Open ron1 opened 4 years ago
Perhaps related to the case in #147 - because if you scale the noobaa-core sts replicas the result is undefined. Is it the same case?
@guymguym No, I encountered this with the out-of-the-box single instance of noobaa-core.
Hey @ron1
OK, so the logs you attached show that we got disconnected from the DB. We've seen this symptom occurring mostly when low on cpu/memory resources, so lets start by checking that first.
Can you get the cpu/memory resources allocated to the noobaa-core-0 pod - both to the core
and db
containers? you can get those with:
oc get pod noobaa-core-0 -o json | jq '.spec.containers[] | .name,.resources'
Then, in order to change the allocated resources you can update the noobaa spec and set it as shown below and the operator will reconcile to the statefulset and restart the pod. The numbers below are currently used in our performance testing, and you can use lower numbers depending on your use case.
apiVersion: noobaa.io/v1alpha1
kind: NooBaa
metadata:
name: noobaa
namespace: openshift-storage
spec:
coreResources:
limits:
cpu: "16"
memory: 32Gi
requests:
cpu: "8"
memory: 16Gi
dbResources:
limits:
cpu: "8"
memory: 32Gi
requests:
cpu: "4"
memory: 16Gi
Environment: OCP 3.11, Rook Ceph 1.1.6 block provisioner, NooBaa 2.0.8-SNAPSHOT including PR #146
Pre-req: Configure two pv-pool BackingStores with one volume each wrapped in a Mirror BucketClass. Then define a StorageClass for this BucketClass and create an OBC. Use rclone to upload a large Maven repository to NooBaa.
Expected Result: Files are successfully uploaded.
Actual Result: After 60 seconds, 500 Internal Errors are reported. A snippet from the nuxeo-core-0 log is included below.