Error in writing to file - Oracle

[former_member184628]

Dear All,

We have to upgrade Oracle from 10.2.0.2 to 11.2.0.2. For this we are applying patches till 10.2.0.4. OS is AIX 6.1.

Before proceeding I have stopped SAP, DB, Listerner and emctl also.

But after starting the patches we found the msg that, Error in writing to file 'oracle/SID/102_64/jdk/bin/libdbgmalloc.a'.

[/oracle/SID/102_64/jdk/jre/bin/libdbgmalloc.a(Cannot open or remove a file containing a running program.)]

Click 'Help' for more information.

Click 'Retry' to try again.

Click 'Ignore' to ignore this error and go on.

Click 'Cancel' to stop installation.

We are in the middle of installation where deinstallation has already passed. Please help us in fixing this issue. We tried ps -ef | grep libdbgmalloc.a to check the running status but didnt find that.

Thanks,

Rableen

[ashish_mishra2 · Accepted solution]

Hi,

This is known issue. For whichever files it is prompting, keep renaming those files and click retry.

Do not ignore it otherwise mentioned files will not come of upgraded version.

for 10.2.0.4 patch below files will prompt you as 'in use' just rename one by one and keep hitting retry button.

After retry if you check on the same location an updated file would be present, put by the tool.

1. cd /oracle/<SID>/102_64/jdk/jre/bin

mv libzip.a libzip.a_old

mv libxhpi.a libxhpi.a_old

mv libnet.a libnet.a_old

mv libjsig.a libjsig.a_old

mv libjitc.a libjitc.a_old

mv libjava.a libjava.a_old

mv libhpi.a libhpi.a_old

mv libute.a libute.a_old

mv libdbgmalloc.a libdbgmalloc.a_old

1. cd /oracle/<SID>/102_64/jdk/jre/bin/classic

mv libjvm.a libcore.a_old

mv libcore.a libcore.a_old

Cheers !!!

Ashish

.

[JPReyes]

Error in writing to file 'oracle/SID/102_64/jdk/bin/libdbgmalloc.a'.
Cannot open or remove a file containing a running program

Check programs running under ora<sid> theres must be a jdk related file that is still open, you can kill it and resume the upgrade

Regards

Juan

[former_member184628]

Hi Juan,

After giving command

ps -ef | grep orasid

it didnt gave and java related process running.

orasid 7929994 15990886 0 17:21:42 - 0:00 /usr/dt/bin/dtterm

orasid 14024732 18284588 0 19:13:56 pts/1 0:00 -bin/csh

orasid 14221402 6947040 0 18:14:07 - 0:00 /usr/dt/bin/dtterm

orasid 15466656 14221402 0 18:14:07 pts/3 0:00 -bin/csh

orasid 15990886 20774920 0 17:21:42 - 0:00 /usr/dt/bin/dtexec

orasid 17105072 1 0 15:59:47 - 0:00 /usr/dt/bin/ttsession -s

orasid 17235978 7929994 0 17:21:42 pts/0 0:00 -bin/csh

orasid 17301616 16253004 0 15:59:52 - 0:00 dtfile

orasid 17498186 14024732 0 20:00:52 pts/1 0:00 ps -ef

orasid 18284588 17039578 0 19:13:56 - 0:10 /usr/dt/bin/dtterm

orasid 20512784 18022608 0 15:59:47 - 0:00 /usr/dt/bin/dtsession

orasid 20774920 20512784 0 15:59:48 - 0:00 dtwm

orasid 20316314 1 0 17:27:20 pts/0 1:26 /tmp/OraInstall2011

Only one java related process is running.

root 6619202 1 0 Dec 15 - 1:09 /usr/java5/jre/bin/java -

Thanks.

Rableen

[JPReyes]

can you see if theres any other users pointing to a similar path is ps -ef?

Regards

Juan

[former_member184628]

Hi Juan,

After ps -ef, I dont see even a single process running from /oracle location.

also I checked the location /oracle/SID/102_64/jdk/jre/bin and did not even found soft link to other files.

Looking for a reply.

Thanks,

Rableen

[JPReyes]

Check Ashish answer, I remember something similar happening to me a couple of times...

Regards

Juan

[ashish_mishra2]

Hi Rableen,

You can go ahead with the solution provided above. There would be no impact on system as we have done the same on our whole landscape and did not encounter any issue after multiple round of testings.

This happens because semaphore which uses jdk library files at OS level. You would not be able to locate any process related to that.

Once you rename libzip.a to libzip.a_old and click on retry, another file libzip.a of newer timestamp will be put by upgrade tool.

hostname:orasid 7> ls -rtl|grep libzip.a

-rwxr-xr-x 1 orasid dba 109588 20 Aug 2005 libzip.a_old

-rwxr-xr-x 1 orasid dba 109803 21 Apr 2008 libzip.a

hostname:orasid 7> pwd

/oracle/SID/102_64/jdk/jre/bin

hostname:orasid 7>

Nothing to worry after renaming as upgrade tool will put a new file of new date at the same place.

Also as you are going to upgrade 11G, directory 102_64 will not be used at all after 11G upgrade. A new directory structure will be created as /oracle/SID/11202.

Cheers !!!

Ashish

[former_member184628]

Hi Ashish,

Upgrade finished. Thanks a lot.

I have not applied interim patches till now and running catupgrd.sql script.

Do I need to run this catupgrd.sql script again after applying interim patches?

Thanks

Rableen.

[ashish_mishra2]

> I have not applied interim patches till now and running catupgrd.sql script. 
> 
> Do I need to run this catupgrd.sql script again after applying interim patches?
> Rableen.

If you are upgrading your 10g database to 11g, I am not sure if you really need to apply interim patches.

If you will be applying patches now, this will patch the 10204 database and not on 11G software.

When you install 11g software, on top of that you have to install latest SBP which will be actually used by your system.

So on 10g database, I do not think that it is required to apply interim patches.

Regarding catupgrd.sql, if you have are running it now, and you have decided to apply interim patches on 10g, then it is not required to re-run catupgrd.sql after interim patches.

Check this URL.. might be bit helpful in your activity.

http://www.dbforums.com/oracle/1632779-oracle-10-2-0-2-10-2-0-4-a.html

cheers !!!

Ashish