The "OSM Standard tile layer" looks wrong (white lines, abusive comments etc.)

Yes, that is why I try to convince everyone that just instructing to do a hard reload won’t help if you are not on www.osm.org.

(We don’t know the whole VSL ruleset the OWG uses at Fastly, that is fine).

And also: if the VSL rely on the cache headers of the tile render servers, your issue at mod_tile is spot on.

Indeed - I’m loosely quoting what DWG correspondents are saying. As numerous comments above have said, it’s complicated. Generally speaking, people using a web browser to see OSM map tiles don’t fully understand the delivery mechanism that allows them to see those**, and a comment above suggests nor can anyone else - even those who do understand how normal web server access works!

** to them there is just “their computer” and “sites on the internet”. If something is wrong on a site on the internet, clearly that site needs to just “fix the data on their site” and all will be well. :slight_smile:

Maybe the OWG could remove that restriction for seven days and until mod_tile#441 is fixed?

1 Like

This might be dragging us off topic, but to me it feels like we need some sort of “deeper” cache. So we can have some way of marking tiles “to be burned” and if a request comes in for them we serve day old tiles (rather than the latest ones) if the server is too busy to render fresh ones.

3 Likes

Is this vandalism and cache management related to the fact that osm is not accessible half the time since yesterday (June 14-15)?

if you mean the website www.openstreetmap.org: that one had a DDoS attack as well. (Can’t tell if this was because of ransom, asininity or an api abuser or related to the vandalism act).

1 Like

Hi,

In this moment at 20.06 UTC is a new big attack.

1 Like

Was there any discussion about restrictions on rendering long roads? Why try to render segments of roads half the length of the planet?

1 Like

new attack by KYtyQuaIihonsT | OpenStreetMap, looks to be reverted now

i do think the best solution here would be to start serving tiles from a different date to external servers (without an obvious vandal attack), only showing the latest tiles on OSM.org etc. this way the vandalism isn’t shown on loads of sites that use osm tiles, which is the main reason imo that they’re targetting OSM

3 Likes

I see a couple of accounts scrambling the contents of relations now (1, 2). Both already blocked by DWG.

2 Likes

I’ve never seen it to be this bad (though I’m also looking at farther out than before):

Screenshot

These already have been fixed but jeez, it becomes worse and worse…

Small edit: Actually, this is that bad recently and the ways weren’t indeed that bad before.

What is going on with this industrial scale vandalism? Seems like OSM needs to up its game against cyberattack! :confused:

2 Likes

right now its looking bad in Algeria, a huge area is now covered in roads
p

new vandalism: Changeset: 152736657 | OpenStreetMap

Now Hirschaid in Bavaria pretty much looks like what Manuel posted about Rüsselsheim. Dozens, maybe a hundred lanes, half of them with English street names, the other half with Ukrainian anti-Russian slogans (I sympathize a lot with the Ukrainian cause but this is definitely not the right place for this).

Make that Hirschau, not Hirschaid. Relation: ‪Hirschau‬ (‪1028538‬) | OpenStreetMap

It is exactly this. The amplification is the reason the bad actor feels gratification for his vandalizing doings. If that gratification would fail, motivation would be gone. But this means: do not serve these tiles with the vandalism. Stop being the megaphone for the vandalizing bad actor.

9 Likes

Are you sure the bad actor is a human? Surely attacks on this scale are automated.

Yes, but the systems are run by a human :wink:

[We are not in the age yet where an AI decides to write the code to do this :wink: ]

5 Likes