The same applies for the fact that systems should be sized for what they really need and one should not oversize the systems. For this reason I am a personal fan of just enough operating system (JEOS) kind of deployments and just enough hardware resources.
One of the downsides is that if you use this line of thinking and use a bare minimal operating system deployments with a limit set of compute resources you sometimes run into the issue that you miss things you actually would like to have. One of the things that you might run in at first when using this line of reasoning on the Oracle Public Cloud is that of swap space.
No swap space
When deploying a templated bare minimal system in the Oracle Public Cloud using the Compute Cloud Service you will notice that you do not have swap space. Depending on the goal you have for a specific instance this might be an issue or you might not even notice. Some applications are perfectly fine not having swap space while some even demand it during installation.
By default you willl not have swap space, this means you will have to add swap space at run time or you have to make sure that your automated deployment will take care of adding swap space for those instances where it is required.
Give me swap space
In cases where you need swap space you can simply add swap space. In effect there are two main ways of adding swap space. You can use a swap file you can create with the dd command or you can add a entire disk to your machine and use that for swap space. It used to be that case that adding additional disks to your machine was a task that was not as simple as only executing commands and it would involve actual hardware.
In the era of virtualization and cloud, and in all reality since we started using SAN solutions for storage, adding more diskspace to a machine is not that hard anymore. claiming and adding more disk space in the cloud era is simply requesting more space from your cloud provider.
Creating a disk in the Oracle Cloud
When we decide to use a disk to use for swap space the first thing we need to do is to ensure we have a disk to add. To create a disk we navigate to the storage tab in the compute cloud service console. Here we can create a new disk as a storage volume, an example of this is shown below
After the disk is created you can attached the disk to an instance in the Oracle Compute Cloud Service. This will result in a screen like the one below. You have to select the instance name from a list of values and you have to select as which number you add the disk to the instance.
Selecting the number to which you add the disk is important, it will result in the device name under which the new disk is known in the Oracle Linux instance. By default the first disk will be known device /dev/xvdb, the second device will be /dev/xvdc, the third will be /dev/xvdd etc.
Creating the swap space
As soon as you have attached the disk to the instance it will be known as a new device on the instance. This means you will have to tell the instance how you like to use it. You could use it for storage which would require you to mount it as a filesystem. However, in this case we like to use it as swap space which will require a bit of a different approach than using it as a filesystem
First we check the current amount of swap space that is available on the instance at this moment. As can be seen below we do currently do not have any swap space added.
[root@pocapp2 ~]#
[root@pocapp2 ~]# free
total used free shared buffers cached
Mem: 7657252 5632492 2024760 0 72088 5244236
-/+ buffers/cache: 316168 7341084
Swap: 0 0 0
[root@pocapp2 ~]#
Now we have to see where the disk is we just created and added to the instance. As we have selected that the disk should be added as the secondary disk we now should be able to find a new disk as device /dev/xvdc
[root@pocapp2 ~]# [root@pocapp2 ~]# ls /dev/xvd* /dev/xvdb /dev/xvdb1 /dev/xvdb2 /dev/xvdc [root@pocapp2 ~]#
As you can see from the above example we now have a device /dev/xvdc available on the system which we can use for swap space. Now we have to make the device a swap device by using mkswap
[root@pocapp2 ~]#
[root@pocapp2 ~]# mkswap /dev/xvdc
mkswap: /dev/xvdc: warning: don't erase bootbits sectors
on whole disk. Use -f to force.
Setting up swapspace version 1, size = 10485756 KiB
no label, UUID=8ac3eacf-42e6-43a4-8a53-d33f29767dee
[root@pocapp2 ~]#
Now we have ensured that we can use the new disk as swap space we have to enable swap in the system by making use of this new swap device. A simple swapon command on the device will ensure that the swap is used.
[root@pocapp2 ~]# [root@pocapp2 ~]# swapon /dev/xvdc [root@pocapp2 ~]#
After executing the swapon command the device should now be acting as a device to provide swaps space to the system. You can check so by again executing the free command and you will notice that additional swap space is now active.
[root@pocapp2 ~]#
[root@pocapp2 ~]# free -m
total used free shared buffers cached
Mem: 7477 5508 1969 0 70 5121
-/+ buffers/cache: 316 7160
Swap: 10239 0 10239
[root@pocapp2 ~]#
Even though we now have the swap space available on the system we have not made it persistent. Meaning, next time we reboot the machine we will lose the swap space again. To ensure the swap space is persistent we have to add a line to /etc/fstab like the one shown as an example below.
/dev/xvdc swap swap defaults 0 0
Now we have ensured that our system is equipped with additional swap space and that it is done so in a persistent manner to ensure that swap space is available every time we reboot the machine.
No comments:
Post a Comment