Site not found errors

[Uzziah]

Are you guys or the host site having bandwith troubles? Several times a day, for several minutes at time, I can't get to the site. I most commonly use Firefox with a cable modem behind a secured router and have no other problems with other sites. I also work in IT so I've confirmed things at this end.

Just curious if it is something you're aware of and working on. Or???

Thanks,
Greg

[SE84]

http://www.ducati.ms/forums/showthread.php?t=70949

[scott]

I will remind the head honchos pronto. Sorry everyone.

[Uzziah]

Originally Posted by SE84

http://www.ducati.ms/forums/showthread.php?t=70949

I had read that thread but since I joined after the majority of the discussion, I figured the "new servers" weren't the issue, nor were stored cookies or DNS history.

When it works, it does indeed fly. But there's no rhyme or reason for when it does not. Sometimes it's in the middle of a post, sometimes it's just clicking on one...nothing to which I can tie it down. (I've taken to copying my posts before I click the Submit button so I don't lose any more).

(Actually I came back to edit this right after posting and got it again for about three minutes.)

Thanks Scott, hope it's resolved easily.

[SE84]

Originally Posted by Uzziah

I had read that thread but since I joined after the majority of the discussion, I figured the "new servers" weren't the issue, nor were stored cookies or DNS history.

When it works, it does indeed fly. But there's no rhyme or reason for when it does not. Sometimes it's in the middle of a post, sometimes it's just clicking on one...nothing to which I can tie it down. (I've taken to copying my posts before I click the Submit button so I don't lose any more).

(Actually I came back to edit this right after posting and got it again for about three minutes.)

Thanks Scott, hope it's resolved easily.

It's DNS, Verticalscope has the TTL set to 300 seconds, so the cache times out quickly, and your edits (especially if they take longer than 5 minutes) will require a DNS lookup to post, but may fail if the server doesn't respond.

[Uzziah]

Originally Posted by SE84

It's DNS, Verticalscope has the TTL set to 300 seconds, so the cache times out quickly, and your edits (if they take longer than 5 minutes) will require a DNS lookup to post, but may fail if the server doesn't respond.

I realize you're trying to help, but this isn't my first rodeo. I've been on forums for several years, different formats, and I'm a network admin.

Just to clarify: I wasn't writing the edit, rather I read the post, realized I should add a bit and clicked the edit button...and it hung. There was maybe a 20-second span between the page load for the post and the time I clicked the edit button. Three minutes later or thereabouts, I was able to get back to it.

There is nothing normal, typical or predictable about it, and nothing I can see that would allow me to replicate it at will.

[arai_speed]

Originally Posted by Uzziah

Are you guys or the host site having bandwith troubles? Several times a day, for several minutes at time, I can't get to the site. I most commonly use Firefox with a cable modem behind a secured router and have no other problems with other sites. I also work in IT so I've confirmed things at this end.

Just curious if it is something you're aware of and working on. Or???

Thanks,
Greg

I get the same error every day, multiple times a day. Sucks.

[SE84]

Originally Posted by Uzziah

I realize you're trying to help, but this isn't my first rodeo. I've been on forums for several years, different formats, and I'm a network admin.

Just to clarify: I wasn't writing the edit, rather I read the post, realized I should add a bit and clicked the edit button...and it hung. There was maybe a 20-second span between the page load for the post and the time I clicked the edit button. Three minutes later or thereabouts, I was able to get back to it.

There is nothing normal, typical or predictable about it, and nothing I can see that would allow me to replicate it at will.

Well, if your DNS cache on your ISP's server had fetched the IP and the timeout for the cache was right before you clicked 'edit,' then you would need another lookup and if it timed out...well.

If you add the site to your local hosts file all the problems go away. But, be my guest...I'll just take my twenty years of Unix and Network experience and be over in the corner.

Cheers

[SE84]

Scott, here's a little more useful data that you can take back to the V/S folks.

