Implement parallel writes

This is a big feature that would allow Etcher to be able to stream the same image to multiple drives at the same time. This feature should be implemented by extending the already existing etcher-image-write NPM module. The module currently accepts the path to a drive device as an argument to the main function. If the user passes an array of drives, the multi-writer logic should come in.

Some of the challenges for this task are:

• Cross platform support.
• Validating the drives at the moment the writing starts.
• Handling validation correctly for all of them (there are some quirks to make validation work in certain operating systems, like in OS X, where we have to use OS X specific frameworks to prevent the drives from being mounted right when the flash ends, so the operating system doesn’t write files like .DS_Store that messes up the checksum).
• Handling error conditions correctly (e.g: if a certain drive fails, the rest should continue).
• Making sure the whole stream is streamed to all devices (e.g: no data is lost because it has been piped to another drive).
• Imposing a reasonable limit of drives that can be written at the same time, depending on the capabilities of the operating system to prevent potential kernel panics, etc. If a user tries to write to more images than he should, we should make of use of a queue.

The module returns an EventEmitter instance which emits state information. When multiple writings are taking place, the emitted events could have a property containing the drive they correspond to.

It’s important that this feature contains extensive end to end tests.

This project should not contain native addons by any means, since it’s used by other projects that are installed by end users without necessarily having a toolchain in place.

[jhermsmeier]

Resolved by #2124

[sonttran]

Same issue, same setup with @unphased on Etcher Version 1.5.100. Any way to improve speed to full in parrallel, @jviotti ? That would be great.

I’ve flashed about 20 cards now and I have noticed that the time taken more or less scales linearly with the transfer speed of one card. e.g. if I have 3 cards that are independently capable of flashing @21MB/s, flashing 3 of them simultaneously has them proceed at almost exactly 7MB/s, making it so that I have to wait 3 times as long for the batch. Now, this is a process improvement and already a big win, however I am confused as to why there is no speedup as you would see if you just ran concurrent dds on multiple drives. The reason I use etcher is so I can avoid building a dd runner and figuring out how to properly verify a disk, but this seems… not ideal.

This is using a MacBook with multiple usb-c to sd card adapters.