0

Background of the issue : We are using Magnolia CMS with customized UI. As a first step of using Magnolia, we are migrating old content including documents to Magnolia. Content migration is working fine. Issue with DAM. We have more than 50 GB of historical content to be stored in Magnolia. Storing in DB became very expensive. We decided to save them in File system. Magnolia running in Kubernetes cluster as a service.

Problem: After we migrate dam from old system to magnolia, we are trying to publish asset from author to public instance. After MAx 200 documents published, It is killing the pod. In logs, we are not seeing anything. Last message is Catalish.sh killed...

Could you please suggest if anyone come across such scenarios. Best practices to implement dam and other configuration.

In Datadog, have seen that threadcounts are always more and GC is going crazy.

Thank you in advance.

Mamatha Shivanna
  • 15
  • 3
  • 2
    do you have set any request limit ? can you describe the pod and check the status code and reason of pod crash and restart? – Harsh Manvar Feb 11 '21 at 08:38
  • In one sentence you mention migrating assets from DAM to FS and in another you mention trying to publish assets from DAM, so which is it? Did you move the assets out or not? As comment above mention, you need to figure out reason for pod crash. I'd guess it gets killed because it becomes unresponsive. Why is that? What is your metric? Describe the healthcheck call you make to figure out if instance is still alive? It might be that by publishing a lot of assets in parallel (if that is what you are doing) you overwhelm the instance and it has no bandwidth left to respond to your healthcheck call – Jan Feb 11 '21 at 12:16
  • Thank you all for your feedback. we configured JMX configuration for this services. We found that old GC was going high during both migration and Publication process. We analyzed our code and found that some sessions were not closed as expected not taken care by default. We explicitly closed connection with respect to DB session and some IO operations. Then GC is fine and it is not crashing the system. – Mamatha Shivanna Feb 15 '21 at 17:04

0 Answers0