Error messages with Oracle 7.3.4

[Verónica Muñoz Segovia]

Hello everybody!

Last weekend we had a maintenance window in order to do a Database reestructure and we have to a different problem on two databases.

The fist one error was this:

Doc ID: Note:1024704.6
Subject: ORA-04030 -07324: WHEN EXECUTING DML
Type: PROBLEM
Status: PUBLISHED
Content Type: TEXT/PLAIN
Creation Date: 13-DEC-1996
Last Revision Date: 24-MAY-2001


Problem Description:
====================

You are either on the HP/UX, Digital UNIX or Unixware operating system and
when you try to run a query, do an insert or make an update, you receive one
or more of the following errors:

ORA-04030: out of process memory when trying to allocate %s
bytes (%s,%s)
Cause: Operating system process private memory has been
exhausted
Action: See the database administrator or operating system
administrator to increase process memory quota.
There may be a bug in the application that causes
excessive allocations of process memory space.

ORA-07324: smpall: malloc error while allocating pga
Cause: Malloc library routine returned an error.
Action: Check errno. Possibly out of swap space.

HP-UX Error 12

OSF Error 12


Solution Description:
=====================

On HP/UX, increase the UNIX Kernel parameters "MAXDSIZ" and "MAXSSIZ" by 50%
of the existing value, relink the kernel, and reboot your machine.

On Digital, increase the UNIX Kernel parameter "max_per_proc_datasize" and
"max_per_proc_stacksize" by 50% of the existing value, relink the kernel, and
reboot your machine.

For Kernel configuration assistance, please contact your operating system
vendor.


Explanation:
============

The "MAXDSIZ" kernel parameter, which governs the data segment size (process
memory size), is set too low.


Search Words:
=============

hp, 10.x, 9.04, ORA-4030, ORA-7324, Intel



04030, 00000, "out of process memory when trying to allocate %s bytes (%s,%s)"
*Cause: Operating system process private memory has been exhausted

I don't undastand very well the solution INCREASE BY 50% what is this mean? 50% of what?

And the second error is this one:

%Internal DCE Threads problem (version CMA BL10+), terminating execution.
% Reason: hp_preinit_sig_handler:1
% See 'cma_dump.log' for state information.

The DBA said that when she was trying to create a table space on a specific filesystem this error appears, she tried to do this twice and the third time she did the table space on a different filesystem. And when she was doing the import of the database she tried to do something else on the same filesystem where the error appears and again the error will come out.

Some body can help me? What am I need to do in order to avoid again those errors?

Both box are HP9000 K370 with 3 and 3.5 Gb respectively and running HP-UX 10.20.

The actual kernel parameters are:

maxdsiz 67108864
maxssiz 8388608
dbc_min_pct 2%
dbc_max_pct 10%

Thanks in advance

Veronica

[Rita C Workman]

I have a K570 with only 1.5gb running a 7.3.4 database...

maxdsize=67108864
massize=8388608
dbc_max_pct=10
dbc_min_pct=5

but now here's what you didn't mention:
shmmax=1073741824

Also what about your swap space? What is set up...and is pseudo swap turned on???

Just a thought,
Rita

[Rita C Workman]

One more thing....

also note that ora-4030 mentions there may be a bug whereby the application is allocating excessive memory.

So you may want to look at the code if you know the kernel is set right.

Just a quick thought,
Rita

[Ian Dennison_1]

Veronica,

There is so much here, it is hard to know where to start!

Firstly, check that you have the most up to date patches for the Database (Oracle) and the Operating System (HP-UX). Might be difficult for the older levels, but worth a try.

have you checked how much memory space you have left? try 'vmstat 5 30' when the problem is occurring.

How big is the swap space on your system? Are you noticing the system starting to deactivate processes when the problem happens?

Try searching on 'mazdsiz' in the Documents area of ITRC, there are some very good white papers on memory management that can explain this.

best of luck, Ian

[linuxfan]

Hi,

Looks like you have to increase your kernel parameters maxdsiz to 96468992 and maxssiz to 1258912.

For a better understanding of configurable kernel parameters,
check this out

http://docs.hp.com/hpux/onlinedocs/os/11i/kcparams/KCparams.OverviewAll.html
-HTH
Ramesh

[Wim Rombauts]

I think you need to increase maxdsiz.

I have a C160 with 640L RAM and maxdsiz = 536870912 (or 0x20000000). Maxssiz is 67108864 (0x04000000). Both a lot higher than your configuration. I would say, if you have been running fine previously, just double both values. it will give you a safety margin.

Another parameter that I know that is important is semmns. Oracle can return curious errors when this parameter is too low. I have 500 for this parameter an I am running 2 oracle 8i databases.

I hope this helps.

[Rita C Workman]

Here's one last thought...before lunch.
as was mentioned above..semmaphores.
Check w/your DBA and see if they made any changes to their ~init.ora file. If they did, then did you alot for these requests by calculating how much you need to set your semm* parms up to... here's a thread that may help: http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x10ca6af52b04d5118fef0090279cd0f9,00.html

Rgrds,
Rita

[Sanjay_6]

Hi Veronica,

Try to implement the hp tuned kernel parameter set for General OLTP / database server system. To do this use
sam --> kernel parameter --> configurable parameters --> action --> apply tuned kernel parameter --> General OLTP / database server system.

Rebuild the kernel and reboot. Try the oracle jobs again. Do you still get the error. Post any new errors. Before you make any kernel parameter change, do take a copy of the current kernel parameters. You may have to set a few kernel parameters to their original values, if any application of your demands that.

hope this helps.

Regds

[Verónica Muñoz Segovia]

Firstly I want to thank you for all your response, and to complete the information posted my shmmax is just like yours Rita 1073741824 and semmns 6640 but I have not enable de pseudo-swap = 0.

I don't understand very well how to calculate the sems.

Now I have some ideas to start to review again, unfortunally I can't now move any kernel parameter because the server is on production.

Thanks a lot to all of you.

Veronica

[Verónica Muñoz Segovia]

Just another thing, about the second error message on the other machine, Any comment?

This was the error:


%Internal DCE Threads problem (version CMA BL10+), terminating execution.
% Reason: hp_preinit_sig_handler:1
% See 'cma_dump.log' for state information.