-
Notifications
You must be signed in to change notification settings - Fork 20
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
feat: add hor autoscale docs #298
Changes from all commits
68e543b
daf071d
2ff5dc7
f8d23fa
bc444fa
9695037
aabc223
511a135
cba1c6a
0add07e
ca50eaf
6a719b4
43ea4a5
4cccc87
beddeb2
60a1860
ccdacc7
05d1d27
42e2a59
95665ab
2de8cbc
1b3bcb3
f25106e
4c7cdcf
64582a5
e363e5c
635f341
894e94c
2fefb86
8ee7d12
ca8a01e
3cb1fd5
e0a1fd7
68a06ec
6c1e851
11e5c8b
File filter
Filter by extension
Conversations
Jump to
Diff view
Diff view
There are no files selected for viewing
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,204 @@ | ||
--- | ||
myst: | ||
html_meta: | ||
description: Learn how autoscaling dynamically adjusts server resources based | ||
on CPU usage metrics, allowing efficient workload management without manual | ||
intervention. Configure thresholds, durations, and understand the workflow for | ||
horizontal autoscaling without downtime on Hypernode. | ||
title: How does horizontal Autoscaling work? | Hypernode | ||
--- | ||
|
||
# How does horizontal Autoscaling work | ||
|
||
This article explains how Horizontal autoscaling works, what the requirements are and how you can enable it. | ||
|
||
## How does it work | ||
|
||
With horizontal autoscaling, additional Hypernodes will be added automatically to your main Hypernode at peak times. The extra Hypernodes will be used as PHP fpm workers. | ||
Since there's no configuration or IP changes needed, Horizontal autoscaling has zero-downtime. | ||
|
||
For Horizontal autoscaling you can configure some conditions. If the conditions are met from the configured settings in the Control Panel, Horizontal autoscaling will be triggered. | ||
You can configure the following conditions: | ||
|
||
- **CPU Load Threshold:** CPU load that is continuously monitored by the autoscaling agent. The percentage value you see is the actual load divided by the number of CPUs you have. For example, if your actual load is 2, but you have 4 CPUs, your CPU load percentage is 50%. The default setting is 70%. | ||
- **Minimum Duration in Minutes:** Minimum amount of time the server is overloaded before autoscaling gets triggered. We monitor your CPU load every minute; autoscaling will be triggered once the CPU load condition is met for the minimum amount of time set in the Control Panel. The default setting is 15 minutes. | ||
|
||
## Horizontal autoscaling process | ||
|
||
### Upscaling | ||
|
||
Once the configured conditions are met and autoscaling is triggered, we spin up extra Hypernode(s) which meets the extra calculated amount of resources you need. | ||
We create a snapshot of your main Hypernode and attach that snapshot to the extra provisioned Hypernodes. | ||
All the services, such as MySQL, Elasticsearch and Redis, keep running on your original Hypernode. We add the additional Hypernodes as PHP-fpm workers and will handle the incoming requests. | ||
The load of the different autoscaling nodes will be balanced from your original Hypernode. | ||
|
||
If the Hypernodes remains overloaded post-autoscaling, the extra needed resources will be added to handle the traffic. The autoscaling trigger is determined based on these criteria. | ||
The greater of either: | ||
|
||
- User-defined minimum duration for CPU threshold surpassing: Allows time to assess newly allocated resources' impact on CPU performance. | ||
- A cooldown period of 15 minutes: Ensures a minimum interval for reassessment when the user-defined duration is less than 15 minutes. | ||
|
||
This approach selects the longer duration between user-defined settings or the fallback duration before initiating the next autoscaling event. It ensures adequate time for evaluating resource changes on CPU performance. Autoscaling is capped at the biggest plan available on each cloud provider. | ||
|
||
### Downscaling | ||
|
||
If the monitor agent notices that the used resources are below the configured thresholds, the downscale operation will begin. This will lead to removing the additional servers that are no longer needed. | ||
|
||
Downscale will happen by detaching the autoscaling instances from the main Hypernode one by one. | ||
After a downscale operation has been completed, there will be a cooldown of 15 minutes to monitor the behaviour of the newly available resources before downscaling again. | ||
|
||
