search

influxdata / influxdb-relay

Service to replicate InfluxDB data for high availability
MIT License
853 stars 350 forks source link

data inconsistency #69

Open tessus opened 5 years ago

tessus commented 5 years ago

I was looking at the architecture and I'm a bit puzzled by the fact that data inconsistency can occur and I'm wondering how I'm supposed to fix those.

Here's an example:

So now I have 2 databases with different data. Although A has all the correct and complete data. Now imagine the same process as above again, but B gets the successful write and the client also gets the success message, because the write to B was first (and successful). In this case now both databases have incorrect and incomplete data.

Or am I missing something here?

kashifmin commented 5 years ago

@tessus You're right. I've noticed the same. Do you know of any alternatives?

tessus commented 5 years ago

Nope, but I guess that's the reason why they are selling the influxdb-cluster. They don't even want to solve this in the free relay code....

rockyluke commented 5 years ago

Please take a look on https://github.com/veepee-moc/influxdb-relay.

We tried to implement some useful features without conflicting with the paid version in InfluxDB (which is really great).

jhg03a commented 5 years ago

That's a lot of really interesting work @rockyluke, was there ever a PR to upstream it?