Loading...

XML

Word

Printable

Story Points:
3
Blocked:
False
Blocked Reason:

Hide

None

Show
None
Ready:
False
Release Note Text:

Hide
= Out-Of-Memory errors in the {product} Operator

To resolve Out-Of-Memory errors in the {product} Operator, adjust memory settings manually:

. Edit the deployment:
+
----
kubectl edit deployment rhdh-operator -n <namespace>
----

. In the editor, set the memory request to 1 GiB and the limit to 2 GiB:
+
[source,yaml]
----
spec:
  containers:
  - name: rhdh-operator
    resources:
      requests:
        memory: "1G"
      limits:
        memory: "2G"
----

. Reapply this change whenever the Operator is actively reconciled. The pod should become healthy, using about 1.07 GiB of memory.

Show
= Out-Of-Memory errors in the {product} Operator To resolve Out-Of-Memory errors in the {product} Operator, adjust memory settings manually: . Edit the deployment: + ---- kubectl edit deployment rhdh-operator -n <namespace> ---- . In the editor, set the memory request to 1 GiB and the limit to 2 GiB: + [source,yaml] ---- spec:   containers:   - name: rhdh-operator     resources:       requests:         memory: "1G"       limits:         memory: "2G" ---- . Reapply this change whenever the Operator is actively reconciled. The pod should become healthy, using about 1.07 GiB of memory.
Release Note Type:
Known Issue
Release Note Status:
Done

Sprint:
RHDH Install 3270, RHDH Install 3271, RHDH Install 3272, RHDH Install 3273, RHDH Install 3274, RHDH Install 3275, RHDH Install 3281, RHDH Install 3282, RHDH Install 3283, RHDH Install 3284, RHDH Install 3285
Severity:
Critical

Description of problem:

After installing the Operator version 1.4.1 the deployment rhdh-operator becomes briefly healthy before being killed and restarted with an OOM error.

The operator manages a single Backstage Instance.

Install the Operator version with a single Backstage instance present
Observe that the rhdh-operator deployment requests 128Mi of memory with a 1Gi limit
Observe that the operator becomes healthy for less than 10 seconds before being OOM killed
Manually set the memory request to 1Gi and the limit to 2Gi
Observe that the pod now becomes healthy and that the memory tops out at 1.22 Gi and stabilizes at 1.07 Gi
With the next operator update/reconciliation the memory is reset and the game starts anew

see above

The operator should have a valid memory configuration out of the box for all common use cases.

Always