Do you keep access logs? How long do you keep them?

I imagine that if you ever got a request, you’d understandably just give in and hand out the data. Have you thought of a canary?

Thanks for all your work!

  • snoweM
    link
    fedilink
    75
    edit-2
    10 months ago

    I’m not in the business of collecting user data and don’t really want to be. In regards to logs, we restart our containers every 6 hours and the logs are wiped at that time, so the furthest back logs I can actually find in our system are from an hour ago.

    And nah, I wouldn’t give in. There’s no real reason to request that information, as accessing a url means absolutely nothing. I did so just now to verify things and the same could be argued by any real user (oh, I clicked on the link and didn’t know what it was going to). I very much doubt the past 6 hours of logs would be useful anyway, as by the time I got the request the logs wouldn’t matter anymore.

    But, I’m still going to see if I can turn off logging for requests. I do not think we need them at all, and if we do, we can simply turn it on for a few minutes to get the info we need.

      • snoweM
        link
        fedilink
        1110 months ago

        Thank you for the good post. And at this point, most of the work is being done by Ategon and the other admins. I am mostly here for infrastructure support and general direction of the instance.

    • Die4Ever
      link
      fedilink
      6
      edit-2
      10 months ago

      as accessing a url means absolutely nothing.

      Hmmm yes I see Google’s search indexer has accessed this URL, they should get sued!

  • @IphtashuFitz
    link
    English
    1210 months ago

    The company I work for probably doesn’t see as much traffic as Reddit, but we provide services via the web in the US and roughly 15 other countries. We make use of Akamai for CDN, security, etc. and one of the things they do is provide us with raw logs of every request made to our sites. That generates a lot of data that we feed into Splunk for analysis, debugging, etc.

    One of the nicer things Akamai does in their logs is to classify if they believe the request came from a bot, and if so then what bot it was. They are able to identify over 1000 individual bots, and can also detect traffic from new/unknown bots. There is a LOT of bot activity on the internet these days, and many originate from cloud providers like AWS, where it’s clear it’s a machine making the request and not a human.

    If we had a legal request for logs I’d have to look at the data to see how to respond. If Akamai showed a lot of bot activity from consumer ISP IPs then I’d likely include that data in an effort to show that end users may be victims of botnets. But if bot activity was mostly originating from cloud providers etc. then I probably wouldn’t include it. Let the lawyers try to figure out from the raw data what traffic originated from humans vs bots.

    • @[email protected]OP
      link
      fedilink
      English
      310 months ago

      Dunno if fediverse instances would be willing to get a lawyer to fight such requests. IMO, the best way to counter it is not collect such data in the first place, but you make a good point about bots. Honestly, I’m just curious what the maintainers will say. I might just start interacting with the fediverse over TOR and be done with it.

      CC BY-NC-SA 4.0