Out of Memory Error on Linux Server

shows us your kernel parameters

Quote:

---

Originally Posted by pando

shows us your kernel parameters

---

Hi,

Physical RAM Is 16GB

4P Intel(R) Xeon(TM) MP CPU 2.20GHz

---> /etc/sysctl.conf

kernel.shmmax = 4294967295
kernel.shmmni = 4096
kernel.shmall = 2097152
kernel.sem = 250 32000 100 150
fs.file-max = 65536

----> /etc/security/limits.conf

oracle hard nofile 65536
oracle hard nproc 16384

---> SQL> show sga

Total System Global Area 1662588704 bytes
Fixed Size 452384 bytes
Variable Size 587202560 bytes
Database Buffers 1073741824 bytes
Redo Buffers 1191936 bytes

didnt you say your SGA is 2.8GB...?

show us the EXACT error and the error which appears in your alert.log

Quote:

---

Originally Posted by pando

didnt you say your SGA is 2.8GB...?

show us the EXACT error and the error which appears in your alert.log

---

I Also, said, i have reduced the SGA now that i faced that problem.

WHile startup a database, I got the following error:

ORA-27123: unable to attach to shared memory segment

reduce to less than 1.6GB and try

Quote:

---

Originally Posted by pando

reduce to less than 1.6GB and try

---

Hi,

Tried making SGA < 1.6 GB and booted through Huge Mem Kernel, still same error:

print/show us the init.ora

Quote:

---

Originally Posted by yxez

print/show us the init.ora

---

Hi,

Let me ask my question in a different way.

1. The question is, when i am booting the Linux using Normal Kernel, my database is taking upto 2.8GB of SGA max size. Now, inspite of 2.8 GB Sga, i am facing a problem of "Out of Memory Error".

Linux Kernel allocates memory to each and every request it recieves (malloc).

2. Now, i boot my Server using Huge Mem Kernel and the database itself refuses to start gives:

ORA-27123: unable to attach to shared memory segment

Now, if you are going to tell me go on decreasing SGA, that itself is not going to help my cause. Even if i do that, my database doesnt start under Huge Mem Kernel.

I think, i am more clear now.

Hi

Did your oracel instance shutdown properly ?

Can you try a reboot of the server is that a Viable option

Alternatively you can try and list the shared memory segments using ipcs -mt and remove them using ipcs -st

You might also want to

cd $ORACLE_HOME/rdbms/lib
genksms -s 0x50000000 >ksms.s (lower SGA base to 0x15000000)
make -f ins_rdbms.mk ksms.o (compile in new SGA base address)
make -f ins_rdbms.mk ioracle (relink)

if you had originally used genksms -s 0x15000000 >ksms.s to lower the SGA base to support greater then 1.7Gb SGA.

regards
Hrishy

yeah...ur memory might have been locked up by the previously
aborted instance