Is this supposed to be the most reliable database server? A msg to oracle.

Question

Hi, it's been 3 weeks now that I'm fiddling with oracle 11g and I am surprised with the amount of installation/startup/shutdown/management issues that I am facing. I need to evaluate oracle as a possible replacement for sql server, I am supposed to test oracle's scalability vs sql server scalability but so far I was messed up by all kind of installation/management issues. Is Oracle a product always in alpha stage?
I am not new with oracle, I've used oracle 10g in the past and I went through all the oracle library documentation (concepts, dba guide, sql reference etc). By now I am pretty confident with db concepts like control files, redo logs, spfile, tablespaces etc.
I've installed it on 4 machines:
1. windows xp 64bit, 40GB Ram
installation went ok, database + listener + dbconsole started correctly. I did a few things but after rebooting the machine the db did not start! (it complained that it tried to allocate 0 bytes of memory where it needed 2mb for some reason).
dbca is supposed to configure an instance but guess what? To configure the instance if first must start the instance and ofcourse in my case it couldn't start the instance!!!
After reading docs, googling etc, I exported spfile to pfile , manually edited the pfile, imported the pfile as spfile and the db finally started working.
Why on earth a product costing so much can't survive a reboot?
2. centos 5.3 64bit on a P4 machine with 2gb ram
installed ok, database + listener + dbconsole started ok after install. There is no script that I am aware which starts or shuts down the db nor the installer prompted me to install the db as a service (/etc/init.d). So I manually had to create scripts to start/shutdown the db by exporting ORACLE_HOME, ORACLE_SID and using dbstart.sh and dbshut.sh and also calling emctrl start dbconsole.
Which means that if someone doesn't shut down the instance before rebooting linux, will the db shut down abnormally?
I would expect a product that costs £13k to install itself properly on linux. mysql is free but comes with proper installations in most linux flavors.
3. centos 5.3, 64bit on an athlon 64 3000+ with 1gb ram.
installed ok, database + listener + dbconsole started ok after install. As you can guess, it didn't survive a reboot. Anyway to make long story short, I managed to create startup/shutdown scripts and all started working.
Then I tried to play admin for a while, i.e. tried to move tablespaces, backup/restore db, backup tablespaces and so on, using dbconsole. Oh well, I can't count the times dbconsole was failing... Are we the beta testers of dbconsole?
But then I needed to change the domain name of the machine and bang! All db services broken again. After fiddling with listener.ora and other config files that I don't remember, I managed to make db and listener work again but not the dbconsole which doesn't want to start and also I don't know where are the log files (there are so many log folders....) Anyway I am still digging into the documentation.
4. fedora 9 32 bit, 2gb ram. Same issues like #2,#3.
I've the following suggestions for oracle:
1. Less configuration files for oracle's next version.
2. Installation should "do it all" for the user in all operating systems
3. a dedicated testing team for dbconsole - it's just buggy
4. improvements on db,tablespace backups and movements. Is it so hard to create a utility that create a zipped backup of tablespaces or the database which can be used on a different server to restore the files?
I've still not worked on my main task : to test scalability. More comments will follow...
Regards,
Kostas

Answer 1

Hi, found the log file, it is under .../dbhome_1/oc4j/j2ee/OC4J_DBConsole_server.lan_orcl/log
It appears that oc4j was trying to load the encryption keys file from OC4J_DBConsole_localhost_orcl, didn't find it, oc4j was then exiting but emctl was constantly trying to restart oc4j!!! One more cool "feature" of oracle.
Anyway, I managed to fix that issue but now oc4j/dbconsole still doesn't start, dunno why, the log says:
<MESSAGE>
<HEADER>
<TSTZ_ORIGINATING>2010-01-07T21:35:39.517+00:00</TSTZ_ORIGINATING>
<COMPONENT_ID>oc4j</COMPONENT_ID>
<MSG_TYPE TYPE="WARNING"></MSG_TYPE>
<MSG_LEVEL>1</MSG_LEVEL>
<HOST_ID>server.lan</HOST_ID>
<HOST_NWADDR>192.168.1.4</HOST_NWADDR>
<MODULE_ID>network</MODULE_ID>
<THREAD_ID>13</THREAD_ID>
<USER_ID>ariskk</USER_ID>
</HEADER>
<CORRELATION_DATA>
<EXEC_CONTEXT_ID><UNIQUE_ID>192.168.1.4:94967:1262900139517:43</UNIQUE_ID><SEQ>0</SEQ></EXEC_CONTEXT_ID>
</CORRELATION_DATA>
<PAYLOAD>
<MSG_TEXT>Exception in NIOServerSocketDriver:selectForRead</MSG_TEXT>
<SUPPL_DETAIL><![CDATA[java.nio.channels.ClosedChannelException
     at java.nio.channels.spi.AbstractSelectableChannel.configureBlocking(AbstractSelectableChannel.java:252)
     at oracle.oc4j.network.NIOServerSocketDriver$SelectorThreadTask.selectForRead(NIOServerSocketDriver.java:331)
     at oracle.oc4j.network.NIOServerSocketDriver.selectForRead(NIOServerSocketDriver.java:58)
     at oracle.oc4j.network.ServerSocketAcceptHandler.persistConnection(ServerSocketAcceptHandler.java:389)
     at oracle.oc4j.network.ServerSocketAcceptHandler.endReadHandlerRun(ServerSocketAcceptHandler.java:409)
     at oracle.oc4j.network.ServerSocketReadHandler$SafeRunnable.run(ServerSocketReadHandler.java:275)
     at oracle.oc4j.network.ServerSocketAcceptHandler.procClientSocket(ServerSocketAcceptHandler.java:234)
     at oracle.oc4j.network.ServerSocketAcceptHandler.access$700(ServerSocketAcceptHandler.java:29)
     at oracle.oc4j.network.ServerSocketAcceptHandler$AcceptHandlerHorse.run(ServerSocketAcceptHandler.java:879)
     at com.evermind.util.ReleasableResourcePooledExecutor$MyWorker.run(ReleasableResourcePooledExecutor.java:303)
     at java.lang.Thread.run(Thread.java:595)
]]></SUPPL_DETAIL>
</PAYLOAD>
</MESSAGE>
I also got an other problem. I have a heap table and I've created the same table as index organized. I am trying to copy the data over :
insert into header2 select * from header nologging;
But after around 5000 secs, sql developer gives me this error:
SEVERE     117     1314807          java.io.PipedInputStream.checkStateForReceive(PipedInputStream.java:246)
and the dml stops running. I've run that 3-4 times and it fails from 4800 to 5200 secs so I recon a time out is in place, any idea how to remove this timeout?