September 9th, 2009
st0ma
I noticed an IP conflict today on a windows box hosted on the ESXi. In the events viewer under system I checked the MAC address of the system trying to hijack my IP address. I wanted to find a quick way and check if this mac address is in my existing ESXi Virtual Machines or it’s outside somewhere..
I wanted to find a quick and dirty way to check this since there is number of machines on the ESXi host.
Here is what I did…
I opened VMWare VI-TOOLKIT. After I connected I decided to try some commands that I regularly use such as get-vm and get-vmguest. After I found nothing interesting I checked out the CI Toolkit Cmdlets Reference Document.
And there it was… the perfect command for what I wanted..
Read more…
How to extract the IP addresses in use by the Guests in our VMware ESX host?
This was the question that I asked myself yesterday, when I had to go over a long list of IPs and check if any of the listed IPs was in use by any Guest Operating system on our ESX server. I immediately thought of the VMware VI Toolkit and all the nice commands that I have seen there, but none of those came to me immediately. A colleague was swifter and used python to get any matches after he copied all IPs manually using the Infrastructure Client, but since I had few spare minutes today, I decided to solve this one and post the answer. Here it is:
[VI Toolkit] C:\Program Files\VMware\Infrastructure\VIToolkitForWindows> get-vmg
uest -vm (get-vm *) |select IPAddress
IPAddress
———
{192.168.128.110}
{192.168.128.113}
{}
{192.168.128.127}
{}
{}
{192.168.128.125}
{}
{}
{192.168.128.175}
{}
{}
{}
{}
{192.168.128.186}
{192.168.128.153, 192.168.128.163, 10.10.128.153}
{}
{192.168.128.154, 192.168.128.164, 10.10.128.154}
{192.168.128.102}
{192.168.128.196}
{192.168.128.236}
{192.168.128.213}
{192.168.128.103}
{192.168.128.254}
Read more…
1. ESXi 3.5 Extending the VMDK file [Virtual Machine Hard Disk]
The main reason behind this is the fact that I needed more space for 2 more Oracle databases on a SLES10 Linux system. The partition mounted on /u01 was initially created as 21GB but I quickly depleted the space with 3 oracle 10g databases that took more than 17GB and the space left was not sufficient for the 2 new databases that I had to create.
Before I advanced with the extension of the VMDK file I connected to each separate instance and performed “shutdown immediate” command as sysdba.
Then I stopped the listener, dbconsole, isqlplus and once I confirmed that no oracle related processes were present on the system I performed a shutdown “shutdown -h now”.
In order to extend an existing hard drive attached to a virtual machine you have to make sure no snapshots of the virtual machine are present. I know that this is quite uncomfortable considering the risky operation that you are about to perform but there is a work around. (The work around is not in this post, so please let me know if you are interested or simply search for it. There is a good chance that I will have an article concerning that topic)
Read more…
Categories: Database, ESX, ESXi, Linux, Oracle, SLES, VI, VMware Tags: ESXi, fdisk, Linux, Oracle, SLES, VI, vmdk, VMware
February 20th, 2009
st0ma
When I finished the clusterware install in the lab environment I have continued with the database software itself but it appears that I needed a VMware ESXi Update 3 required! The kernel version of SLES failed verification and upon an upgrade of the kernel through YAST both machines were unusable since they would not boot.
This is an obstacle until completed since I can’t continue with the How-to until this is done.
In Part 9 of the Oracle RAC How-to we have completed successfully the installation of the Oracle Clusterware services on the shared storage for the two Suse Linux Enterprise Server 9 SP5 nodes.
Read more…
Categories: Database, ESX, ESXi, Linux, Oracle, SLES, VI, VMware Tags: Database, ESX, ESXi, Kernel, Oracle, VMware
February 16th, 2009
st0ma
Verify the Oralce Clusterware Installation
With the introduction of Oracle RAC 10g, cluster management is controller by the evmd, ocssd and crsd processes.