If the amount of used resources is still below the configured threshold, another downscale operation will be triggered. | ||
|
||
If there is need for extra resources instead, extra resources will be added again. | ||
|
||
## Horizontal autoscaling requirements | ||
|
||
Horizontal autoscaling is available on all the Falcon cloud plans (OpenStack). | ||
|
||
Next to the provider, horizontal autoscaling does have a couple of additional requirements. | ||
|
||
We will divide them between Hypernode-specific and Application-specific requirements | ||
|
||
### Hypernode Specific Requirements | ||
|
||
#### Operating system | ||
|
||
- The operating system of the Hypernode should be Debian Bookworm. If you would like to upgrade the os of your Hypernode, feel free to contact our support team for help. https://www.hypernode.com/en/contact/ | ||
|
||
#### Make sure the Hypernode is a production plan | ||
|
||
For now, we don't support horizontal autoscaling for development plans. | ||
|
||
#### Enable and configure Varnish | ||
|
||
To make use of Horizontal autoscaling, Varnish should be enabled and configured on the Hypernode. | ||
You can check if Varnish is enabled on your Hypernode by running | ||
|
||
```console | ||
hypernode-systemctl settings varnish_enabled | ||
``` | ||
|
||
Example output if Varnish is enabled: | ||
|
||
```console | ||
varnish_enabled is set to value True | ||
``` | ||
|
||
If Varnish is not enabled, you can [enable Varnish](../varnish/how-to-enable-varnish-on-hypernode.md) by following the documentation | ||
If varnish is enabled on your Hypernode, your Magento store should also be configured to make use of varnish. | ||
You can verify if the Varnish host is configured correctly by running the following command from the Magento root: | ||
|
||
```console | ||
php bin/magento config:show system/full_page_cache/varnish/backend_host | ||
``` | ||
|
||
The output should show `varnish` as backend. If it is configured as something else (like `localhost` or `127.0.0.1`), you can update it by running which sets the backend host to `varnish` instead. | ||
|
||
```console | ||
php bin/magento config:set system/full_page_cache/varnish/backend_host varnish | ||
``` | ||
|
||
Additionally make sure the IP range `10.0.0.0/24` is set to the `acl_purge` section in the Varnish VCL file. The `acl_purge` section should look something similar:loaded Varnish VCL. | ||
|
||
```console | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Do we also need to include the following in vcl_recv for handling PURGE requests? sub vcl_recv { Curious if this is required ? There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. No that's completely up to the users |
||
acl purge { | ||
"localhost"; | ||
"10.0.0.0/24"; | ||
} | ||
``` | ||
|
||
#### Make sure to use MySQL 5.7 or higher | ||
|
||
The configured MySQL version should be 5.7 or above. You can check the enabled MySQL version by running the following command. | ||
|
||
```console | ||
hypernode-systemctl settings mysql_version | ||
``` | ||
|
||
Example output if MySQL version is 8.0: | ||
|
||
```console | ||
mysql_version is set to value 8.0 | ||
``` | ||
|
||
If your MySQL version is still set to 5.6, you can consider [upgrading](../mysql/how-to-use-mysql-on-hypernode.md) the MySQL version to a supported version for autoscaling. | ||
|
||
After the version validation, please verify the MySQL host is set to `mysqlmaster`. You can verify this by running `cat app/etc/env.php | grep -i mysql | grep -i host` from the magento root. | ||
|
||
You should see something similar to `'host' => 'mysqlmaster',`. If this is not the case please make sure the database connection host is set to `mysqlmaster` instead of `localhost` or `127.0.0.1` in the magento configuration file at `<magento_root>/app/etc/env.php`. | ||
|
||
#### Disable supervisor services | ||
|
||
Make sure supervisor is disabled and that there are no supervisor services configured. | ||
|
||
#### Disable Podman services | ||
|
||
Make sure podman is disabled and that there are no podman services running. | ||
|
||
#### Configure hostnames correctly | ||
|
||
The database, cache, session and queue of the application must be configured with correct hostnames instead of `localhost` or `127.0.0.1`. This way the services will be available across multiple hypernodes when the app is autoscaled. | ||
|
||
