8i | 9i | 10g | 11g | 12c | 13c | 18c | 19c | Misc | PL/SQL | SQL | RAC | WebLogic | Linux

Home » Misc » Here

Comments for Oracle Database 12c Release 2 (12.2) RAC On Oracle Linux 7 Using VirtualBox


Rijesh Chandran said...

"ADF Driver" => It should be AFD Driver right? ASM Filter Driver? Minor correction :)

Tim... said...

Hi.

Thanks for pointing that out. I've corrected the typo now. :)

Cheers

Tim...

Li said...

Hi Tim,

on the udev rule for ASM disks, it appears they are owned by root.root, as it appears:
# ls -al /dev/oracleasm/*
lrwxrwxrwx. 1 root root 7 Sep 18 08:19 /dev/oracleasm/asm-disk1 -> ../sdb1
My questions is: shouldn't it be owned by oracle.dba? from my experience, they are like that on Linux 6 and below...Thanks!

Tim... said...

Hi.

The symbolic links are owned by "root:root", but the devices are owned by "oracle:dba". I've added an extra example in the article to show you what I mean.

Cheers

Tim...

Jamal said...

Need to keep more patience in Configure GIMR part , its really time consuming and worst part we don't know what is happening in background

Tim... said...

Hi.

Yes. Especially on under-powered VMs. It can take a very long time.

Cheers

Tim...

Manuel Donoso said...

Hi,
I've got a question about config of shared disk.
I've configurated disks for asm in the same manner, but without partitioning disk, so I haven't /dev/sdb1. I've got AFD configurated over /dev/sdb, /dev/sdc/ ...
In this manner if I expand one disk on virtual box, I'll be able to expand the corresponding asm diskgroup in ASM.
Is this correct? Disatvantages?
Many thanks.

Tim... said...

Hi.

The main advantage of using the partition is it is obvious to all people and tools the disk is being used. Without it, some tools will see it as an unused disk that can be reused. All Oracle docs seem to assume you will partition, even if it is a single full size partition. Technically, it doesn't seem to matter either way. I partition. :)

Cheers

Tim...

Akhlaq said...

Hi,
After cloning the VM when i try to create the second node it give error saying cannot create already exist.

Tim... said...

Hi.

OK. Try putting the virtual disk in a different location, so the sub-directory VirtualBox wants to create is not already there.

Cheers

Tim...

Akhlaq said...

Dear Tim,
Excellent article. Our consultant says it is recommended by Oracle to install Oracle binaries on shared storage and to keep data on single tablespace.
Need your expert advice on this.

Thanks&Regards

Tim... said...

Hi.

Shared binaries save space, but you may have to turn off whole cluster for some patches. I prefer not to.

One tablespace? Unlikely. Are you sure they didn't mean one ASM disk group?

Cheers

Tim...

Akhlaq said...

Hi,
Great so we will install our Binaries on local servers as we have no space issue.
Ya he said one tablespace for data and one for indexes. According to him this is for ease of manageability.
But our main concern is performance.
Thanks for the quick reply.

Regard

Tim... said...

Hi.

Which tablespace an object is in is not really that big a deal these days. There is little in the way of performance difference by putting in the same tablespace or splitting on modern storage. Manageability is my primary concern, so I split things by functional area, not data/index so much.

Cheers

Tim...

Jay said...

Hi Tim,

Please correct a small typo. When adding shared disks on VM, each disks should be 20Gb.
Given disk size is correct for Linux environment, but for window it is just 5Gb each disk.

Thanks.

Tim... said...

Hi.

Great. Thanks for the heads-up. That's corrected now. :)

Cheers

Tim...

marcin said...

Great article, but do you remember on with kernel version you installed this stuff.
I have problems with running root.sh scripts due to
modprobe: ERROR: could not insert 'oracleoks': Unknown symbol in module, or unknown parameter (see dmesg)?

Tim... said...

Hi.

Sorry. No idea what kernel version I was on. I typically use the latest UEK version available for any new installation.

Cheers

Tim...

Terry P. said...

Hi Tim,
Great article, very helpful, thank you very much! However, the section that creates the udev rules file 99-oracle-asmdevices.rules has one missing bit: you did not pre-create the /dev/oracleasm directory and make it owned by oracle:dba. Once I did that, it all worked great, and it works as-is if you remove the 'oracleasm/' path from the SYMLINK+= lines as the udev article has it.

Tim... said...

Hi.

If you don't have the "oracleasm" directory, but you have it in the SYMLINK, the path is part of the link, so it is fine. If you don't have "oracleasm" in the SYMLINK, you have to have the directory present. So your experience is dependent on the approach you take. It's totally fine to have an extended path as part of a symbolic link.

Cheers

Tim...

Mike said...

Hi Tim, talking about net config.

first you said:enp0s9 (eth2): IP=192.168.1.102, Subnet=255.255.255.0, Gateway=, DNS=, Search= (Connect Automatically)

The during check:

enp0s9: flags=4163 mtu 1500
inet 192.169.1.101 netmask 255.255.255.0 broadcast 192.169.1.255

Is it a mistake talking prior to 192.168.1.102 then 192.169.1.101 ?


Tim... said...

Hi.

Remember, there are 2 nodes. On node 1 enp0s9=192.168.1.101. On node 2 enp0s9=192.168.1.102. Believe this is what I have expressed in the article if you check which node I am discussing at the time. :)

Cheers

Tim...

Mike said...

Hi tim, i know, you config 2 nodes.

But during first node setup:

1) config enp0s9 = 192.168.1.101
2) after some steps (BUT ONLY in first node setup) you check via ifconfig on same card and you wrote: inet 192.169.1.101

is it only a write error ?

Thanks MIke

Tim... said...

Hi.

