5 ms·
We use RethinkDB in production and our main frustration lies around the lack of automatic failover. We're looking forward to 2.0, which is supposed to bring aut
by kolencherry 12y ago
We use RethinkDB in production and our main frustration lies around the lack of automatic failover. We're looking forward to 2.0, which is supposed to bring automatic failover (using Raft for consensus) to RethinkDB.
- tracker1 12y agoThat was one of my two bigger issues earlier on... the other one being geographic based searches (which is now implemented iirc).
- coffeemug 12y agoYep, geospatial support has been in since 1.15.
- coffeemug 12y agoSlava @ RethinkDB here. Unfortunately automatic failover won't be a part of 2.0, but it will happen very quickly after that. Please hang in there, we expect to ship this feature some time in May. I just saw a demo of the failover feature yesterday from Tim Maxwell (the lead engineer on this), and it's really impressive! Another side benefit of this feature is live reshards -- you'll be able to reshard/rebalance data without any availability loss on the cluster. The code is there and just needs a bit more polish and a lot of testing. I'm very excited to get this out, it's probably the last part of RethinkDB that I'm not 100% proud of yet (but will be in a month or two).
- TylerE 12y agoYou guys are killing it. Wish I had a product I could write around rethink....currently at the day job our stuff is mostly Mongo....all layered under django-nonrel with lots of mongo crud so a port wouldn't really be an option I don't think.
- hardwaresofton 12y agoAs a person who agrees -- maybe you could write a port/adapter for django-nonrel to Rethink? Also, why not start a new greenfield project to test out Rethink? a something something realtime something geospacial something app should be a fantastic way to kick the tires, since that's one of the things that Rethink does really well out of the box (as of 1.15) compared to other databases (relational or not)
- StavrosK 12y agoI am rather horrified to hear of your setup, how has it worked out for you in practice?
- TylerE 12y agoIt works, I guess? I was brought on well after the system was originally developed. The websites are mostly our internal admin tools anyway. Most of the real work is run through cronjobs or task queues (Celery). The biggest annoyance is that the Django version the stable django-nonrel is based on is ancient (1.3). There are non-stable branches to newer Djangos (1.5, I think?). When I investigated there were some issues with them so we're still on 1.3.