Fucking elastic appsearch, too many requests per seconds and it dies. It doesn't go slow first, it just dies. No warning just a timeout that lasts until a manual restart in elastic cloud console. Besides being apparently the shittiest product in the elastic stack, it's also the worst documented. And yes I just scaled it up but not being able to handle indexing 100 documents per second with 8 vcpu and 8 gb memory is a shame.

I have no idea about appsearch but ES can scale really well with high read/write traffic given enough power and cluster configuration

@asgs that's making it so weird. Elastic search is a technological wonder. Have been self hosting it for a couple of years on a bare metal kubernetes cluster. And never had any issues with it at all. It always self healed and scaling was never a issue. However appsearch is a complete different piece of shit. It runs on top of elastic search, but has so many unaddressed issues. Really bad extremely scarce documentation is one of them. Unmaintained drivers and examples is another one. And in general it tries to abstract so much that it gets a pain in the ass to work with. Anything es is doing really great is not available in appsearch. It is just a dumb abstraction.

@romperstomper ah, good to hear you did manage ES on your own

So, why doesn't an 8 vCPU, 8 GB RAM machine handle your indexing load? Are you able to get any insights on the underlying ES setup? The shards are all as uniformly distributed with data as possible? Or is that abstracted and made opaque?

📌

You should probably learn how to implement denouncing.