0

I would like to try and better understand how to analyse the following situation (which is no longer occurring)

  • CPU pinned at 50% all night (100% of 1 core I suppose) according to Cloudwatch.

  • Network usage all night: almost completely flat at 1MB/5 mins

  • Http logs mostly empty.

  • My computer left on all night, connected to the instance with SSH and a database connection.

  • All of the above stops when I unplug my ethernet connection in the morning.

  • Has not occurred before, no longer happening.

I'm wondering if there's a way to analyse the network usage by IP, port, or the cpu usage by process, retrospectively, or if there's any other approach that I could use to learn more about what happened.

Tom
  • 101
  • Sorry for not answering the question, but in case you catch the problem in flagranti, consider installing interactive tools: `top` or its derivatives like `ntop` or `htop`. Also `iptraf` to see which remote IP addresses are involved. You can also run them in batch mode. – berndbausch Jan 23 '21 at 01:46
  • 1
    VPC Flow Logs can show where traffic is going. They're a bit tricky to read but the AWS documentation helps. top or similar is good for working out CPU use. – Tim Jan 24 '21 at 06:47

0 Answers0