I have a small program written in C that crafts a DNS packet containing the lookup request for www.ducati.ms:

#include <stdio.h>
main(  )
{
FILE *fp;

fp=fopen("lookup", "w");
fprintf(fp, "%c%c%c%c", 0x00, 0x01, 0x01, 0x00);
fprintf(fp, "%c%c%c%c", 0x00, 0x01, 0x00, 0x00);
fprintf(fp, "%c%c%c%c", 0x00, 0x00, 0x00, 0x00);
fprintf(fp, "%c%s", 0x03, "www");
fprintf(fp, "%c%s", 0x06, "ducati");
fprintf(fp, "%c%s%c", 0x02, "ms", 0x00);
fprintf(fp, "%c%c%c%c", 0x00, 0x01, 0x00, 0x01);
fclose(fp);
}

This generates a file called "lookup" that contains a valid DNS request.

Next, hping is a useful tool that can ping alternate ports and protocols. An interesting feature of hping is that it can read protocol data from a file. So, I used hping to "ping" the DNS server with the crafted packet. This isn't sending ICMP data, but actual DNS traffic. Normal ICMP ping shows no differences between the three servers and their response is each 100%. But, when we look at DNS traffic:

The results speak for themselves:

To ns1.verticalscope.com:

[~]$ sudo /opt/local/sbin/hping3 -2 -p 53 -E lookup -d 31 64.34.231.90
HPING 64.34.231.90 (en0 64.34.231.90): udp mode set, 28 headers + 31 data bytes
len=208 ip=64.34.231.90 ttl=48 id=60863 seq=0 rtt=90.4 ms
len=208 ip=64.34.231.90 ttl=48 id=60882 seq=1 rtt=102.0 ms
len=208 ip=64.34.231.90 ttl=48 id=60915 seq=2 rtt=90.6 ms
len=208 ip=64.34.231.90 ttl=48 id=60963 seq=3 rtt=92.3 ms
len=208 ip=64.34.231.90 ttl=48 id=61000 seq=4 rtt=94.3 ms
len=208 ip=64.34.231.90 ttl=48 id=61038 seq=5 rtt=107.2 ms
len=208 ip=64.34.231.90 ttl=48 id=61082 seq=6 rtt=110.9 ms
len=208 ip=64.34.231.90 ttl=48 id=61100 seq=7 rtt=91.6 ms
len=208 ip=64.34.231.90 ttl=48 id=61128 seq=8 rtt=92.4 ms
len=208 ip=64.34.231.90 ttl=48 id=61179 seq=9 rtt=93.3 ms
len=208 ip=64.34.231.90 ttl=48 id=61230 seq=10 rtt=91.3 ms
...
^C
--- 64.34.231.90 hping statistic ---
126 packets tramitted, 126 packets received, 0% packet loss
round-trip min/avg/max = 90.0/94.1/110.9 ms

To ns3.veritcalscope.com:

[~]$ sudo /opt/local/sbin/hping3 -2 -p 53 -E lookup -d 31 208.43.51.3
HPING 208.43.51.3 (en0 208.43.51.3): udp mode set, 28 headers + 31 data bytes
len=208 ip=208.43.51.3 ttl=54 id=28584 seq=0 rtt=61.3 ms
len=208 ip=208.43.51.3 ttl=54 id=28724 seq=1 rtt=61.1 ms
len=208 ip=208.43.51.3 ttl=54 id=28889 seq=2 rtt=62.4 ms
len=208 ip=208.43.51.3 ttl=54 id=28972 seq=3 rtt=60.4 ms
len=208 ip=208.43.51.3 ttl=54 id=29311 seq=4 rtt=86.9 ms
len=208 ip=208.43.51.3 ttl=54 id=29448 seq=5 rtt=60.1 ms
len=208 ip=208.43.51.3 ttl=54 id=29526 seq=6 rtt=61.6 ms
len=208 ip=208.43.51.3 ttl=54 id=29685 seq=7 rtt=60.0 ms
len=208 ip=208.43.51.3 ttl=54 id=29809 seq=8 rtt=61.4 ms
len=208 ip=208.43.51.3 ttl=54 id=30063 seq=9 rtt=60.0 ms
len=208 ip=208.43.51.3 ttl=54 id=30154 seq=10 rtt=60.6 ms
...
^C
--- 208.43.51.3 hping statistic ---
141 packets tramitted, 141 packets received, 0% packet loss
round-trip min/avg/max = 56.2/61.4/86.9 ms

However, when we test ns2.verticalscope.com:

