can't say why. I just see delivery errors in logs. maybe some network problems, maybe some protocol inconsistencies between platrorms/servers, maybe connectivity data is wrong in servers' public interfaces like webfinger, or something. just the fact.
at the moment, I have contacts that followed me on these servers and I'm pretty sure they can't read this post. as well as I cannot reach the admins of these servers because I get HTTP errors and/or other weird answers from their servers. so I just make it public, pin the post and just leave it here. if you can see this post and see your server in the list, please check the connectivity to my server or ask your admins. if they blocked my address or something I want to know because I need to block them in turn to stop this strange and bothersome undelivered posts resend attempts on my side.
if you're an administrator of these servers and want to solve connectivity problems - I'm ready to get into details and see what's going on from logs on my side, etc. I'm programmer and I always stand for normal server interconnection.
some servers are banned on my side. I do this just in case a server breaks the network exchange rules, has spam/malware/porn content or has banned my server by some I-don't-really-care reasons to prevent the posts resend cycles. I'm not aggressive dork but roaches in somebody's heads are not my problem. I may dislike some users and I may block them but usually this does not come to blocking whole servers. blocking servers is stupid. but when I get blocked I have no other choice but block them in my turn because futile delivery attempts waste network traffic and CPU on my server. I don't see reasons for this.
some IPs may get banned on my side: if I see problems from some network address like ddos-like messages attacks, scraping, script bots, malicious behavior, attempts to break protocols, etc. I block such things without a special notes. I think Fediverse is not a place for spam and malicious attacks.
I have a problem to report. I take it, server name epsilon-ix.masto.host is located on your hosting.
this server spreads malicious ActivityPub attacking malware: its peers list contains 100886 faked domain records of *.activitypub-troll.cf format (they all are NXDOMAIN).
@mastohost@strizhechenko I tried to think out some euristic function for checking peers list but for a while could not come up with any good solution. I just added check for blacklisted patterns in peers discovery. but many other platforms/servers don't do this and seriously suffer from this malicious peers lists.
@mastohost@strizhechenko maybe something like this could work. but peers lists are not updated instantly and 10 is an abstract number. for servers with thousands of peers this may be not good solution. maybe the percentage could be used: nonexisting domains to total records number. something of the kind. I will try some when I have more spare time. I have to do my work. PHP is definitely not my work, although I added some extra checks to avoid extra load on exploring and/or processing of the blocked or nonexistent domains.
@VGM inspiration is an abstract term, not having juridical definition. so lawers cannot operate with such terms. and exploitation is getting some material profit from something. without material profit even AI is not °criminal°.
@phend why do they consider that a model was trained on copyrighted data? the most models in research are actually trained on publicly accessible data that are barely copyrighted.
@phend it's like to say that any person that has seen a copyrighted image can draw the same picture. but this is not a digital copy of any kind, by definition.
@lxo you got me wrong. it's not just tools. it's centralization. one cannot build any search without it. and this is something that is unacceptable for such networks. as well as centralized blocklists that are peddled recently over here. centralization means surveillance and control, whatever they might claim. thinking one step ahead is not enough. we should consider the global consequences. I know very well what security is. and I never did anything that could deliver troubles to people. programmers (and any other specialists) should care what they do. we can write anything. that does not mean we should write anything without thinking.
well, the good news is it will never work out. it's too bloated. any search engines are bloated and resource consuming. there's no variants. so this is a kind of guarantee that it won't spread over decentralized networks. bloated data means coprorations invasion and centralization, their commercial interest and big data. it won't work here, thank goodness. but this does not mean people may be shallow and don't think what they do. wrong ideas lead to bad consequences. we should be careful.
@r000t I know what I post and where, but the less tools fascists have - the better it is. don't think they're smart. far from that. they use plain search tools like google, etc. and then this may lead to tortures and murders of people for any words they might find. so I think decentralized networks must stay decentralized.
@r000t I also think that search is a good but dangerous feature. in this country people get in prison for 10 years for a simgle post about war. and we should not give them the tools for repressions. alas, search that is intended for information lookup is de facto used for opressing governments, tyrannies and different crooks. so I closed the web access and block spiders and unknown crawlers on my server. this is weird but it works in current situation.
@sophia_sardegna it cannot be "censored" because users can choose any servers and run their own instances. and Mastodon is just a little part of a big Fediverse community which at the moment contains not less than 15000 servers (my count may be not exact, the real number may be more).
@d_pechkin@erabo and I emphasize that I write in C all these years. so the code clarity and code style is important and should not allow things like 'int* x'. because they're principally wrong and may lead to errors if the code is changed later. and it means that people don't understand the C typing and type assotiation rules. and then they cannot understand a little more complicated things like 'int *(*f)(int *x);' and so on. because they have a chaos in their head and mistake common declarations and evaluations. and note that (*f) means a pointer to function, because pointer is attached to the defined object, not to the type it dereferences. the example of proper code style is the Kernel coding style: clear and fine. I think that coding style is important.
@d_pechkin@erabo well, if you dealt with assembler and binary machine codes you can clearly understand the thing with addressing and the difference between direct and addressed access to data, and that there's no really 'int*' or 'float*' types but intXX_t that is defined by the address bus. and on modern PC machines I can use int64_t type for addresses. C is flexible enough to allow this without problem (and this is why I like it). pointers are just a convenience.
I'm not a teacher kind of person. I always easier dealt with hardware and systems than with people :) but what I know for sure is C is plain and simple. I learnt it when I was 12 and I barely knew English then. I had only a K&R book and text file with description of standard functions for borland C. and it wasn't that difficult. now people say 'it is unsafe and error prone' but this is something wrong with their head, not with C :) C is there since 1972. and I'm 5 years younger than it. it always was plain and simple. and it still is.
I'm a professional C/C++ programmer, I use Linux and deal mainly with system programming, back-ends, networks, high-load data processing. I fond of music, play all kinds of basses, cello, learn to play tenor-sax and take lessons of academic vocals (wide ranged big voice). I've got reptile pets. I fluently speak Russian and English, learn Italian and know some other languages a little.