everydns.net go BOOM!
Dec. 1st, 2006 04:26 pmSo, I run a few hobbyist/non-profit domains, and I've been using the DNS services of EveryDns.net.
At 3:45 PM EST today, all 4 of everydns's nameservers stopped responding. All 4 servers are on seperate networks and in seperate physical locations.
This has "DDoS attack" written all over it.
The following domains are currently having interruptions in service:
- anthrocon.org
- claws-and-paws.com (including pafurry.claws-and-paws.com and lists.claws-and-paws.com)
- jbadger.org
- saveardmorecoalition.org
Not much we can do, other than to wait things out. Unless this DDoS lasts for more than 48 hours (typical time to propagate new NS records for domains), switching DNS providers probably won't help.
"What is DNS?"
DNS is the service that maps machine names to network addresses. That way, if you move a website onto a new machine, you simple need to update the DNS entry with the new address. Think of it like a phonebook.
"So the websites themselves are not affected?"
Correct. The websites and the machines they are running on are just fine. Though they aren't processing much traffic right now. :-P
"So, who is responsible for this?"
I don't know.
"How long will this continue?"
I don't know.
"How is this even possible?"
The main tactic for DDoS attacks these days seems to involved 0wned Windows boxes that are used as zombies. Thousands of 0wned boxes will be "commanded" at the same time to flood a specific host (or hosts in this case) with traffic. Since the traffic is coming from computers all over the world, it's non trivial to filter out, especially if it is disguisedto look like normal HTTP or DNS traffic.
Boxes typically get owned through spyware, viruses, and/or worms introduced through trojan horses.
"Are there any preventative measures we can take?"
Yes, if you own a Windows box, make sure it is kept up to date. Also, make sure that you are running both anti-spyware and anti-virus software. Having Internet access is a wonderful thing, but certain responsibilities come with that. If you don't know how to keep your machine up to date, find someone to help you out.
[Edit: Things are back to normal now.]
At 3:45 PM EST today, all 4 of everydns's nameservers stopped responding. All 4 servers are on seperate networks and in seperate physical locations.
This has "DDoS attack" written all over it.
The following domains are currently having interruptions in service:
- anthrocon.org
- claws-and-paws.com (including pafurry.claws-and-paws.com and lists.claws-and-paws.com)
- jbadger.org
- saveardmorecoalition.org
Not much we can do, other than to wait things out. Unless this DDoS lasts for more than 48 hours (typical time to propagate new NS records for domains), switching DNS providers probably won't help.
"What is DNS?"
DNS is the service that maps machine names to network addresses. That way, if you move a website onto a new machine, you simple need to update the DNS entry with the new address. Think of it like a phonebook.
"So the websites themselves are not affected?"
Correct. The websites and the machines they are running on are just fine. Though they aren't processing much traffic right now. :-P
"So, who is responsible for this?"
I don't know.
"How long will this continue?"
I don't know.
"How is this even possible?"
The main tactic for DDoS attacks these days seems to involved 0wned Windows boxes that are used as zombies. Thousands of 0wned boxes will be "commanded" at the same time to flood a specific host (or hosts in this case) with traffic. Since the traffic is coming from computers all over the world, it's non trivial to filter out, especially if it is disguisedto look like normal HTTP or DNS traffic.
Boxes typically get owned through spyware, viruses, and/or worms introduced through trojan horses.
"Are there any preventative measures we can take?"
Yes, if you own a Windows box, make sure it is kept up to date. Also, make sure that you are running both anti-spyware and anti-virus software. Having Internet access is a wonderful thing, but certain responsibilities come with that. If you don't know how to keep your machine up to date, find someone to help you out.
[Edit: Things are back to normal now.]
(no subject)
Date: 2006-12-01 09:35 pm (UTC)Would it be more effective to talk to their parents and otehr family members about turning of theirs? ;)
Or do we need to start targeting random corporations?
(no subject)
Date: 2006-12-01 09:37 pm (UTC)It's home users in general who tend to be the least computer saavy.
And turning off machines won't help. An infected machine can still participate in a DDoS while it is being used at the keyboard.
