VPU ip routing problem causing - 502 bad gateway message

[lucky1]

Hi admins,

it seems that there is a network routing error that causes a 502 - bad gateway message from time to time.

If I run a traceroute normally and the site is working it looks like this

  8   175 ms   176 ms   176 ms  ae-5.r23.lsanca07.us.bb.gin.ntt.net [129.250.3.189]
  9   176 ms   176 ms   176 ms  ae-2.r00.lsanca07.us.bb.gin.ntt.net [129.250.3.238]
 10   177 ms   177 ms   177 ms  ae-0.a00.lsanca07.us.bb.gin.ntt.net [129.250.2.104]
 11   178 ms   178 ms   178 ms  xe-0-0-14-3.a00.lsanca07.us.ce.gin.ntt.net [128.241.1.18]
 12   181 ms   178 ms   178 ms  192.228.109.129
 13   178 ms   186 ms   178 ms  192.228.109.50
 14   179 ms   178 ms   178 ms  dns.katustation.com [72.46.128.154]

but when the 502 error appears it looks like this

 9  ae-5.r23.lsanca07.us.bb.gin.ntt.net (129.250.3.189)  175.743 ms  176.180 ms  175.099 ms
10  ae-2.r01.lsanca07.us.bb.gin.ntt.net (129.250.4.107)  189.986 ms
    ae-2.r00.lsanca07.us.bb.gin.ntt.net (129.250.3.238)  173.314 ms
    ae-2.r01.lsanca07.us.bb.gin.ntt.net (129.250.4.107)  190.057 ms
11  ae-0.a02.lsanca07.us.bb.gin.ntt.net (129.250.2.186)  192.832 ms  193.218 ms
    ae-1.a02.lsanca07.us.bb.gin.ntt.net (129.250.3.234)  170.163 ms
12  xe-0-0-14-3.a00.lsanca07.us.ce.gin.ntt.net (128.241.1.18)  170.627 ms  166.672 ms  167.289 ms
13  192.228.109.129 (192.228.109.129)  168.568 ms  167.218 ms  163.835 ms
14  192.228.109.50 (192.228.109.50)  165.354 ms  168.872 ms  174.982 ms
15  * * *
16  * * *
17  * * *
18  * * *
19  * * *

which means that some kind of communication error between ip 192.228.109.50 and the vpu server with ip 72.46.128.154 seems to cause the problem.

Maybe it helps to find the root of the problem  if you forward that information to your provider. 

[Dazz]

Lucky thanks for this.  I just sent this over to the host.  Hopefully, they can fix this soon... driving me crazy.

[BeachBumTN]

Thanks Lucky1 and Dazz.    Dazz, hope you are well and thanks for keeping the site going for all of us even with the recent issues.

[Thalamus]

To me it seems to be that they are running a nginx proxy in front of the real site. Or I should maybe say site(s) - plural. I expect is that one of the hosts behind is having issues and the other one is not, and they haven't properly configured the nginx proxy to handle failed hosts. If the front would stop sending requests to the "failed" host, I expect this to resolve itself. Having more than one site is of course a good thing - but only if configured properly.

Why am I saying this ? I just tried to download 2 roms. First the page loaded, then I saw the two roms, I tried to open those in tabs, I got gw timeout on both tabs. reload and the page loaded. Pressed the rom and another gw error, then reload and I was given the file. So, it feels like a 50% hit and miss thing. I've seen similar things from my daily work.

Can't of course guarantee that this is what I'm seeing,