do you do that much money from nitro? I actually never seen that many adds in discord about nitro and I didn't even knew it exists (i'm using discord for over a year) and I mean we actually migrated from a 10€ teamspeak to discord. while it is unlikely that everybody in my group would buy nitro, I find it amazing that you do not have many adds in your app, not even about your own service. (i mean sometimes there is the update pop up and some messages between the chat messages, but we basically never use that much of the chat anyways...)
It definitely drops the last word for the cell phone app. I often talk using the desktop client to someone using the cell phone app, and every last word of a sentence she says seems dropped
We use the desktop client and our group often has problems with words being dropped not only at the end but also in the middle of someone talking if they're using voice activation. :-/
It's usually specific users whose words get dropped, maybe their activation threshold is just low enough. But still, seems to me like this should not happen, especially in the middle of sentences, since the beginning of the sentence is recorded just fine.
The messages struggles have been sadly due to issues with Cassandra and GC pauses caused by bugs within it. We have been trying to work with the Cassandra developers to resolve these.
Voice issues should not be happening. Please contact our support with more information and we will gladly investigate.
Thanks for the response, it's good to know what's causing the problems with messages and that it's being worked on. I'll try to contact support next time I have voice issues with my group.
I'm currently far down the database rabbit-hole and have to ask: What's so great about Cassandra that you can't get with CouchDB or other AP (yeah, I know...) databases?
Solid ingestion story. Very very good write throughout. Linear scaling. Easy expansion / contraction. Complete flexibility in consistency vs availability tradeoff.
And most importantly:
It actually works at scale. Huge scale. Thousand node cluster and hundreds of thousands of instances scales.
Because a good chunk of the active maintainers actually run this shit in prod.
We are currently in the process of testing ScyllaDB with double writes for our fixed data clusters. It is very scary to transition to something so new :)
Our message storage clusters have a very large set of data that keeps increasing and using Scylla without incremental repair will suck so we are waiting on that.
What's funny is I got rejected from YC with Guildwork years ago. After flailing around trying to make it work I reflected on why it didn't work, Discord was the result. :)
Ha, interesting. Glad to see you doing so well with Discord, I used to spend a ton of time on the BG forums back in around 2009 or so, and vaguely remember giving some feedback on the very first versions of Guildwork.
Cassandra has a snapshat command that creates a directory by symlinking files that hold data (this is safe cause Cassandra files are immutable). Then you just upload them to your backup storage. This is obviously for recovery scenarios that are catastrophic.
Normally though since the data is replicated on 3 nodes, you can technically loose a node completely and rebuild it from the other nodes.