I don't think so. There are six references to enp0s9 in the article.

Node1: References 1, 2 & 3 all say 192.168.1.101
Node2: References 4, 5 & 6 all say 192.168.1.102

There are only 2 "ifconfig" commands. First one for node1. Second for node2. They are consistent. The writing above the ifconfig output says if it is for the first or second node. :)

Cheers

Tim...

Tim... said...

If you still think there is a problem, please email me using tim{at}oracle-base.com and explain where exactly you see the issue, because I don't. :)

Tim... said...

Hi.

I see what you are getting at. In your example you said 192.168.1.102 then 192.169.1.101, and I think what you meant was 192.168.1.101 then 192.169.1.101. I was just looking at the last number (101/102), which was consistent. I didn't notice the 168/1699 switch. That is a mistake. I'll correct it. :)

Thanks for the heads-up. :) It should be 168 all the way through. I must have captured the details with a mistake in, then gone back to correct it to get the pings to work and forgotten to recapture the ifconfig output. :)

Cheers

Tim...

Marcin said...

At your configuration you have got multiple listeners like LISTENER and LISTENER_SCAN.
In my configuration these two listeners are based on the same port 1521 is it correct?
Because when I add the new PDB database it occurs in both liseteners.
lsnctl status LISTENER and lsnrctl status LISTENER_SCAN. Is it correct?

Tim... said...

Hi.

It's normal. Read about what the SCAN listener does and you will understand why.

Cheers

Tim...

Hanh said...

Hi Tim,
Would Oracle 12.2 RAC works with Oracle Linux 7.5? Just wonder, since I've already have Oracle Linux 7.5 downloaded from Oracle18c RAC installation.
Hanh,

Tim... said...

Hi.

Yes. Any OL6 or OL7 spin.

Cheers

Tim...

Nic said...

Looks like latest uek is not yet supported and grid install fails while running the root.sh:

CRS-8505: Oracle Clusterware CLSECHO process with operating system process ID 11111 encountered internal error ACFS-09319
> ACFS-9319: Unknown OS kernel version '4.1.12-124.19.2.el7uek.x86_64' detected.

Nic said...

Havent been to explicit: its 12.2.0.1 Grid release on OL 7.4 :

ACFS-9505: Using acfsutil executable from location: '/u01/app/12.2.0.1/grid/usm/install/cmds/bin/acfsutil'
> CRS-8505: Oracle Clusterware CLSECHO process with operating system process ID 11111 encountered internal error ACFS-09319
> ACFS-9319: Unknown OS kernel version '4.1.12-124.19.2.el7uek.x86_64' detected.

Tim... said...

Hi.

Sorry for the delay. I had to build a test case. I don't see this problem. Can you try my Vagrant build as a test to see if that works for you?

https://github.com/oraclebase/vagrant/tree/master/rac/ol7_122

It builds fine with 4.1.12-124.18.6.el7uek.x86_64.

Cheers

Tim...

SK said...

Tim, is this missing "-silent" from the command

cd /u01/app/12.2.0.1/grid
./gridSetup.sh -responseFile /tmp/grid_config.rsp


Tim... said...

Hi.

Yes. Added it in. Thanks for the heads-up. :)

Incidentally, I rarely use the response files anymore. I tend to specify all parameters on the command line, as I do in the Vagrant builds.

https://github.com/oraclebase/vagrant/tree/master/rac/ol7_122

Cheers

Tim...

Anonymous said...

I'm facing issues on
nslookup ol7-122-scan
;; connection timed out; no servers could be reached.

Tim... said...

Hi.

It appears you've either setup your networking incorrectly, or you've not set up your DNS properly.

If you check out my hands-free vagrant build you will see exactly what I do for the complete build.

https://github.com/oraclebase/vagrant/tree/master/rac/ol7_122

Cheers

Tim...

SK said...

Tim,

I had to add below parameter to get my silent install to work because of "PRVG-13606 : chrony daemon is not synchronized with any external". Did silent install work by using separate diskgroups for GIMR and OCR ?

cd /u01/app/12.2.0.1/grid

./gridSetup.sh -silent -responseFile /tmp/grid_config.rsp

TO
./gridSetup.sh -silent -ignorePrereqFailure -responseFile /tmp/grid_config.rsp

Tim... said...

Hi.

I didn't use that parameter, but that's fine. I edited the Chrony config. That should sort out your failure. My hands-off vagrant build has all the scripts for the silent build.

https://github.com/oraclebase/vagrant/tree/master/rac/ol7_122

This install uses one big disk group for everything.

Cheers

Tim...

Anonymous said...

AAAAAAAA+++++ article. Everything worked flawlessly on my KVM Linux box with 2 CentOS guest systems. The only thing which I changed was udev rule (because scsi_id returned null while trying to query virtio drives, I assigned a fake serial num to my drives and used ATTRS{serial}=="123" expression instead). Thanks Tim for the great article. May the force be with you!

Tim... said...

Glad you like it. :)

billy said...

after all stuf
when i run dbca i take this error

dbca does not support this operation in a grid infrastructure oracle home

Tim... said...

Hi.

Looks like you are in the grid home, rather than the DB home when running that command.

Cheers

Tim...

billy said...

rhanx for responce Tim

how can i switch to db home for run dbca


i follow up just your documents

Tim... said...

Hi.

You set up "grid_env" and "db_env" to allow you to switch.

I'll add it to the article.

Cheers

Tim...

billy said...

thank you tim
it is work now :)

DO NOT ask technical questions here! They will be deleted!

These comments should relate to the contents of a specific article. Constructive criticism is good. Advertising and offensive comments are bad and will be deleted!

If you post personal information (name, email address etc.) you are agreeing to them being stored and displayed. Feel free to remain anonymous.