[~]$ sudo /opt/local/sbin/hping3 -2 -p 53 -E lookup -d 31 64.34.231.71
HPING 64.34.231.71 (en0 64.34.231.71): udp mode set, 28 headers + 31 data bytes
len=208 ip=64.34.231.71 ttl=47 id=43688 seq=1 rtt=93.3 ms
len=208 ip=64.34.231.71 ttl=47 id=43696 seq=2 rtt=91.6 ms
len=208 ip=64.34.231.71 ttl=47 id=43705 seq=3 rtt=89.6 ms
len=208 ip=64.34.231.71 ttl=47 id=43718 seq=4 rtt=93.3 ms
...
len=208 ip=64.34.231.71 ttl=47 id=43881 seq=12 rtt=107.7 ms
len=208 ip=64.34.231.71 ttl=47 id=43957 seq=15 rtt=110.3 ms
len=208 ip=64.34.231.71 ttl=47 id=43968 seq=16 rtt=91.1 ms
len=208 ip=64.34.231.71 ttl=47 id=44003 seq=17 rtt=93.6 ms
len=208 ip=64.34.231.71 ttl=47 id=44024 seq=18 rtt=92.5 ms
len=208 ip=64.34.231.71 ttl=47 id=44046 seq=19 rtt=91.9 ms
len=208 ip=64.34.231.71 ttl=47 id=44055 seq=20 rtt=90.6 ms
...
len=208 ip=64.34.231.71 ttl=47 id=44785 seq=60 rtt=96.4 ms
len=208 ip=64.34.231.71 ttl=47 id=44820 seq=62 rtt=93.6 ms
len=208 ip=64.34.231.71 ttl=47 id=44838 seq=63 rtt=94.5 ms
^C
--- 64.34.231.71 hping statistic ---
64 packets tramitted, 44 packets received, 32% packet loss
round-trip min/avg/max = 88.9/93.6/110.3 ms

Looks like ns2 is the culprit there with significant packet loss. The packet loss, coupled with the low TTL values are causing DNS caches frequent lookups against ns2.verticalscope.com, which is having trouble responding in a timely fashion. If you can't lock down why the packet loss is occurring (at the node, or in the network), then a work-around is to adjust your TTL for www.ducati.ms to something higher, like one hour (3600).

Cheers

[Uzziah]

Originally Posted by SE84

Well, if your DNS cache on your ISP's server had fetched the IP and the timeout for the cache was right before you clicked 'edit,' then you would need another lookup and if it timed out...well.

If you add the site to your local hosts file all the problems go away. But, be my guest...I'll just take my twenty years of Unix and Network experience and be over in the corner.

Cheers

I wasn't trying to insult or condescend, just to let you know I'm not unfamiliar with either the specifics or the vagaries of IP traffic. Seems like I'm not the only one suffering from the issue.

Curious that adding it to the localhosts file would be the proper fix, though. Hard to believe this site would need that, but none of the others I frequent. Or that the rest of the folks would be required to do that, much less even know what we're talking about.

I should add that it doesn't seem to ever be a problem during the evening hours, and seemed much more pronounced during US daytime hours today. The errors seem almost like a bandwidth problem trying to handle all the traffic on a segment, sort of like when apple released 3.1 for iPhone or when there is a DOS attack on a site or network. (FWIW, I actually took about thirty minutes to compose this post...and it worked the first time.)

I'm know I'm just a new member, and a free one at that. But this is indeed an atypical experience for me, and one that also seems more typical for *other* members here. If the tech guys behind this site can find the issue and resolve it, great. If not, I'll just end up avoiding the site during what seem to be peak traffic hours when the issue is prevalent. It will probably also affect my decision whether to continue as an unpaid member or directly support the site. It seems to have plenty of advertisers and supporting vendors, so my paltry contributions aren't needed anyway. (Probably in the post sense as well as the financial.)

In the end, the dual points of starting this thread was to inquire about the issue so that if they weren't aware of a problem they could check into it, and also to see if others (at least the normal folks who hadn't figured out a site-and-tech-specific workaround) were experiencing difficulties. I believe both purposes have been served, so I'll just hope the former is resolved so the latter can be, too.

EDIT: In the time I was waiting, I see you took some time to do some very focused detective work, so one way or the other it looks like they'll get to the bottom of it.

Take care and enjoy the ride,
Greg