#fail2ban

7 posts · Last used 26d

The #crawlers are pretty annoying, especially when looking at irrelevant stuff digging deeper than necessary. Fortunately, in #OpenBSD using the #fail2ban pattern can be applied as well. My #GoT web server got hammered and first I just banned all the found crawler names. However, now, every action they do on gotwebd is remembered with an effort number and if that adds up to 100, the IP is banned. Additonally, connection storms are also banned if they follow certain patterns. #pf, #perl and I am done using only on-board tools. Why? I didn't want to install #anubis. Not because I don't like it, it is just because I like to do the minimum.
1
0
2
0
troduction: I'm Arber — 17 years running infrastructure that can't afford to be down (ISP backbones, national broadcasters, government systems), based in Tirana, Albania. For the past year I watched fail2ban block tens of thousands of attacks… and nothing ever happened to the attacker. Almost every source is someone's hijacked VPS, and the provider never hears about it. That silence started to feel like the real vulnerability. So I built something about it. RIPOSTE turns every firewall ban into an automated, evidence-backed abuse report: → fail2ban / QNAP QuFirewall push events to a local FastAPI collector (no log scraping) → attribution via Abusix abuse-contact DNS with RDAP fallback, cached per CIDR → one aggregated X-ARF report per responsible network per day, full timestamped evidence → delivery tracked end-to-end: sent / bounced / acked / human reply / takedown Month one on my own infra: 2,847 blocked attacks → 214 responsible networks → 31 reports → 4 compromised hosts confirmed offline. The per-provider spread is the interesting part — some suspend within 48h, some abuse mailboxes literally hard-bounce. A public per-provider accountability dashboard is next. No hack-back — it never sends a packet at the attacker. Only professional reports to registered abuse contacts. Docs & architecture: https://github.com/arberormeni2022/riposte Beta access for operators: https://buymeacoffee.com/securitysystem Which firewall should get an adapter next? Happy to talk WHOIS/RDAP swamp-draining. #infosec #fail2ban #selfhosted #sysadmin #abusedesktroduction: I'm Arber — 17 years running infrastructure that can't afford to be down (ISP backbones, national broadcasters, government systems), based in Tirana, Albania. For the past year I watched fail2ban block tens of thousands of attacks… and nothing ever happened to the attacker. Almost every source is someone's hijacked VPS, and the provider never hears about it. That silence started to feel like the real vulnerability. So I built something about it. RIPOSTE turns every firewall ban into an automated, evidence-backed abuse report: → fail2ban / QNAP QuFirewall push events to a local FastAPI collector (no log scraping) → attribution via Abusix abuse-contact DNS with RDAP fallback, cached per CIDR → one aggregated X-ARF report per responsible network per day, full timestamped evidence → delivery tracked end-to-end: sent / bounced / acked / human reply / takedown Month one on my own infra: 2,847 blocked attacks → 214 responsible networks → 31 reports → 4 compromised hosts confirmed offline. The per-provider spread is the interesting part — some suspend within 48h, some abuse mailboxes literally hard-bounce. A public per-provider accountability dashboard is next. No hack-back — it never sends a packet at the attacker. Only professional reports to registered abuse contacts. Docs & architecture: https://github.com/arberormeni2022/riposte Beta access for operators: https://buymeacoffee.com/securitysystem Which firewall should get an adapter next? Happy to talk WHOIS/RDAP swamp-draining. #infosec #fail2ban #selfhosted #sysadmin #abusedesktroduction: I'm Arber — 17 years running infrastructure that can't afford to be down (ISP backbones, national broadcasters, government systems), based in Tirana, Albania. For the past year I watched fail2ban block tens of thousands of attacks… and nothing ever happened to the attacker. Almost every source is someone's hijacked VPS, and the provider never hears about it. That silence started to feel like the real vulnerability. So I built something about it. RIPOSTE turns every firewall ban into an automated, evidence-backed abuse report: → fail2ban / QNAP