scaling down issues about k8s-mongo-sidecar HOT 6 CLOSED

Sorry for the delay! Should be fixed with v0.5.0

from k8s-mongo-sidecar.

The issue with this was the parsing of the DateTime of the last heartbeat. It was coming back NaN for me, at least. I submitted a PR with fixes, and verified it was removing. In the meantime, I've posted an image on DockerHub at jhughes2112/k8s-mongo-sidecar

from k8s-mongo-sidecar.

@rosslutsch Any information when you run kubectl logs YOUR-POD-NAME mongo-sidecar?

from k8s-mongo-sidecar.

@MarsssBlkm here is what I understand so far.

• With a member missing because of the scaling down the quorum is lost and it logs the following every few seconds

Pod has been elected as a secondary to do primary work
Pod has been elected as a secondary to do primary work

• as there is no primary all the remaining pods become slave
• it seems the remove member function isn't working or it's not picking up that there is a member missing . it doesn't appear in the members to remove [] in the log either.

from k8s-mongo-sidecar.

@rosslutsch
I have met the same problem. It seems like the problem is with inReplicaSet function in src/lib/worker.js.
It only checks if (member.state === 1) while after scaling down the members which are supposed to be removed has state == 8.
Once it has found a member with state === 1 the loop breaks.. so it doesn't look at the others anymore...?
This is just my guess and I am new to k8s and node.js so I could be wrong..

from k8s-mongo-sidecar.

@rosslutsch and @MarsssBlkm I will have a look at this and see what I can do. Thanks for pointing out a starting place @MarsssBlkm. Not sure if they accepting pull requests at the moment but I will fork this on my side.

from k8s-mongo-sidecar.