1

For the last week we've been struggling to get connected from inside our network to a VPS where we host some files for users of our product. It's up for everyone else, down for us and not blocked by our firewall. We can ping IP's above it and blow it (not owned by us) but tracert give us this

[root@webserver ~]# tracepath 205.178.186.176
 1:  10.10.10.85 (10.10.10.85)                              0.086ms pmtu 1500
 1:  24-158-1-81.static.mdsn.wi.charter.com (24.158.1.81)   2.732ms
 1:  24-158-1-81.static.mdsn.wi.charter.com (24.158.1.81)   1.782ms
 2:  10.137.194.1 (10.137.194.1)                            7.067ms
 3:  dtr01mazowi-tge-0-7-0-7.mazo.wi.charter.com (96.34.22.50)   9.244ms
 4:  crr01euclwi-tge-0-5-0-4.eucl.wi.charter.com (96.34.17.38)  20.057ms
 5:  bbr01euclwi-bue-4.eucl.wi.charter.com (96.34.2.4)     20.499ms
 6:  bbr02euclwi-bue-5.eucl.wi.charter.com (96.34.0.7)     43.499ms
 7:  bbr01chcgil-bue-1.chcg.il.charter.com (96.34.0.9)     27.379ms
 8:  prr01chcgil-bue-2.chcg.il.charter.com (96.34.3.9)     22.129ms
 9:  eqix.chcg.twtelecom.NET (206.223.119.36)              22.177ms asymm 10
10:  atl1-ar3-ge-1-1-0-0.us.twtelecom.net (66.192.241.138)  51.657ms asymm 16
11:  66-193-226-118.static.twtelecom.net (66.193.226.118)  43.431ms
12:  209.17.112.34 (209.17.112.34)                         47.507ms
13:  209.17.114.78 (209.17.114.78)                         43.511ms
14:  no reply
15:  no reply
16:  no reply
17:  no reply

and so on till it times out. It's always been up when I check downforeveryoneorjustme, it works on my phone and from my home (a few of our customers may reported an outage, although I haven't confirmed it).

In between hop 11 and 12, there is a long pause. I've heard that Time Warner has been blocking somehow, and 209.17.114.78 is actually at network solutions (or web.com) who hosts our VPS, we're not even Time Warner subscribers, and we tried connecting using our backup ISP (not Charter) and that didn't work either, which is strange. I've called Network Solutions support a few times and I actually agree with them that this is not their problem (even though it happened right after an automated update parallels update).

So, the question is, what could make traceroute fail like this, and what could gum up the process seemingly just for us.

Here's another traceroute (and you could do your own) from a co-worker outside our network 

Tracing route to 205.178.186.176
over a maximum of 30 hops:

  1    <1 ms    <1 ms    <1 ms  192.168.0.10
  2     8 ms     8 ms     6 ms  10.135.192.1
  3     9 ms    14 ms     7 ms  dtr01ftbgwi-tge-0-6-0-3.ftbg.wi.charter.com [96.34.25.58]
  4    26 ms    11 ms    11 ms  crr01ftbgwi-bue-4.ftbg.wi.charter.com [96.34.18.108]
  5    22 ms    23 ms    23 ms  crr01euclwi-bue-1.eucl.wi.charter.com [96.34.16.77]
  6    19 ms    15 ms    15 ms  bbr01euclwi-bue-4.eucl.wi.charter.com [96.34.2.4]
  7    20 ms    15 ms    16 ms  bbr02euclwi-bue-5.eucl.wi.charter.com [96.34.0.7]
  8    23 ms    31 ms    23 ms  bbr01chcgil-bue-1.chcg.il.charter.com [96.34.0.9]
  9    23 ms    21 ms    21 ms  prr01chcgil-bue-2.chcg.il.charter.com [96.34.3.9]
 10    69 ms    20 ms    21 ms  eqix.chcg.twtelecom.net [206.223.119.36]
 11    53 ms    54 ms    54 ms  atl1-ar3-ge-1-0-0-0.us.twtelecom.net [66.192.246.238]
 12    49 ms    49 ms    47 ms  66-193-226-118.static.twtelecom.net [66.193.226.118]
 13    52 ms    49 ms    57 ms  209.17.112.38
 14    49 ms    61 ms    47 ms  209.17.114.78
 15     *        *        *     Request timed out.
 16    50 ms    49 ms    48 ms  176.186.178.205.netsolvps.com [205.178.186.176]

It also worked when using a proxy. Also, when I do SSH into my VPS, I cannot connect back to my office (where the connections need to be made to authenticate between the two servers)

Peter Turner
  • 2,178
  • 9
  • 33
  • 45
  • can you SSH to the VPS (from another location) and check iptables `sudo iptables -L`? Do you have any other firewall or security at the VPS location? Who is hosting the VPS, did you ask them about it? – user16081-JoeT Jan 29 '14 at 00:05
  • do you know what "reject-with icmp-host-prohibited" means? – Peter Turner Jan 29 '14 at 03:53
  • (not that it has anything to do with this, I was running `iptables -L` on the wrong server. Our VPS is with Network Solutions and doesn't have anything in iptables (the firewall is configured with Parallels and didn't look like it had anything in it either) – Peter Turner Jan 29 '14 at 03:58
  • yes based on the routing issue mentioned below, iptables has nothing to do with this - I would go through whatever channels to have your ISP(s) escalate the issue. – user16081-JoeT Jan 29 '14 at 04:10

1 Answers1

1

With mtr check what kind of packet loss you are getting to your VPS, if any. I'm getting a lot of loss at 209.17.114.78, so it could very well be a configuration there is dropping your packets.

c4urself
  • 5,530
  • 3
  • 28
  • 39
  • So, that's off in network solutions land, I did notice that there are two hops in their subnet, if it goes from .38 to .78 it's busted, but if it goes from .34 to .78 it works. There's no way to fix this internally, is there? – Peter Turner Jan 29 '14 at 03:50
  • @Peter I think you have to try to work through the ISP, tell your VSP host to escalate the issue to their routing provider – user16081-JoeT Jan 29 '14 at 04:07