Recent searches
Search options
This is an experiment. Please boost.
Here's the idea: This post is going first to my followers, then, if they boost it, to other people. This domain has been registered for only this experiment. I should see in my web server's logs when mastodon instances start crawling the site for info. Then maybe also some curious humans.
I just want to play with my monitoring a bit :)
OK, it's bedtime. Will do some numbers tomorrow :)
@mdione
I prefer not to open the link, just to not pollute the logs :-)
@caprieldeluca I can in fact differentiate a crawler from a human. Or so I should :)
@mdione is this GDPR compliant tho 
@djh the only thing I will have for a while is some IPs. `logrotate` will get rid of them in some 15 days. I'll just play with prometheus and grafana tonight :)
@djh OK, I also get referers/referrers/referees :-P
@mdione it told I'm either a crawler or human - but I'm a dragonfox :<
(share statistics? :3)
@littlefox I'm sorry, I'm rather new to non binary-ism :) Fixed.
Thank you all for being inclusive of the nonhuman crawlers.
@mdione Do you plan to publish your findings?
@marcel yes, although they don't seem to be _that_ interesting...
@mdione @franskeijer yeah, because ddns is noip.com subdomain... I would expect some blacklists to do this tbh
@fbievan @franskeijer I didn't know/remember that DynDNS domains were badly regarded nowadays. My home sever was running on one for 10 years between 2003 and ~2012.
@mdione @franskeijer I dont even think they are, I just think organizations are going a bit loony with filters tbh.
@mdione @franskeijer funnily enough, it is blocked at my college for 'dynamic DNS', so it seems that for some reason these filters really hate that
@danielittlewood ufff, you should see how many typos didn't make it :)))
@danielittlewood (fixed, thanks :)
@mdione I am heathily curious UwU
@mdione replying cause an interested in said numbers if you intend to post them 
@sbrl yes I do.
@mdione I have GotoSocial, not Mastodon, though…
@mirabilos yeah, corrected to 'fediverse'. I have seen things I didn't even heard about.
@mdione
one curious human here - just saw the index.html (clicked a 2nd time to check if index page's name) (for your numbers'exactitude) `w;7[)
@mdione did you see the similar experiment I did a couple of years ago..
@keefmarshall not sure if I want to read it before I do my own take :)
What are you monitoring?
@futurebird just my home server, nothing nefarious :) But more seriously, I wanted a better description of the 'thundering herd'¹ that popular accounts sometimes complain about. Also, see how well the tools I have handle this for monitoring. I already know that Prometheus is not made for this.
¹ yes, I know I'm not using the term properly :)
@mdione @futurebird Ask @jwz to boost it 
@aerique @futurebird @jwz TBH I think he blocked me.
@mdione I'm curious, why isn't Prometheus suitable?
@viq short version: if you don't configure your log exporter correctly and you add the URL or referrer to the labels the cardinality of the metrics explode. Is not that it won't handle it, but your instance would have to be properly sized for this, and you might want to duplicate the metrics for querying them when you don't care about those labels.
Also, prom can't reveal other info like geoip or visit duration like other tools do.
@mdione@en.osm.town Misskey instances will have user agent containing "summaly bot" (i don't remember the exact string sorry)
@puppygirlhornypost2 yes, I saw it in the logs :)
@mdione
Clicked the link, but no boost... hey, I like mixing up the data for you <grin>
@mdione will there be some visualisation over classical music?
@jonn hmmm, I don't think I'll do that level of production. My writeups are pretty spartan, and lately I've even used automatic transcription of audios I dictate to my phone, which sounds more advanced, but it just saves me typing (but not editing) time.
@mdione Lack of SSL/TLS support might keep done systems from connecting.
@mikemccaffrey @mdione which is really silly let's be honest
@mikemccaffrey yeah, didn't think that through. I should have prepared better, but again, I didn't expect ... this :)
摏 @mdione If you wind up with a good list of user agents, please share. I'll send a DM to explain why.
@mdione Clicked on it out of curiosity. LOL
@mdione what is the experiment, exactly?
@mdione 
That's some tiny, tiny text...
@BlippyTheWonderSlug it should be your browser's default text, the page has 3 attrbibute-less tags, `html`, `body` and `p`. No CSS, no JS, nothing. Granted, it's not 100% up to spec, but should be handled just fine by browsers.
@mdione
Dunno how, then. I was under the assumption my default was 10pt Arial. I could be mistaken on that; technical German is hard.
TBF, if I tip the Handy to landscape mode, the point size does go up.
@mdione healthily curious human it is. Although, I did ignore a warning about an unsafe connection. :)
@JorisMeys right. This will make musing about real life traffic impact more difficult (having to add CPU usage and response time). I was lazy and didn't set up Let's Encrypt on the new domain, but honestly I didn't expect this to be this 'popular' (1.4k and counting).
@mdione will you be creating a graph for distribution over time or a tree of distribution?
@Djh1997 what's a tree distribution?
@mdione sorry meant to say tree diagram of the distribution path something like this but with the instance’s being the branching points
@mdione Always happy to help people with producing pretty graphs. 
@mdione bookmarking this and hope you’ll reply later with a write-up on what you learned.
@mborus many people has asked for this, so I'll just do an edit of the original toot and every body will get the update :)