Skip to content

Conversation

@andy-stark-redis
Copy link
Contributor

@andy-stark-redis andy-stark-redis commented Jan 23, 2025

DOC-4753 (based on this Confluence page).

Is there any more to add to this before merging? Also, the K8s installation section says that you should download redis-di - can you actually download this separately?

Copy link
Collaborator

@dwdougherty dwdougherty left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just one minor recommendation. Otherwise, LGTM.


### Recovering from failure during a Kubernetes upgrade

If you get an error during the upgrade or some deployments are not OK, then
Copy link
Collaborator

@dwdougherty dwdougherty Jan 23, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If OK here is referring to a state shown in a command's output, then using OK is okay. :) Otherwise, use okay. This is not a Google style guide issue per se (it uses both OK and okay to mean the latter), but I've seen this rule on other style guides. Up to you.

Copy link
Contributor

@galilev galilev Jan 30, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@andy-stark-redis maybe we should add:

Run the command sudo k3s kubectl get all -n <namespace> and verify that all the pods are running and that the READY column for all the pods is 1/1. E.g. fornot okstate:

<pod_name> 0/1 CrashLoopBackOff 1881 (91s ago) 6d17h

Copy link
Contributor Author

@andy-stark-redis andy-stark-redis Jan 30, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done. (I've combined this section with the one above about verifying the installation, btw, since the recovery info is quite short and straightforward.)


### Recovering from failure during a VM upgrade

If the previous version is v1.4.4 or later, go to the `rdi_install/<NEW_VERSION>` directory
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

my mistake, it's not the NEW_VERSION, it's the previous version

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Fixed.

[deploy]({{< relref "/integrate/redis-data-integration/data-pipelines/deploy" >}})
again after this step.

1. Download the latest `redis-di`.
Copy link
Contributor

@galilev galilev Jan 23, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@yaronp68 can we remove it? They don't need it for the k8s installation.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, let's remove it. Also, they cannot just get the latest CLI, it's part of the installation package.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@andy-stark-redis please remove line 109

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done.

docker pull redis/rdi-operator:tagname
docker pull redis/rdi-api:tagname
docker pull redis/rdi-monitor:tagname
docker pull redis/rdi-collector-initializer
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@yaronp68 what about the collector-api?

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, we should also include it. It will not be an optional component anymore, so let's just add it here.

@galilev , please provide the command so that @andy-stark-redis can add it here.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@andy-stark-redis please add docker pull redis/rdi-collector-api

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done.

- integrate
- rs
- rdi
description: Learn how to install RDI
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
description: Learn how to install RDI
description: Learn how to install and upgrade RDI

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done.

### Recovering from failure during a VM upgrade

If the previous version is v1.4.4 or later, go to the `rdi_install/<NEW_VERSION>` directory
and run `sudo redis-di upgrade`.
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@galilev If the previous version is higher than 1.4.4, shouldn't we be using the upgrade.sh script instead of the CLI?

Copy link
Contributor

@galilev galilev Jan 30, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done.


### Upgrading a VM installation with High availability

If there is an active pipeline, the upgrade process will involve upgrading RDI on the active
Copy link
Collaborator

@ZdravkoDonev-redis ZdravkoDonev-redis Jan 30, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@galilev Why upgrading first the active version? Wouldn't you have 2 downtimes this way?

Scenario 1 - upgrade first the active RDI (RDI_instance_1) then the passive RDI (RDI_instance_2):

  1. Upgrade the Active RDI (RDI_instance_1) ---> Downtime, the passive RDI (RDI_instance_2) becomes the active one, since the active is now down
  2. Upgrade the newly active RDI (RDI_instance_2) ---> Downtime, the newly passive RDI (RDI_instance_1) becomes the active one, since the newly active RDI is down

Scenario 2 - upgrade first the passive RDI (RDI_instance_2) then the passive RDI (RDI_instance_1):

  1. Upgrade the passive RDI (RDI_instance_2) ---> no downtime, upgrade succeeds
  2. Upgrade the Active RDI (RDI_instance_1) ---> Downtime, the passive RDI (RDI_instance_2) becomes the active one, since the active is now down

Or am I missing something?

Copy link
Contributor

@galilev galilev Jan 30, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@ZdravkoDonev-redis The upgrade won't trigger a switchover. We plan to upgrade the active RDI first, followed by the passive. @yaronp68 asked whether switching to RDI instance 2 and then upgrading RDI instance 1 would result in zero downtime. However, since a switchover itself requires downtime, this approach wouldn’t be beneficial

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Okay

@andy-stark-redis andy-stark-redis merged commit 47d101d into main Feb 7, 2025
5 checks passed
@andy-stark-redis andy-stark-redis deleted the DOC-4753-rdi-upgrade-guide branch February 7, 2025 13:28
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

6 participants