Run the ps command on both nodes to make sure that the processes are running.
rac01:/u01/clusterware/cluvfy # ps -ef |grep d.bin
root 4694 1 0 Feb13 ? 00:00:00 /u01/crs1020/bin/crsd.bin reboot
oracle 5242 4692 0 Feb13 ? 00:00:00 /u01/crs1020/bin/evmd.bin
oracle 5344 5326 0 Feb13 ? 00:00:00 /u01/crs1020/bin/ocssd.bin
root 20078 10946 0 09:44 pts/1 00:00:00 grep d.bin
Next you should check the /etc/inittab file, which is processed whenever the runlevel changes:
Read more…
Categories: Database, ESX, ESXi, Linux, Oracle, SLES, SSH, VMware Tags: Clusterware, ESXi, Linux, Oracle, SLES, VMware
February 14th, 2009
st0ma
Linux OS Parameters
Here is the list of the required parameters for clusterware and oracle database 10g
Parameter |
Value |
File |
semmsl semmns semopm semmni |
25 032 000 100 128 |
/proc/sys/kernel/sem |
shmmax |
The minimum of the following (4 GB – 1 byte), or half the size of physical memory (in bytes), whichever is lower. |
/proc/sys/kernel/shmmax |
shmmni |
4096 |
/proc/sys/kernel/shmmni |
shmall |
2097152 |
/proc/sys/kernel/shmall |
file-max |
65536 |
/proc/sys/fs/file-max |
ip_local_port_range |
Minimum: 1024 Maximum: 65000 |
/proc/sys/net/ipv4/ip_local_port_range |
rmem_default |
262144 |
/proc/sys/net/core/rmem_default |
rmem_max |
4194304 |
/proc/sys/net/core/rmem_max |
wmem_default |
262144 |
/proc/sys/net/core/wmem_default |
wmem_max |
4194304 |
/proc/sys/net/core/wmem_max |
In order to check the values in your system use the sysctl command.
You will probably get the following results from the default kernel configuration:
Read more…
Categories: Database, ESX, ESXi, Linux, Oracle, SLES, VMware Tags: ESXi, Linux, Oracle, SLES, sysctl, VMware
February 14th, 2009
st0ma
Configuring SSH access for the oracle user for Remote installation
Login as the oracle user and generate keys for ssh authentication without passwords.
I usually check the /etc/ssh/sshd_config file to make sure I have the proper configuration in place.
I usually disable the DNS check since it causes the ssh to timeout.
UseDNS no
Now I will generate the ssh keys with no password
Read more…
Categories: Database, ESX, ESXi, Linux, Oracle, SLES, SSH, VMware Tags: ESXi, Oracle, ORACLE_HOME, SLES, SSH, VMware
February 13th, 2009
st0ma
Required UNIX groups and users. User environment and Shell Limits
1. Creating an oracle user, a dba and an oinstall group on each node.
# groupadd -g 500 oinstall
# groupadd -g 501 dba
# useradd -u 500 -d /home/oracle -g oinstall -G dba -m -s /bin/bash oracle
Reset the oracle user password to something you want
# passwd oracle
Changing password for oracle.
New password:
Re-enter new password:
Password changed
Check that the nobody user exists:
#grep nobody /etc/ passwd
[I had to add space between the / and passwd since mod_security of this server would not allow me to post.. ]
Read more…
Categories: Database, ESX, ESXi, Linux, Oracle, SLES, VMware Tags: ESXi, Linux, Oracle, SLES, VMware
February 13th, 2009
st0ma
Hardware and package requirements
1. Hardware
Here is a small list of hardware requirements that we have satisfied by creating our virtual machine as it is:
at least 1 GB of RAM
minimum of 1GB of swap space
/tmp directory is said to be minimum 400MB which we have spare in our / mount point
2. package requirements
Since time will be the most important variable for the Oracle RAC it is a must to have all nodes up to date using NTP deamon.
In my case I use a local ntp server and I configure the /etc/ntp.conf to poll that server on boot.
In addition to this I have added a cronjob as root to restart the xntpd service every hour at the twentieth minute, to make sure every node will be generally up to date even If I perform snaphosts and reverts forgetting about the time.
Read more…
Categories: Database, ESX, ESXi, Linux, Oracle, SLES, VMware Tags: ESXi, Linux, Oracle, SLES, VMware
February 13th, 2009
st0ma
Network and hostname configuration of the nodes
Once you have converted the the first virtual machine that you have installed, you have a mirrored copy of it.
This means that the hostnames of the two machines will be identical and this has to be fixed.
Each Node must have at least 2 network adapters: one for the public and one for the private interconnect. In addition , the interface names associated with the network adapters for each network must be the same on all nodes.
For the public network each adapter must support TCP/IP. For the private network, the interconnect must support UDP for Linux. Gigabit ethernet or an equivalent is recommended.
Read more…
Categories: Database, ESX, ESXi, Linux, Oracle, SLES, VMware Tags: ESXi, Linux, Oracle, SLES, VMware