], [], [], [], [], [], [], []
----- Call Stack Trace -----
ORA-21500: internal error code, arguments: [], [], [], [], [], [], [], []
ORA-21500: internal error code, arguments: [17099], [], [], [], [], [], [], []
Errors in file :
ORA-21500: internal error code, arguments: [], [], [], [], [], [], [], []
ORA-21500: internal error code, arguments: [17099], [], [], [], [], [], [], []
----- Call Stack Trace -----
calling call entry argument values in hex
location type point (? means dubious value)
calling call entry argument values in hex
location type point (? means dubious value)
Cannot seek to string table section header in /proc/8712/exe.
Cannot seek to string table section header in /proc/8712/exe.
Segmentation fault
its compiled against gcc version 4.1.2 20080704 (Red Hat 4.1.2-46)
I've tried the following:
Installing the occi download from http://www.oracle.com/technology/tech/oci/occi/occidownloads.html and compiling with gcc-3.4.6
I'm using the following commands to compile & link:
g++ -DMOXA_DEBUG -I/database/u00/app/oracle/product/10.2.0/db_1/rdbms/demo -I/database/u00/app/oracle/product/10.2.0/db_1/rdbms/public -I/database/u00/app/oracle/product/10.2.0/db_1/plsql/public -I/database/u00/app/oracle/product/10.2.0/db_1/network/public -O0 -g3 -Wall -c -fmessage-length=0 -fPIC -MMD -MP -MF"src/DB_Table.d" -MT"src/DB_Table.d" -o"src/DB_Table.o" "../src/DB_Table.cpp"
g++ -L/database/u00/app/oracle/product/10.2.0/db_1/rdbms/lib -L/database/u00/app/oracle/product/10.2.0/db_1/lib -shared -o"libmoxa.so" ./src/DB_Table.o ./src/Dialup.o ./src/Modem.o ./src/Rtu.o ./src/Rtu_Type.o ./src/Session.o ./src/Session_Log.o ./src/Soe_Point.o ./src/Soe_Point_Event.o ./src/Thread.o ./src/Utils_Time.o ./src/db.o -locci10_343 -lclntsh -ldl -lm
I've tried to do a -locci10 instead of the -locci10_343, still nothing..
Intesting to note, the older version of the application on the same environment doesn't have this problem..
I do know OCCI is a bit tempremental, especially when creating threads, as i've had to put a 100ms delay otherwise i get other errors
the back trace looks like, it allways core dumps when releasing the connection..
#0 0x00002b01b9fd8cae in slrac () from /database/u00/app/oracle/product/10.2.0/db_1/lib/libclntsh.so.10.1
(gdb) bt
#0 0x00002b01b9fd8cae in slrac () from /database/u00/app/oracle/product/10.2.0/db_1/lib/libclntsh.so.10.1
#1 0x00002b01b9fd8b3d in kgdsaaddr () from /database/u00/app/oracle/product/10.2.0/db_1/lib/libclntsh.so.10.1
#2 0x00002b01b9fd83b8 in kgdsdst () from /database/u00/app/oracle/product/10.2.0/db_1/lib/libclntsh.so.10.1
#3 0x00002b01b9c7f0b0 in skgudmp () from /database/u00/app/oracle/product/10.2.0/db_1/lib/libclntsh.so.10.1
#4 0x00002b01b9fb27ac in kgeriv () from /database/u00/app/oracle/product/10.2.0/db_1/lib/libclntsh.so.10.1
#5 0x00002b01b9fb2fab in kgesiv () from /database/u00/app/oracle/product/10.2.0/db_1/lib/libclntsh.so.10.1
#6 0x00002b01b9fb2ace in kgesic0 () from /database/u00/app/oracle/product/10.2.0/db_1/lib/libclntsh.so.10.1
#7 0x00002b01b9ace52d in kpuhhfre () from /database/u00/app/oracle/product/10.2.0/db_1/lib/libclntsh.so.10.1
#8 0x00002b01b9ad9a3a in kpuhmcfre () from /database/u00/app/oracle/product/10.2.0/db_1/lib/libclntsh.so.10.1
#9 0x00002b01b9a6ca5a in kputac () from /database/u00/app/oracle/product/10.2.0/db_1/lib/libclntsh.so.10.1
#10 0x00002b01b9b8265a in kpuspsessionrelease () from /database/u00/app/oracle/product/10.2.0/db_1/lib/libclntsh.so.10.1
#11 0x00002b01b9be1d6f in OCISessionRelease () from /database/u00/app/oracle/product/10.2.0/db_1/lib/libclntsh.so.10.1
#12 0x00002b01b95452f1 in oracle::occi::ConnectionImpl::do_destroy () from /home/fep/moxa_1.3/libmoxa/Debug/libmoxa.so
#13 0x00002b01b954d78c in oracle::occi::ConnectionPoolImpl::terminateConnection () from /home/fep/moxa_1.3/libmoxa/Debug/libmoxa.so
#14 0x00002b01b953c47c in DB::release_connection (conn=0xed36498) at ../src/db.cpp:126
I'm downloading the 11.2g client in the hope it works there without a hitch..
Any help/suggestions whould be much appreciated..
Cheers
James

Answer 1

Shiyer,
libmoxa.so is the library I have created which is performing the oracle queries..
The pool is created from the main thread by:
Environment _env = Environment::createEnvironment (Environment::DEFAULT);
ConnectionPool connectionpool = _env->createConnectionPool(
username, passwd, tnsname, 1, 10, 1);
And then each query is used run inside a connect initiated with:
Connection *conn = connectionpool->createConnection(_username.c_str(), passwd.cstr());
Statement *stmt = conn->createStatement("select 1 from dual;");
ResultSet *rset = stmt->executeQuery();
while( rset->next() == rset->DATA_AVAILABLE ) {
std::cout << rset->getInt(1) << std::endl;
stmt->closeResultSet (rset);
conn->terminateStatement (stmt);
and connection released with:
connectionpool->terminateConnection(conn);
Hope this helps!
cheers
James