UploadManager | PHP upload manager , with ability to chunk upload | File Upload library

I am unable to run the repro you provided because it appears to require private API keys.

However, inspecting the code, here's what appears to be happening:

1. The component renders for the first time with reporting state as false. This causes the component to render, and a new instance of Uppy to be created by the effect within that component and stored in this instance of the 's state.
2. The user uploads a file, and hits the "continue" button, which sets reporting state to true. This causes the component to re-render, and show the component instead. The original component is unmounted, and the things that were present in its state, notably, the instance of Uppy that had the user's uploaded file, are lost.
3. The user hits the "Back To Photos" button, which sets reporting state back to false. A this causes a new instance of to render, and the effect that creates an instance of Uppy re-runs. The instance that is created is new as well, and thus won't show what it contained the last time it was rendered.

One way to solve this would be to lift up state that needs to remain the same as the is mounted and unmounted into the parent component. So the component would be responsible for creating the Uppy instance, and pass that as a prop to the .

See also this part of the Uppy docs:

Functional components are re-run on every render. This could lead to accidentally recreate a fresh Uppy instance every time, causing state to be reset and resources to be wasted.

The @uppy/react package provides a hook useUppy() that can manage an Uppy instance’s lifetime for you. It will be created when your component is first rendered, and destroyed when your component unmounts.

You could also re-factor your app in other ways if it feels like too much is happening in a single component. But the guiding principle would be: "don't create the uppy instance more than once in a single user flow".

as far as I know this is why the first exception arises.

When ActiveMQ Artemis uses replication, the live and the backup servers do not share the same data directories, all data synchronization is done over the network.

Upon start-up the backup server will first need to synchronize all existing data from the live server before becoming capable of replacing the live server should it fail. So unlike when using shared storage, a replicating backup will not be a fully operational backup right after start-up, but only after it finishes synchronizing the data with its live server.

Why does a single live-backup pair look working?

The backup node at start-up executes the following steps at SharedNothingBackupActivation:

• Initialize (it looks running)
• Waiting on cluster connection (it is starving here)
• Starting backup manager

So in a scenario with a single live-backup pair the backup node doesn't complete the initialization.

The fundamental problem is noted in the log from the master broker: