New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
You're not supposed to assign 1k IPv6 addressess to a single server.
Why not?
What you're doing is actually called NDP table exhaustion attack on the upstream router.
They may be detecting that via the flood of DAD requests (duplicate address detection) for newly added IPs.
Try assigning them not on eth0, but on some VPN interface at your server, or make a dummy interface (see modinfo/modprobe dummy) and add them there. I believe adding them to 'lo' could also work.
As to why /48, for running multiple VMs on multiple servers and assigning a /64 or more per VM.
They may be detecting that via the flood of DAD requests (duplicate address detection) for newly added IPs.
That makes sense. I'm not trying anything else for awhile since their support is probably irritated with what I'm doing. I'll just stick to a couple of IPv6's when my server comes back.
Edit: Looks like Cisco switches have NDP table <= 6500, which is around where I crashed. It'd be pretty bad if I crashed an entire rack.
So basically when you run that allocation script at 50 Dedi's or more you could possible kill one of the upstream routers?
Who was the guy with 50 Boxes and had no use for it?
I think it only takes 1, and it wasn't my intention to crash a layer 3 switch
I was just playing around with IPv6.
I don't think you crashed anything, they just detected the unusual activity and cut off your port. Also when the NDP table exhausts it will only lead to a slow-down for everyone and maybe some transient packet loss, but not a complete crash of the switch.
Why 6k IPv6 addresses ?
Just going to chime in and say I got one of these earlier. AutoSnipe was very kind and prompt in responding when I submitted a ticket, and we worked out a deal I was very happy with
CPU model : VIA Nano processor U2250 (1.6GHz Capable) Number of cores : 1 CPU frequency : 1595.806 MHz Total amount of ram : 1980 MB Total amount of swap : 975 MB System uptime : 8 min, Download speed from CacheFly: 56.7MB/s Download speed from Coloat, Atlanta GA: 6.17MB/s Download speed from Softlayer, Dallas, TX: 6.88MB/s Download speed from Linode, Tokyo, JP: 3.39MB/s Download speed from i3d.net, Rotterdam, NL: 35.9MB/s Download speed from Leaseweb, Haarlem, NL: 9.20MB/s Download speed from Softlayer, Singapore: 3.38MB/s Download speed from Softlayer, Seattle, WA: 5.92MB/s Download speed from Softlayer, San Jose, CA: 5.76MB/s Download speed from Softlayer, Washington, DC: 9.82MB/s I/O speed : 75.3 MB/sWhat kind of a deal, if you don't mind me asking? I was interested in it as a $3.00/month backup but paying 5Euro(according to the site) I'd rather go with a KS-1, better proc/more space but only 100Mbps.
@im_jmz - The deal was $10USD/q (8.82€) if you would like this offer just send in a ticket @ evobilling.com
@YellowHummingBird - Glad you're happy with the service
Regards,
Ryan
That's not a bad deal at all... Might be a customer soon
Their switches seem all to easy to crash.
I dont think he crashed, more like some alarm was triggered and they acted on it.
I bet the french techies were like Oh-la-la, 6k IPv6? Cet mec est fou!
I bet the french techies were like Oh-la-la, 6k IPv6? Ce mec est fou!
Perhaps. However this week I attempted to saturate my 2x kidis 1Gb/s ports and was told that these actions took out a number of other customers servers for a brief amount of time.
That is understandable. If one customer of ours uses more than 600 mbps for some hour or more than 800 mbps for any amount of time we consider it abuse.
That is normal in a shared environment, I am sure you did not expect a dedicated 1 gbps port.
kcaj is a permanent pain in the online.net irc channel... ;-)
That is normal in a shared environment, I am sure you did not expect a dedicated 1 gbps port.
I don't expect a 1Gb/s dedicated port at this price point but if I a 1Gb/s port is advertised I expect to be able to attempt to use it and if the capacity is there, not find myself causing problems.
If a customer using 1Gb/s is going to cause issues for your switches, don't give me a 1Gb/s port and limit me to whatever your infrastructure is realistically able to handle.
This was a one time occurrence!
A daily, one time occurence!
Haha you made my day!
CPU model : VIA Nano processor U2250 (1.6GHz Capable)
Number of cores : 1
CPU frequency : 800.000 MHz
Total amount of ram : 1970 MB
Total amount of swap : 975 MB
System uptime : 1 day, 14:34,
Download speed from CacheFly: 47.2MB/s
Download speed from Coloat, Atlanta GA: 6.71MB/s
Download speed from Softlayer, Dallas, TX: 6.67MB/s
Download speed from Linode, Tokyo, JP: 3.25MB/s
Download speed from i3d.net, Rotterdam, NL: 41.3MB/s
Download speed from Leaseweb, Haarlem, NL: 35.9MB/s
Download speed from Softlayer, Singapore: 2.86MB/s
Download speed from Softlayer, Seattle, WA: 5.67MB/s
Download speed from Softlayer, San Jose, CA: 5.04MB/s
Download speed from Softlayer, Washington, DC: 10.0MB/s
I/O speed : 79.0 MB/s
Is this normal on the 2€ dedi?
Looks fine - is anything there causing you concern?
On centos 500mB file downloaded in 20mins and sometimes the speed went to KB/s or even Bytes/second. I changed the Os to Ubuntu and got the above output.
What command was used to download the file?
Ah.. So that was you..
Not on the 26th.
@kcaj I had 3 of mine go down for ~15 mins, all at the same time. Thought I had my answer to why that happened...
Maybe you do, any 1 customer can saturate their port and the switch for other customers at anytime.
My server was up on the 26th last month.
I hope nobody really believes ALL those thousands of boxes are wired in the same 1 Gbps switch.