1
I host a personal website on an old desktop that is LAMP based. There are several strange things about this problem so I will break it down into steps.
Since I have a dynamic IP, I use no-ip to make sure I have a working domain name at all times. I use the automatic update client, but logged in and checked and my no-ip domain has the proper IP tied to it. Here is a link to the homepage through the no-ip domain for reference. Also, I do a ping and a traceroute on the no-ip domain and get:
[eckertzs@localhost ~]$ ping -c 1 endradil.noip.me
PING endradil.noip.me (65.24.215.99) 56(84) bytes of data.
64 bytes from endradil.noip.me (65.24.215.99): icmp_seq=1 ttl=64 time=2.23 ms
--- endradil.noip.me ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 104ms
rtt min/avg/max/mdev = 2.233/2.233/2.233/0.000 ms
[eckertzs@localhost ~]$ traceroute endradil.noip.me
traceroute to endradil.noip.me (65.24.215.99), 30 hops max, 60 byte packets
1 . (192.168.2.1) 1.755 ms 5.409 ms 5.380 ms
2 endradil.noip.me (65.24.215.99) 6.297 ms 9.543 ms 10.324 ms
Using this domain, I can connect to my webserver without issue or interruption(the https is required to avoid a redirect serverside, but it works).
I also have a domain I have bought on GoDaddy where I have a CNAME record forwarding the www subdomain to my no-ip domain.
CNAME Record
Host: www
Points to: endradil.noip.me
TTL: 1 hour
For the past several weeks, I never had an issue using the GoDaddy domain to connect (ssh or https). As of the past few days, however, the GoDaddy domain has only worked intermittently, for a few minutes at a time and then will go down for hours at a time. I get server not found errors most of the time. Also, if I happen to be using the GoDaddy domain for an ssh connection, the connection will freeze.
I have run online tests of the DNS and have seen that the website is visible by external servers and resolved to the correct IP. I also contacted GoDaddy support but they had no issues connecting to the website, and therefore did not see any issues. My personal computers (Windows desktop, linux laptop, android phone) all fail to connect when on my personal wifi. If I disconnect my phone from the wifi and use my AT&T wireless data, it can connect with both domains without issue. When I attempt to use Google webmaster tools to crawl the site using the GoDaddy domain, Google can not find the site.
From my linux laptop, I have found some interesting results when I ping or traceroute the domain. The results from these:
[eckertzs@localhost ~]$ ping -c 1 www.endradil.com
PING www.endradil.com.Belkin (198.105.244.228) 56(84) bytes of data.
--- www.endradil.com.Belkin ping statistics ---
1 packets transmitted, 0 received, 100% packet loss, time 10000ms
[eckertzs@localhost ~]$ traceroute www.endradil.com
traceroute to www.endradil.com (198.105.244.228), 30 hops max, 60 byte packets
1 . (192.168.2.1) 1.918 ms 2.806 ms 2.772 ms
2 cpe-65-24-208-1.insight.res.rr.com (65.24.208.1) 29.247 ms 29.654 ms 30.094 ms
3 cpe-69-23-24-117.new.res.rr.com (69.23.24.117) 15.597 ms 23.218 ms 23.581 ms
4 agg24.clmcohib01r.midwest.rr.com (65.29.1.52) 30.581 ms 30.556 ms 31.192 ms
5 be27.clevohek01r.midwest.rr.com (65.29.1.38) 30.580 ms 31.062 ms 31.038 ms
6 bu-ether25.atlngamq47w-bcr01.tbone.rr.com (107.14.19.38) 37.863 ms 68.844 ms 43.773 ms
7 107.14.17.178 (107.14.17.178) 51.866 ms 51.019 ms 50.989 ms
8 ae0.pr1.dca10.tbone.rr.com (107.14.17.200) 48.467 ms ae-4-0.a0.lax91.tbone.rr.com (66.109.1.113) 49.912 ms *
9 v413.core1.ash1.he.net (209.51.175.33) 60.270 ms 50.842 ms 50.819 ms
10 100ge5-1.core1.nyc4.he.net (184.105.223.166) 55.597 ms 56.045 ms 56.020 ms
11 xerocole-inc.10gigabitethernet12-4.core1.nyc4.he.net (216.66.41.242) 56.001 ms 55.969 ms 55.992 ms
12 * * *
both show the incorrect IP. Also, the traceroute timesout on hops 12 through 255 (output truncated above). The traceroute using site24x7 works and shows reasonable results when run from their california server. From another linux box on a different network but in the same city as me (10 miles away), I still get timeout for traceroute, however the IP resolves correctly for the domain.
From this I believe that the DNS result is incorrectly cached in either my router/modem or perhaps even at my ISP level. My question is, first, how do I find out exactly what is wrong, and second, how do I resolve it.
Godric Seer
Posted 2014-08-24T01:20:22.160
Reputation: 299
I suspect this a really a different issue and not a DNS issue. DNS is cached and stable. If you're reaching the machine initially then DNS is not going to cause a connection problem 10 minutes later. – Tyson – 2014-08-24T02:20:54.770
That's a possibility, but since the one domain works and the other doesn't, I'm not sure where else it could be breaking. Why would my server refuse to respond for certain domains, and not others? – Godric Seer – 2014-08-24T02:25:14.927
If you haven't already, I would purge the DNS cache on the effected machine (note: DNS cache, not the browser's cache/cookies). So by "the one domain" you're saying that the noip sub-domain works perfectly but another domain that is pointed via CNAME to the noip sub-domain breaks? Am I understanding correctly? – Tyson – 2014-08-24T02:38:49.380
Yes, and I am having the same issue on both my computers (linux laptop and windows desktop). Could my router be caching it and causing issues? – Godric Seer – 2014-08-24T02:40:43.433
Do you know what TTL is on both the domain and the no-ip sub-domain? technically they wouldn't need to be the same value, but perhaps the cache is expiring on the CNAME, the resulting query is too slow and the network decides the IP isn't valid anymore either. Only a theory, you have an interesting issue. – Tyson – 2014-08-24T02:52:16.217
what is also interesting is that it appears the htaccess rewrite to switch http requests to https appears to be working (or perhaps my browser is caching that). TTL on the troubled domain is 1 hour, I can't find what it is on no-ip, however my domain rarely changes. And clearing the DNS cache on my windows machine did not resolve it – Godric Seer – 2014-08-24T02:56:36.930
I give up, if TTL is an hour. that's longer than the 10 min it takes to fail. (unless the common TTL human error of setting 60 instead of 3600... 60 would make it expire once a minute and then we might still be onto something....) – Tyson – 2014-08-24T03:06:54.287
No, GoDaddy lists it as "1 hour", so thats not the case. I'm digging through the apache logs, and I don't see any of the requests from when I am using the problematic domain. I will keep hunting and hopefully learn enough to sort something out. – Godric Seer – 2014-08-24T03:09:38.950