Make sure the env variables (db, cache, session, queue) are not using localhost. | ||
|
||
### Application Specific Requirements - Magento2 | ||
|
||
#### Supported CMS | ||
|
||
Horizontal autoscaling is available for Magento 2.4.7 and higher. | ||
To make use of Horizontal autoscaling, there are a couple of other requirements the application should meet. | ||
|
||
#### Enable and configure Redis Persistent | ||
|
||
Redis persistent is another requirement before you can make use of Horizontal autoscaling. | ||
The persistent instance will be used to store the sessions so we can access the same sessions from the Horizontal autoscale Hypernodes. | ||
|
||
You can check if Redis Persistent is enabled on your Hypernode by running | ||
|
||
```console | ||
hypernode-systemctl settings redis_persistent_instance | ||
``` | ||
|
||
Example output if Redis Persistent is enabled: | ||
|
||
```console | ||
redis_persistent_instance is set to value True | ||
``` | ||
|
||
If Redis Persistent instance is not enabled, you can enable the second Redis instance for sessions you run the command: | ||
|
||
```console | ||
hypernode-systemctl settings redis_persistent_instance --value True | ||
``` | ||
|
||
Make sure Redis session is configured as [described](../../ecommerce-applications/magento-2/how-to-configure-redis-for-magento-2.md#configure-magento-2-to-use-redis-as-the-session-store) in our docs | ||
Please notice the Redis host in the setup documentation. The Redis host should be set to `redismaster` instead of `localhost` or `127.0.0.1`. | ||
|
||
#### Make sure Elasticsearch/Opensearch is configured properly | ||
|
||
Please make sure Elasticsearch or Opensearch host is set to `elasticsearchmaster` in the Magento2 configuration file at `<magento_root>/app/etc/env.php` | ||
More information about [Elasticsearch on Hypernode](../../hypernode-platform/tools/how-to-use-elasticsearch-on-hypernode.md) | ||
|
||
#### Make sure RabbitMQ configured properly | ||
|
||
Please make sure RabbitMQ host is set to `rabbitmqmaster` in the Magento2 configuration file at `<magento_root>/app/etc/env.php` | ||
More information about [RabbitMQ o Hypernode](../../best-practices/database/how-to-run-rabbitmq-on-hypernode.md) | ||
|
||
There is a rabbitmq user provisioned by Hypernode called hypernode-admin as a non-default user. But you can also configure RabbitMQ with a new different user of your own. | ||
But please make sure to configure RabbitMQ without the default guest user. | ||
|
||
#### Make sure Database storage is disabled & Remote storage is enabled and configured. | ||
|
||
Please make sure to enable remote storage for your application and configure it correctly as only AWS-s3 remote storage drivers are supported. | ||
|
||
Make sure that the `remote_storage` key is present in the Magento2 configuration file at `<magento_root>/app/etc/env.php` with the correct config. | ||
|
||
More information about [S3 Remote Storage with Magento2](https://experienceleague.adobe.com/en/docs/commerce-operations/configuration-guide/storage/remote-storage/remote-storage-aws-s3) | ||
|
||
## Enabling Horizontal Autoscaling | ||
|
||
For detailed steps on enabling horizontal Autoscaling, please refer to our documentation: [How to enable horizontal Autoscaling?](how-to-enable-horizontal-autoscaling.md) | ||
|
||
## Horizontal Autoscaling Pricing | ||
|
||
To know more about the pricing, you can visit our documentation: [Horizontal Autoscaling Pricing](pricing-for-horizontal-autoscaling). |
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,40 @@ | ||
--- | ||
myst: | ||
html_meta: | ||
description: Step-by-step instructions on enabling horizontal Autoscaling through | ||
the Control Panel | ||
title: How to enable horizontal Autoscaling? | Hypernode | ||
--- | ||
|
||
# How to Enable horizontal Autoscaling? | ||
|
||
## What is horizontal Autoscaling? | ||
|
||
Horizontal autoscaling means increasing the amount of resources when your Hypernode needs it, so you don’t need to worry about constant monitoring. | ||
|
||
The amount of extra resources your Hypernode will upgrade with, is calculated by our agent based on the requirements. | ||
|
||
The upgrade is effective as long the extra resources are needed and being used. For more technical overview you can check our docs [How Horizontal Autoscaling Works](how-does-horizontal-autoscaling-work.md). | ||
hrithikkothari1234 marked this conversation as resolved.
