search

gaffneyc / sunspot-queue

Background search indexing using existing worker systems
MIT License
35 stars 37 forks source link

Index and remove using sunspot autocommit #16

Closed lucasas closed 9 years ago

lucasas commented 10 years ago

I've done some refactoring, fixed some specs that are failing and finally Index and Removal classes are using sunspot autocommit by default.

lucasas commented 10 years ago

Hi @gaffneyc do you have any considerations about my pull request? Thank you.

gaffneyc commented 10 years ago

@lucasas Hey, I'm really sorry. I haven't had a chance to review this yet. Quick look seems good but never that easy. I'll get to it in the next day or two (for sure).

Worth noting the Sidekiq breakage was due to some refactoring inside Sidekiq (moved something but didn't require it's dependencies).

lucasas commented 10 years ago

Thank you @gaffneyc. Let me know when you see the code :smile:

ghempton commented 10 years ago

Any update on this?

gaffneyc commented 10 years ago

@lucasas @ghempton Wups! Looking at it now.

@lucasas I'm seeing a couple fixes (rails version, sidekiq require issue). But it looks like the main goal is that when a change is made it's auto committed to the solr index. If that's the case then it's working as intended. I would take a look at this commit

gaffneyc commented 10 years ago

Also this comment from the readme. We ran into issues with each update causing a commit of the index in production and it quickly overwhelmed Solr.

lucasas commented 10 years ago

Hi @gaffneyc this makes a lot of sense. I'll try this option. I don't have problems with performance, so I didn't realized this problem.