OOM catalog

[kfox1111]

I started a minikube with 8g of ram.

I installed the olm:

curl -sL https://github.com/operator-framework/operator-lifecycle-manager/releases/download/0.14.1/install.sh | bash -s 0.14.1

Most things came up but the catalog was in a crash loop for a while:
NAME READY STATUS RESTARTS AGE
catalog-operator-86f777cc59-4qrq7 1/1 Running 0 16m
olm-operator-5fd48d8cd4-gcm44 1/1 Running 0 16m
operatorhubio-catalog-ghkjm 0/1 Running 7 16m
packageserver-75946c9b59-2wrmj 1/1 Running 0 16m
packageserver-75946c9b59-vnfjw 1/1 Running 0 16m

doing a minikube ssh dmesg:
[ 738.836068] Memory cgroup out of memory: Kill process 10034 (registry-server) score 2018 or sacrifice child
[ 738.836350] Killed process 10034 (registry-server) total-vm:195796kB, anon-rss:92640kB, file-rss:14324kB, shmem-rss:0kB

There is plenty of ram left on the system, so it seems like the limit set on the pod is too low.

[AndrienkoAleksandr]

operatorhubio-catalog-ghkjm 0/1 Running 7 16m

+1. operatorhubio-catalog is very unstable and
CrashLoopBackOff restarting...

[andresmgot]

I am also finding this issue. I see it was fixed here: #1389

But the latest release (0.14.2) doesn't include this change. When can we expect a new release? I am using latest for the moment for the image in the catalog-operator deployment but that's far from ideal.

I see in this comment: #1415 (comment) that @ecordell is planning a release, any update?

[kfox1111]

I think this is fixed, or at least worked around in 0.15+.