Skip to content
Register

Contabo killed my NAT VPS network with rsync, now SSH hangs

VPS Hosting by SamAlvi 14 replies 8.3K views
SamAlvi

Member OP

Self-Host Everything

#1

I was doing what any sane person does and rsyncing my 400GB photo collection off my Contabo NAT VPS to my home NAS. Why pay for cloud backup when you can self-host it with a simple docker compose stack and a reverse proxy. About 20 minutes in the transfer stalled. Now SSH just hangs. The control panel shows the box as "running" but I cannot get a connection back. I had no rate limits set because I wanted it done fast. Has anyone seen a provider kill your ethernet entirely from sustained outbound? I am on their 512MB plan if that matters. No reply from their ticket yet after 6 hours.

my cloud. my rules. my 3AM alerts.
lucgone

Member

Le Baguette

#2

Yikes bro u pwned ur own box gg

Noob move saturating the pipe on a nat vps gg Contabo probably has some janky fw rule that auto drops u when u hit the limit

Vive la résistance... électrique
declan6

Member

Ex-Telco

#3

This is why I run my own ethernet at home and only use someone else's computer when I have to. Back in the day we had a T-1 and you learned real fast what happens when you flood it. Your provider's ethernet port is probably not dead, their upstream just null-routed you for abuse. Check if you have a different v4 address on the control panel at https://my.contabo.com, they might have moved you. Rate limit your rsync next time. --bwlimit=10000 is your friend.

sam3

Member

#4

I have seen this exact behavior. It is almost always a bandwidth trigger on shared upstream. The provider does not kill the interface. They shape or drop your traffic at their edge router. You are still sending packets into the void. SSH cannot complete the handshake. The solution is always rate-limiting and running transfers overnight. I run my rsyncs through a reverse proxy tunnel with hard caps; -- Sam

DA

Member

#5

Peering

  • BGP blackhole probably. V4 nullroute.
  • Fw dropped your prefix. Tier-1 upstream dont play.
  • Cfg bwlimit. Cron for 3am. Transit.
lucgone

Member

Le Baguette

#6

Lol this is why I just run robocopy on my IIS box with /IPG inter-packet gap, never had this issue with Active Directory managing the schedule. Linux networking is just fragile I guess Could also set up a scheduled task instead of cron, much more reliable. IIS enjoyer out.

Vive la résistance... électrique
minh1987

Member

#7

What bwlimit did u actually try

phở at 3AM, deploy at 4
SamAlvi

Member OP

Self-Host Everything

#8

To answer minh1987: I never tried any bwlimit, that's the whole problem. I just let it rip.

Still locked out ten days later by the way. Control panel still says running but every port is black-holed. Opened a ticket with Contabo, no response yet. Starting to think the instance is just toast and I should cut my losses. Did anyone else ever get actually un-null-routed by them or is this a permanent "go away" signal

my cloud. my rules. my 3AM alerts.
declan6

Member

Ex-Telco

#9

Minh1987 I didn't set one at all on that transfer, learned the hard way. Was more warning the OP than describing my own setup.

SamAlvi said:
Has anyone seen a provider kill your ethernet entirely from sustained outbound

Did you ever get back into that box or did you have to reprovision? Curious if Contabo auto-restores after a cooldown or if it's a ticket job.

ZH

Member

#10

Did you ever get back into this box or did Contabo have to reimage it

I had something similar on a OVHcloud NAT plan years ago and their panel just showed "running" forever while the network was blackholed, had to open a ticket to get them to flip it back

Post a reply

You need an account to reply. Log in or register to join the conversation.

Post reply Preview Save draft