TODO.md

TODO

• New providers
  • collectl
    • Retrieve data: collectl -P -p /var/log/collectl/vmd46640-20191228-124035.raw.gz
    • You can have per-process stats
      • Make plot of the n% most time/memory consuming processes
    • Database is here http://collectl.sourceforge.net/Data-detail.html
    • what's the problem with normalisation? http://collectl.sourceforge.net/TheMath.html
  • Aws
    • load aws ce data
      • You must use the following policy:

      • "Version": "2012-10-17",
        "Statement": [
            {
            "Effect": "Allow",
            "Action": [
                "ce:*"
            ],
            "Resource": [
                "*"
            ]
            }
        ]
        }```
        

      • Beware! You will be charged 0.01$ per request! aws ce get-cost-and-usage --time-period Start=2019-12-01,End=2019-12-24 --granularity DAILY --metrics "BlendedCost" "UnblendedCost" "UsageQuantity" --group-by Type=DIMENSION,Key=SERVICE Type=TAG,Key=Environment --region us-east-1
    • make sure it is really only called once a day. E.g., write file, make sure it persists (logging will start the second day you use it) and increment it.
    • Furthermore, you could setup ip tables to block connections https://ce.us-east-1.amazonaws.com if there are to many per day
      • Rate-limit, like --limit 5/day https://making.pusher.com/per-ip-rate-limiting-with-iptables/
  • IP Tables
  • Pflogsum / Postfix
  • Apache2 Access / Redis / mysql log
  • Mail
    • Define a simple macro to extract stats from emails
• Improve srcCtx-concept
  • Multiplying graphs is good, but we also want to
    • stack lines of different resolution
    • combine lines from different contexts
• make a widget where you can "retrieve, aggregate and stack the most common lines"
  • Useful to display aws usage or the most consuming processes
  • Make sure the order of stacked lines and labels is the same. Currently it is not!
• Report errors during program evaluation in the html / e-mail