QuFirewall push events to a local FastAPI collector (no log scraping) → attribution via Abusix abuse-contact DNS with RDAP fallback, cached per CIDR → one aggregated X-ARF report per responsible network per day, full timestamped evidence → delivery tracked end-to-end: sent / bounced / acked / human reply / takedown Month one on my own infra: 2,847 blocked attacks → 214 responsible networks → 31 reports → 4 compromised hosts confirmed offline. The per-provider spread is the interesting part — some suspend within 48h, some abuse mailboxes literally hard-bounce. A public per-provider accountability dashboard is next. No hack-back — it never sends a packet at the attacker. Only professional reports to registered abuse contacts. Docs & architecture: https://github.com/arberormeni2022/riposte Beta access for operators: https://buymeacoffee.com/securitysystem Which firewall should get an adapter next? Happy to talk WHOIS/RDAP swamp-draining. #infosec #fail2ban #selfhosted #sysadmin #abusedeskHi, #introduction: I'm Arber — 17 years running infrastructure that can't afford to be down (ISP backbones, national broadcasters, government systems), based in Tirana, Albania. For the past year I watched fail2ban block tens of thousands of attacks… and nothing ever happened to the attacker. Almost every source is someone's hijacked VPS, and the provider never hears about it. That silence started to feel like the real vulnerability. So I built something about it. RIPOSTE turns every firewall ban into an automated, evidence-backed abuse report: → fail2ban / QNAP QuFirewall push events to a local FastAPI collector (no log scraping) → attribution via Abusix abuse-contact DNS with RDAP fallback, cached per CIDR → one aggregated X-ARF report per responsible network per day, full timestamped evidence → delivery tracked end-to-end: sent / bounced / acked / human reply / takedown Month one on my own infra: 2,847 blocked attacks → 214 responsible networks → 31 reports → 4 compromised hosts confirmed offline. The per-provider spread is the interesting part — some suspend within 48h, some abuse mailboxes literally hard-bounce. A public per-provider accountability dashboard is next. No hack-back — it never sends a packet at the attacker. Only professional reports to registered abuse contacts. Docs & architecture: https://github.com/arberormeni2022/riposte Beta access for operators: https://buymeacoffee.com/securitysystem Which firewall should get an adapter next? Happy to talk WHOIS/RDAP swamp-draining. #infosec #fail2ban #selfhosted #sysadmin #abusedesk
0
0
1
0
Not sure why, but with the LAN I've needed to whitelist all hosts using #Jellyfin clients, otherwise their IP gets blocked by Fail2Ban after about an hour of watching. I fear this would happen to remote viewers too, but not sure I've ever watched more than an hour remotely. No idea if any other remote client services do the same, I've not let them run more than a few minutes. Is this perhaps a #debian #ufw #Fail2Ban setting I should know about?
5
3
0
0
So I tried using #fail2ban to ban AI scrapers that are hammering my #Forgejo instance. I currently have 130k IPs in nftables and there's no end in sight. These are from all over, not from any specific ranges. It took fail2ban several hours to re-add the bans after restarting it for a configuration change. It's 21:45 now and it's currently processing hits from 09:11. I don't know if it will even catch the backlog. I suppose it's the wrong tool for this but I wanted to try. #SelfHosting
0
1
0
0
I always remap my sshd daemon to listen to a non-standard port, to reduce a lot of noise. Which has worked fine for years. But every now and then there are attempts. All the #Linux kernel flaws found lately has made remote login attempts more interesting for attackers. And they scan much more broadly now than just port 22. And that's why my second line of defence is to disallow remote root login - and also make use of the AllowGroups feature in sshd_config. Users granted remote access must be member of a specific group. And root is also excluded from this group. That pays off these days. And this is a nice filter match for #fail2ban and similar tools https://termbin.com/0cf6 I have 293 login attempts on "random users" since May 21. And 259 attempts as root. #infosec #ssh #sshd #systemhardening #kernel
6
4
2
0
You've seen all posts