RAM, CPU, OS, if you’re using a cloud service, Dgraph version and all information you can give about that.
So, others are seeing this too…
Well, I’m doing a lot of things at the same time. I’ll have to sync all to work on this reproduction.
RAM, CPU, OS, if you’re using a cloud service, Dgraph version and all information you can give about that.
So, others are seeing this too…
Well, I’m doing a lot of things at the same time. I’ll have to sync all to work on this reproduction.
We are running a Dgraph cluster using k8s on AWS - it’s version 1.2 of Dgraph, there are 3 Zeros and 3 Alphas with such resources:
dgraph:
alpha:
resources:
requests:
cpu: 1
memory: 1Gi
fullnameOverride: intouch-graph-engine
zero:
resources:
requests:
cpu: 100m
memory: 200Mi
I have increased resources for Alphas to address some performance issues we had with our queueries (but this wasn’t related to this issue) as compared to the initial report
I did a local node to test what I understood from this issue. I’ll let it run at the weekend and see how it behaves. I can’t guarantee that I’m doing it right. Cuz I can’t see how I can do a reproducible test. I need concrete information to reproduce and push to the back-log a request to fix it. If it is reproducible.
As far as my test is running, I don’t see errors in eq() func. I have 1793547 nodes at this moment.
I think the case here is that you must run Dgraph for a while, just import some initial set of data, leave the instance running for a few hours and check if eq
based filter works after that time. I will try to simulate the same thing tomorrow.
Thanks!
Hey @lukaszlenart,
Let me know if the issue still persists?
Looks like the problem is gone, but I’m still going to minitor it.