Some indices are corrupted after dgraph server restart


(Jzhu077) #1

an example node in the database:
person{
name: “foo”
age: “50”
gender: “male”
}

After rolling update (kubernetes) that increases the memory configuration from 3GB to 6GB a query that looks for a person that with a filter age equals “50” failed, but a query that searches for a person named “foo” works and the result contains age: “50”.

Since I have a load balancer in front of all dgraph server, a query that looks for a person with filter gender=“male” sometimes returns results about 50% of the time.
That indicates the index “gender” are corrupted in half of the pods.

Any clue about it?


(Pawan Rawal) #2

That’s strange. We’ll need some information about this to reproduce this. Whats your setup like? Are you sure that all your servers are running dgraph/dgraph:master image and not latest?


(Jzhu077) #3

yes, I am using image dgraph/dgraph:master. It’s quite complicated to reproduce. I will see if I can isolate the problematic part.


(system) #4

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.