(no subject)
Date: 2006-12-01 09:59 pm (UTC)(no subject)
Date: 2006-12-01 10:03 pm (UTC)That was against Wall Street.
And everyDNS isn't part of the core infrastructure like say, the root servers. Chances are, someone specifically has a grudge against EveryDNS, or perhaps one of their customers.
(no subject)
Date: 2006-12-01 10:31 pm (UTC)(no subject)
Date: 2006-12-01 10:37 pm (UTC)Holy crap, you're right! (http://64.233.161.104/search?q=cache:M8nznQgcSBwJ:blog.opendns.com/category/everydns/+everydns+botnet&hl=en&gl=us&ct=clnk&cd=1)
(no subject)
Date: 2006-12-01 10:49 pm (UTC)(no subject)
Date: 2006-12-01 10:51 pm (UTC)Yes. If this keeps up when I get home tonight, I'm going to throw the switch for a temporary solution.
(no subject)
Date: 2006-12-01 10:56 pm (UTC)(no subject)
Date: 2006-12-01 11:01 pm (UTC)Thanks for the offer. I might look at that as a potential longer term solution.
Are you running DJB's DNS server by any chance? :-)
(no subject)
Date: 2006-12-02 12:04 am (UTC)I've not actually futzed with DJB's DNS... is it decent?
(no subject)
Date: 2006-12-02 12:08 am (UTC)I got away from BIND and Sendmail years ago because of all of the security issues that they were having. DBJ's qmail is pretty awsome, too. When I test a new build, I can just start the binary from the command line and initiate an SMTP conversation right there.
(no subject)
Date: 2006-12-02 12:32 am (UTC)(no subject)
Date: 2006-12-02 12:49 am (UTC)Most of my experience with Bind was Bind 8 (8.3.3, I think), and I only used Bind 9 for a few months before I ended up switching to UltraDNS, and eventually to EveryDNS.
(no subject)
Date: 2006-12-02 03:10 am (UTC)(no subject)
Date: 2006-12-02 07:09 am (UTC)Bind takes very little overhead for AC's traffic and could easily be slaved to EveryDNS's main server for the site.
Mussing about with the TTL or snapshotting an AXFR would give you the ability to live if this goes again.
Probably something handy to do for your own sites as well.
Oh. I'm more than willing to do redundant DNS for AC and your own sites to give it some geographical redundancy and health. My stuff is currently hosted in Toronto.
(no subject)
Date: 2006-12-02 08:09 am (UTC)(no subject)
Date: 2006-12-02 08:13 am (UTC)I did manage to find another DNS provider, though. I set up the records and I'm going to throw the switch when I get up in the morning.
(no subject)
Date: 2006-12-02 12:27 pm (UTC)(no subject)
Date: 2006-12-02 05:50 pm (UTC)(no subject)
Date: 2006-12-02 07:18 pm (UTC)I don't know yet.
I think EveryDNS is awesome, but from a sysadmin perspective, I have a responsibility to keep those sites up and running.
(no subject)
Date: 2006-12-03 07:41 am (UTC)Now it seems more machines are being attached. I can not seem to ping mail.claws-and-paws... at the moment via the last known IP address. 69.56......
5 packets transmitted, 0 received, 100% packet loss, time 3999ms
Ohh well time to sleep.
(no subject)
Date: 2006-12-03 09:17 pm (UTC)This appears to be something different.
It's actually a problem I've had before with where that machine is hosted. I opened up a trouble ticket with them.
(no subject)
Date: 2006-12-03 09:52 pm (UTC)On investigation, it appears that I borked a couple of DNS entires in my paniced moving over of them.
I'm fixing the problem now. With a TTL of 600, you should have access again soon.
(no subject)
Date: 2006-12-04 12:36 am (UTC)Thanks Giza!
(no subject)
Date: 2006-12-04 11:52 pm (UTC)(no subject)
Date: 2006-12-05 02:48 pm (UTC)but it just seemed to be a bump, back up after 2 hours.
(no subject)
Date: 2006-12-05 03:16 pm (UTC)I don't see any interruption in the machine's uptime. I'm guessing it was just a transient network issue.