Service unresponsiveness when hosted on EC2

[badolamgk]

Hello, long time user of your project.

I have been facing a very peculiar issue which seems to keep coming back on a frequent basis. I run a custom fork of Remark42 on an AWS EC2 instance, which has nginx to handle requests from public internet to the Remark42 app. It chokes up somewhere, which leads to the EC2 instance not responding and all the request URLs throwing Timeout (no headers received) status. I have to then stop the instance and then start it, and then run compose to start the application again. Simply running the Reboot Instance option doesn't work.[1]

The fork doesn't have any custom processing logic, it is just some frontend changes and a new import command for adding MSN comments. There is no pattern found when the downtime have occurred and I have been scaling up instance type but it is still persisting.

To give context about traffic, the website has been having pageviews in the range of 500k-2M, and we use <iframe> tag to render the comment interface.

I need guidance as to how can I detect the issue or reduce the downtime. If I am using orchestration like Kubernetes, how can I ensure that the database statefulness is maintained across all nodes?

Thanks.

[1] Instance choking up at 17:04 PM

[umputun]

The troubleshooting should be done for the EC2 setup, not Remark42. You should probably reach out to AWS support and ask them to help with this diagnostic. From what I see in your monitoring graphs, it looks like you are using some T-family instances and are running out of credits. If this is the case, you can configure your instance to automatically increase the balance; it has an option for this.