Home » RDBMS Server » Server Administration » 10g Upgrade (Oracle 10g, OS - AIX)
10g Upgrade [message #364485] Mon, 08 December 2008 16:44 Go to next message
prashanthgs
Messages: 89
Registered: May 2005
Location: chennai
Member
Hi

We are applying the patch to upgrade the database from 10.2.0.3 to 10.2.0.4.

While running the catupgrd.sql after some time it got failed.

We increased the shared pool size and java pool size to 300M and 150M.

Following are the errors in the log file.

dbms_registry.check_server_instance;
ERROR:
ORA-03114: not connected to ORACLE

Errors in file /u01/app/oracle/admin/JDV2DB1/udump/jdv2db1_ora_413704.trc:
ORA-00603: ORACLE server session terminated by fatal error
ORA-04030: out of process memory when trying to allocate 8168 bytes (callheap,kcbtmal allocation)
ORA-04030: out of process memory when trying to allocate 8168 bytes (callheap,kdbmal allocation)
ORA-04030: out of process memory when trying to allocate 8168 bytes (callheap,kcbtmal allocation)
ORA-04030: out of process memory when trying to allocate 8168 bytes (callheap,kcbtmal allocation)
ORA-04030: out of process memory when trying to allocate 8168 bytes (callheap,kcbtmal allocation)
Doing block recovery for file 1 block 830
Mon Dec 8 17:18:56 2008
Doing block recovery for file 1 block 830
Block recovery from logseq 864, block 8677 to scn 317265642
Mon Dec 8 17:18:56 2008
Recovery of Online Redo Log: Thread 1 Group 6 Seq 864 Reading mem 0
Mem# 0: /u02/oradata/JDV2DB1/redo_6a.log
Mem# 1: /u03/oradata/JDV2DB1/redo_6b.log
Mon Dec 8 17:18:56 2008
Recovery of Online Redo Log: Thread 1 Group 1 Seq 865 Reading mem 0
Mem# 0: /u02/oradata/JDV2DB1/redo_1a.log
Mem# 1: /u03/oradata/JDV2DB1/redo_1b.log
Block recovery completed at rba 865.2590.16, scn 0.317265643
Mon Dec 8 17:20:54 2008
Completed checkpoint up to RBA [0x35e.2.10], SCN: 317248633
Mon Dec 8 17:21:18 2008
Completed checkpoint up to RBA [0x35f.2.10], SCN: 317253628


By referring ORA-00603 error it shows

An ORACLE server session is in an unrecoverable state.

Login to ORACLE again so a new server session will be created


We tried different new oracle sessions but again we are getting same type of errors. How we can fix this issue.

Any suggestions pls.

DBNAME : ORACLE 10G
OS: AIX


Prashanth
Re: 10g Upgrade [message #364487 is a reply to message #364485] Mon, 08 December 2008 17:00 Go to previous messageGo to next message
ThomasG
Messages: 3212
Registered: April 2005
Location: Heilbronn, Germany
Senior Member
You would probably have to decrease the memory used by Oracle, since the error means that Oracle can't get enough memory allocated from the OS.

Are there any more detailed errors in the /u01/app/oracle/admin/JDV2DB1/udump/jdv2db1_ora_413704.trc tracefile?

Are there any ulimits set on the OS level?

Also, have a look at Metalink Note 233869.1 "Diagnosing and Resolving ORA-4030"

Re: 10g Upgrade [message #364488 is a reply to message #364487] Mon, 08 December 2008 17:18 Go to previous messageGo to next message
prashanthgs
Messages: 89
Registered: May 2005
Location: chennai
Member
Hi Thomas,

Thanks for your reply.

I am finding below in trace file.

ORA-04030: out of process memory when trying to allocate 2097224 bytes (joxpp heap,f:Reserved3)
Current SQL statement for this session:

Prashanth
Re: 10g Upgrade [message #364489 is a reply to message #364488] Mon, 08 December 2008 17:24 Go to previous messageGo to next message
ThomasG
Messages: 3212
Registered: April 2005
Location: Heilbronn, Germany
Senior Member
So it seems the Operating system doesn't want to give the oracle process more memory.

Either the ulimits are set to low, or there is not enough memory in the server for the memory settings of Oracle.

Have you checked the ulimits?


Re: 10g Upgrade [message #364490 is a reply to message #364489] Mon, 08 December 2008 17:35 Go to previous messageGo to next message
prashanthgs
Messages: 89
Registered: May 2005
Location: chennai
Member
$ ulimit -n
2000

I increased the PGA_AGGREGATE_TARGET value from 24M 100M and executed it again.



Re: 10g Upgrade [message #364491 is a reply to message #364490] Mon, 08 December 2008 17:45 Go to previous messageGo to next message
ThomasG
Messages: 3212
Registered: April 2005
Location: Heilbronn, Germany
Senior Member
Ummmmmm.. "ulimit -n" is the "number of file descriptors".

Do an "ulimit -a" to see all limits, especially the "max memory size" and "virtual memory"

Quote:

I increased the PGA_AGGREGATE_TARGET



As I already said, the OS isn't giving you more memory, so it probably doesn't help to increase it even more on the Oracle side.

And do you have access to/read that Metalink Document? Perhaps ask the person who gave you the Patch-Files to also give you that document.
Re: 10g Upgrade [message #364666 is a reply to message #364491] Tue, 09 December 2008 08:26 Go to previous messageGo to next message
prashanthgs
Messages: 89
Registered: May 2005
Location: chennai
Member
Yes Thomas you are right!

I checked the ulimit -a and found that

$ ulimit -a
time(seconds) unlimited
file(blocks) unlimited
data(kbytes) unlimited
stack(kbytes) 4194304
memory(kbytes) 32768
coredump(blocks) unlimited
nofiles(descriptors) 2000


I asked my Sysadmin to change memory also to Unlimited.

Once its changed, I will execute the scripts again and let me posted on the results.
Re: 10g Upgrade [message #364720 is a reply to message #364666] Tue, 09 December 2008 15:55 Go to previous messageGo to next message
prashanthgs
Messages: 89
Registered: May 2005
Location: chennai
Member
After changing ulimit -a memory to unlimited, catupgrd.sql execution worked.

Thanks!

Prashanth
Re: 10g Upgrade [message #364814 is a reply to message #364720] Wed, 10 December 2008 02:49 Go to previous message
ThomasG
Messages: 3212
Registered: April 2005
Location: Heilbronn, Germany
Senior Member
Glad it worked, and thanks for the feedback. Thumbs Up
Previous Topic: Change to logging in a hash subpartition
Next Topic: User management(urgent)
Goto Forum:
  


Current Time: Mon Dec 02 07:03:09 CST 2024