Conversation

Notices

1. Embed this notice
   pomstan (pomstan@xn--p1abe3d.xn--80asehdb)'s status on Tuesday, 18-Mar-2025 02:36:10 JST pomstan

   • 

   https://digdeeper.neocities.org/articles/incognet.xhtml

   @mint opinion?

   •  and Phantasm like this.
   • Embed this notice
      (mint@ryona.agency)'s status on Tuesday, 18-Mar-2025 02:39:23 JST 

     in reply to
     @pomstan small indie company please understand
     I think photorec'ing VPS disks has been a thing for ages, could be solved by encrypting VM disks with some key and nuking it after cancelling the service. Either way the only thing my machine runs is wireguard and nginx with reverse proxy, so not much of a concern.
     Phantasm likes this.
   • Embed this notice
     pomstan (pomstan@xn--p1abe3d.xn--80asehdb)'s status on Tuesday, 18-Mar-2025 02:49:15 JST pomstan

     in reply to

     • 

     @mint

     photorec'ing VPS disks has been a thing for ages

     yeah I myself was surprised why I didn't think about this possibility ever

     though I still would expect decent providers to scrub volumes after deletion, it would be nice to see some sort of statistics

      likes this.
   • Embed this notice
      (mint@ryona.agency)'s status on Tuesday, 18-Mar-2025 02:50:49 JST 

     in reply to

     • Tadano
     @tadano @pomstan You aren't as long as you pay for it, if you do want to cancel it, you could boot into live environment and dd if=/dev/urandom a couple times to be safe.
   • Embed this notice
     Tadano (tadano@mt.watamelon.win)'s status on Tuesday, 18-Mar-2025 02:50:50 JST Tadano

     in reply to

     • 

     @mint @pomstan Most I have is a personal email server on an old Vultr VPS. How fucked am I?

   • Embed this notice
     djsumdog (djsumdog@djsumdog.com)'s status on Tuesday, 18-Mar-2025 03:07:34 JST djsumdog

     in reply to

     • 
     • Tadano

     I have one Incognet VM and I just use it as a proxy for torrent traffic via Wireguard. E-mail stays on a dedicated elsewhere.

     Since the article mentions a similar situation on BuyVM, I'm wondering if this is just a case of the hypervisor allocation block storage and not wiping it first. In that case, it's not a vulnerability in getting stuff from other peoples' VMs. It's a situation of reusing the same blocks for a different VM allocation.

     Anything you put on a cloud host (or even a dedicated host) can be fucked with. Even if you use full disk encryption and have to login via VNC to unlock it on boot, the host can always get your disk key out of memory. A dedicated sever can always monitor the remote KVM.

     I can see why a provider wouldn't zero out storage on allocation. That would be incredibly show. But they could zero old volumes x days after cancellation before returning them to the storage pool.I think the real takeaway here is zero out your VM disks before you cancel a service.