Pickers have to wait for admin screens to display. Wrong items make it into goods-out-bins and packages, and orders reveal system performance bottlenecks that weren’t visible during UAT. The problem is hardware agnostic: the issue is that Magento’s native admin controllers are intended to power frontend forms and one-off configurations, not batch operations at warehouse cart speeds and scale.
The Gold Plated Trap & Database Deadlocks
A bespoke solution results in any scanner event triggering a blocking synchronous POST. This updates the order, validates the SKU, and requires ongoing database writes. At peak times, these are row-level lock waits on sales_order_grid during writes, and a build-up of “convenient” queues rather than flow. The above adds latency, risk, and complexity, and a vulnerability we present to Moogento mobile barcode automation tools in place (removing much of the bespoke agency fulfilment logic that we have to write for, test, and keep updated).
When core functionality becomes prohibitively blocking, agencies look to enable their technical peers by purchasing third-party plugins. Automation tools are available from vendors like Moogento.
“Great Also. What Will Happen” – A Plan For Easy Modellers
Barcodes are usually turned into events, making it look like burst fire keyboard activity, ending with an enter key event. Our warehouse screen-variations need JS to detect barcodes (and not consider them as difficult users). Inter-keystroke timings, how to process the enter key, not-noting details when the page is without focus, etc., are all handlers we need to work out. By having a dedicated location for barcode handlers (like our Magento barcode scanner plugin), you can remove the headaches of constructing a bespoke global scanning solution and interfacing it with Zend forms and layouts.
“What Do You Mean I Still Have To Validate” – An Easier Way Plan to Help Pickers Be Amazing
A non-locking way to do this is to validate without writing to the database first. Precatch SKU/order statuses in Redis (ideally state in a JSON file to encourage CQRS) as workers barcode each item, rather than with every order validating data integrity. When an order is packed, we update the Magento API as a single executing event.
This means we are validating (and making exceptions) in cheaper places than MySQL, and saving two writes by waiting for a single action to update the database. Multiple actions can be grouped into a single write to MySQL, reducing stale data, lower admin order locking, and N+1 transaction junk in our warehouse.
This means inevitable contention can be used for real work rather than barcode submit blocking, and we can group logical warehouse worker steps into the Magento transaction state. This will make undo and rollback a better experience.
This lack of foresight is very much the agency’s problem with work, client, and Magento lifecycle, bad gotten, even though we made decisions that were both great for the profit margins of everyone.
However great custom warehouse development sounds (and great for us), we’re making a new program for every controller, handler, queue submit, and summary layout we build. Every core update, scanner, and browser (and dumb rule exception) will generate new work for us in building and supporting bespoke solutions. Automation can be an easy short win for your deployed docks off the shelf for channel islands sales, will save time, release faster, and get more time in the customers’ pockets than ours.