Show resolved
Hide resolved
|
||
|
||
## Autoscaling in the Control Panel | ||
|
||
1. Choose the Hypernode you want to enable autoscaling for and then click on the autoscaling option from the vertical navbar. <br> | ||
![Autoscaling option in the navbar](_res/autoscaling-navbar.png) | ||
1. If you already enabled vertical autoscaling, you will have to disable it in order to enable horizontal autoscaling for that Hypernode. This is because we only support one type of autoscaling for a Hypernode at a given time. | ||
1. If you do not have vertical or horizontal autoscaling enabled, you will see the following two choices with an option to help you choose. If you are still confused about which type of autoscaling you would like for your Hypernode please click on "Help me choose". And if the additional information is not enough, feel free to [contact support](https://www.hypernode.com/en/contact/) for more info. | ||
![Choose autoscaling options](./_res/autoscaling-options.png) | ||
1. If you have decided to enable horizontal autoscaling, you can move ahead with the below steps. | ||
|
||
## Enabling Horizontal autoscaling from the Control Panel | ||
|
||
1. Click Configure on the Horizontal autoscaling option. | ||
1. You will see the price of the autoscaling license. If you wish to continue, click configure. This charge is automatically added as part of your subscription when you enable autoscaling ( at a futher step ). | ||
1. Toggle Horizontal autoscaling. This will enable the form to choose the thresholds and the applications you would like to autoscale with. You can choose the certain conditions when autoscaling should be initiated. By default we recommend when the load on the server is higher than 70 for more than 15 minutes | ||
Note: You can also choose the specific time window in which autoscaling should be initiated, if you prefer not to autoscale 24/7. ( E.g. if you have a cron running at 4am that will increase the load on the server but you dont want that to trigger autoscaling, you can choose the autoscaling window to be 5AM-3AM. This will avoid autoscaling between 3AM-5AM) | ||
![Autoscaling toggled](_res/autoscaling-toggled.png) | ||
1. Now click on applications to check if your Hypernode and all the applications on that Hypernode are eligibile for horizontal autoscaling. We have a list of requirements that your Hypernode and it's applications must meet for you to be able to enable horizontal autoscaling. You can see the list [here](https://www.hypernode.com/). | ||
![Autoscaling eligibility checks](_res/autoscaling-eligchecks.png) | ||
1. If an application is not supported, you can either unselect that application so that its not a part of horizontal autoscaling. Or you can update your application as required to make it eligible and re-run the checks. | ||
1. After you have decided the thresholds and choosen the eligible applications, you can now click save changes at the top right corner. This will enable horizontal autoscaling, charge you for the autoscaling license and now your Hypernode will be monitored 24/7 against the conditions you have set. | ||
1. You can always come back to this page to change the conditions and add/remove applications. | ||
1. You can always disable autoscaling by toggling off and clicking save changes after which your license will be removed. And if you wish to come back and enable it, you can follow the same steps above. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
should we say the location of the file here?
etc/varnish/default.vcl etc
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
No, we don't really know where the VCL file is, that's up to the user - they should know where it's located. Once you load a VCL into varnish, varnish doesn't know where the VCL came from. We do backup the VCL to
/data/var/varnish/default.vcl
so it loads in the same VCL when you restart varnish.