-
Hi,
I'm having a serious problem at the moment and hope someone out there can help!!
I am running oracle 7.3.4 on AIX.
I was upgrading some of our applications via a script when the script bombed out with a an error ORA-03113: end-of-communication channel.
I have tried to log into SQL Plus and Server manager and get the same message every time. I initally checked the background processes and noticed that CHKPT and LGWR where not running.
The alert log is filling up with the follwoing message:
ORA-00600:internal error code, arguments [15722],[1],[],[],[],[],[],[]
I've been in touch with the support DBA who advised to kill the remaining background processes(?) to try and crash the instance but that didnt work.
Has anyone any idea how I can shut the database down?
Hope someone can help!
Thanks in advance
-
Kill SMON at OS level OR have all users log off and reboot server.
-
What happened when you tried to kill the other processes?
Assuming you've already tried to shutdown abort, the last resort would be to shutdown and restart the whole server. I hope Oracle support can help you without resorting to this option.
-
I've killed SMON, PMON and DBWR at the OS level but am still getting the communication channel error when trying to get into svrmgrl.
I've just tried to issue a startup command and am getting:
LCC-00161: ORACLE error (possible syntax error)
Unfortuinaltely I'm the only person here (in the UK!) and and we dont have any UNIX support as its a test box so I cant reboot the machine!!
Any other ideas?
-
Can you hit the reset switch on the server??
-
Unfortunaltey the box is tucked away in the bowels of the basement and us DBA types dont have access (could be one for the management meeting!!). Looks like it will have to wait till the morning.
I'll keep monitoring the thread just in case anyone has any further brainwaves!
Thanx for the replies
-
In Solaris, you can reboot the server using the command: init 6, while logged in as root. I wonder if the same command applies to AIX. Try to logout and then login again.
-
I've got the root password for the box but its part of a SP node so I'm not sure of the consequences of re-booting it. As no-one else is in the building I guess the only solution is for me to go home and get the UNIX boys to re-boot in the morning.
Does anyone know if there are any common(ish) reasons why the LGWR or DBWR processes would fail? Theres likely to be some questions asked in the morning as to why this happened!
-
I C. I've heard of background process(es) dying if they get overworked.
-
MMmmmm - I wasnt doing too much - just running a script that runs some packages into the database. Apart from that there were no users on the system at all. Very starange. Anyway I've managed to fix the problem by trawling through my little blue note book!
I discovered a command ipcs which shows the shared memory usage on the server. Then there is an ipcrm command to remove the memory. I removed the memory held by oracle and now I can log into server manager and restart and stop the database.
Thanx for your time
Fraze
