Good morning, I have been facing a problem that I don't know where its coming from. Datahub works just fine but sometimes the front container falls (then the new container starts without a problem and the front is available again, but with a period of some minutes of failure) and prints the following error:

[Thread-3] INFO  play.core.server.AkkaHttpServer - Stopping server...
[application-akka.actor.default-dispatcher-227] [akka.actor.ActorSystemImpl(application)] Response stream for [GET /assets/static/js/2.f8eb32eb.chunk.js] failed with 'Processor actor [Actor[<akka://application/system/StreamSupervisor-0/flow-2327-0-detacher#-1437369062]]> terminated abruptly'. Aborting connection. (akka.stream.AbruptTerminationException: Processor actor [Actor[<akka://application/system/StreamSupervisor-0/flow-2327-0-detacher#-1437369062]]> terminated abruptly)

I am currently using the last release of such image (

linkedin/datahub-frontend-react:v0.8.44

), does anyone know why this happens or has had this error before?? Thanks in advance!

Hi @microscopic-mechanic-13766 are you still facing this issue ?

Hello @hundreds-photographer-13496, so I have not faced this issue again but, as I haven't changed anything in my deployment, I am expecting to face it again along this week. I will keep you updated if it happens again. Do you know, by any chance, why it happens??

Where is your DataHub running? Docker compose? Kubernetes? Have you set resource requests/limits if set on your containers? I have not seen this particular error so trying to rule out most common problems of limited resources causing the container coordinator sending a kill signal.

It is running on Docker Swarm. I haven't specified a limit to the container so it should be able to take all the resources needed. Initially it didn't happen, but with the latest version changes it has been becoming recurrent. Is any additional configuration needed?? What does the

DATAHUB_PLAY_MEM_BUFFER_SIZE

variable do?? Is it needed to be/not to be on specific values??

I am not very familiar with this part of the code base. We will have to wait until the U.S. team members who are more familiar with this part of the codebases are up and can look at this.

Perfect!