Can't get any logs for GameLift fleet stuck in activating state while game build is constantly crashing

0

I've tested and tried everything I could imagine:

  • I've tested my GameLift server SDK integration with GameLift Anywhere;
  • I've tested my build running in 'amazonlinux' container;
  • I've tried to force log into CloudWatch to get realtime logs (didn't see any logs coming);

The only clue GameLift gives me is this error message: Server process exited without calling ProcessEnding(), exitCode(0), launchPath(/local/game/My.GameServer) It's really hard to iterate, because Fleet is stuck in an Activating state for an hour, can't be deleted, and there are limitations for the types of ec2 instances and, for some reason, there is a quota for only 1 fleet(in a region?) for me (due to new/free-tier account may be?) It's very frustrating and disappointing.

What else could I do to find out the crashing reason?

profile picture
asked 10 months ago341 views
1 Answer
1
Accepted Answer

Hello,

The comments and answers to this question may help your investigation: https://repost.aws/questions/QUOYWPiFAjQ0iwAb6AFNSHJw/initsdk-timeout-using-gamelift-sdk-5-ue-5-1

The main items to checkout:

Lastly, you can check out AWS Service Quotas or the Support Console for requesting increases to your resource limits.

If you're still facing issues, I recommend creating a Support ticket so the Amazon GameLift team can take a deeper look into your specific use case.

Regards,

Jackson

AWS
answered 10 months ago
  • Thank you, Jackson! The problem lay with the ServerSDKVersion provided to the Build resource. (I knew it was something simple as that!) I was creating it with the CDK GameLift Alpha library, which doesn't allow for setting the SDK version for the Build construct (I suppose that's the risk you take when using Alpha libraries). I've reverted to using the lower-level construct for the Build, set the correct ServerSDKVersion (5.0.0), and now my Fleet faces no issues and successfully transitions to the ACTIVE state.

  • However, I'm somewhat confused as to why you recommended connecting directly to the instance, given that it's only possible with an activated fleet (according to the documentation), even though I mentioned that mine was stuck in the activating state?

  • Glad that solved your issue!

    It appears that documentation may be misleading, you are able to connect to the instance in an ACTIVATING fleet. I've alerted the team to look into updating that document to be clearer.

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions