ARTIFACTORY: How to Switch Artifactory to Masterless (Cloud-Native) Architecture

How to switch to cloud native/masterless

Starting with version 7.17.4, Artifactory supports a masterless architecture. This means there is no longer a dedicated primary node, and all nodes in the cluster operate as equal members

The setting is enabled by default from version 7.17.4. That means new installations will have this feature. If we upgrade from a lower version to any version above 7.17.4, we can enable it by navigating to $JFROG_HOME/artifactory/var/etc/system.yaml and adding the parameter taskAffinity to the value any and performing a restart.

shared:
    Node:
        taskAffinity: any

Please refer to the documentation here for more details.



What should you be aware of before switching to Cloud native?

Before Artifactory version  7.17.4, the primary node is responsible for the following tasks:

```
Garbage Collection
```
```
Backups
```
```
Imports
```

Archive indexing (responsible for opening and mapping archived files)

```
Maven indices (for IDE only) generation
```
```
Xray event generation and propagation
```

That means, if the primary node is down, the above tasks will be paused (But the member nodes will be serving all other requests).

Before this version, there was a possibility that the primary node might have been tuned with high resources like memory, in order to accommodate the above tasks.

If we are switching to Cloud native, we should make sure that the member nodes are also tuned according to the load of the server with enough resources, so that all nodes will perform all the tasks. This avoids performance issues further:

Make sure the member nodes are aligned with enough resources, as mentioned in this confluence page.

If your server is under heavy load, you may refer to this knowledge base article for tuning all your nodes.