Recent site problems

chip

MutualFundObserver.com was down for just over 20 minutes today. I filed a report with our hosting provider and they indicated that they'd had some network disruptions which have since been cleared.

I've been monitoring the site for over a month using a service called pingdom.com which sends me alerts if the site is down for more than 5 consecutive minutes. I created a public pingdom page for MFO today, so that everyone can see the odd pattern of very short interruptions that we've been experiencing. You can see it at http://stats.pingdom.com/rz59sk2qql6d.

At this point, I can't really tell if the problem is with the hosting provider or with our site. Today's longer disruption presented an error of "HTTP CRITICAL - Unable to open TCP socket" on pingdom and the hosting company mentioned their network problems. I've seen this error a few times. With some of the shorter outages, there's not a diagnosis from pingdom, but there is a traceroute showing lots of timeouts.

Does anyone have any ideas about this? Or, a suggestion as to whether it's "normal" for a site on a shared virtual host to experience these types of outages?

Thanks for taking the time to look!
chip

Old_Skeet

Hi Chip,

Thanks for letting us know of the problems. There has been times that I would try to make a post and it would just grid and not do much of anything. And, at times I’d try to link into the board and it would just grind away. So from my thoughts your post explains probably much of the problems I was having. I ran a health check on my computer through OMG and everything checked out ok. With this, I’d say there were issues with the host or perhaps even the site itself but at this stage lets go with the host for now. Perhaps pingdom will, in time, provide an answer.

Back to lurking,
Skeeter

catch22

Hi chip,
Your skills in this area far surpass mine.

I was a troubleshooter in a former life. IMHO, if Pingdom can provide data as to where to problem area existed; this then, would be the only valid information as to the problem area.

A server ping during this period would have indicated no connection. A tracert would have narrowed to the point of the last good connection. If Pingdom can provide system data comparable to a tracert, then one would expect valid problem area to be known.

TotalChoice hosting would not find any problem, if the problem existed outside of their servers. "HTTP CRITICAL - Unable to open TCP socket" would indicate no connect to the server, yes? But, where did the connection fail, eh?

My best guess is too much net traffic or someone/organization that may have had a router problem that affected the area in and around Troy, Mi. Keep in mind that the largest reported current data streaming and usage is now taking place during the NCAA basketball playoffs, versus any other events. Multiply this against the streaming video mobile devices now in use, a Saturday morning in and around Troy, Mi (a fairly affuluent area in Mi and lots of internet/mobile users) and that the regional NCAA games were about to start in Auburn Hills, Mi (about 10 miles away, as the crow flies) from Troy. I wouldn't be surprised that a router was brought online and was addressed improperly as a global router or similar, and steamed large amounts of data. It would nice to know if other service was affected in the area; although our home router does stream through a large vendor about 20 miles away from Troy, and our service was not affected.

This does not explain earlier problems, but only a possible consideration for March 23 (Saturday).

Just my 2 cents worth. Back to the projects here.

Take care,
Catch

hank

Reply to @catch22: Hi Catch. Hope you're well. IMHO you need to check in more often. Hope the project's going well. Take care.

Anna

Reply to @catch22: HI, ditto what hank said. I had just finished telling my husband how much I missed your posts and here you are!

Accipiter

Totalchoice - could have BGP routing loops, dns problems, firewall problems, router problems, etc. All could lead to TCP unable to open a socket. Based on what Totalchoice is telling you, it seems they are trying to correct the problem. What you could do, is find someone else hosting on Totalchoice and seeing if they have disruptions at the same time. That would narrow it down. I'm sure @Investor will chime in here, to any gaffs in scenarios presented.

>At this point, I can't really tell if the problem is with the hosting provider or with our site.

I don't see how intermittent TCP open socket problems could have anything to do to any code on your site.

>Or, a suggestion as to whether it's "normal" for a site on a shared virtual host to experience these types of outages?

I think you would see a performance decrease with "a shared virtual host" , not a failure or outage, but since I have no experience, take this with a grain of salt.

Investor

I don't think you can tell if the problem was the host/blade or there was a network problem on any hop between your host and the monitoring service. Try to narrow it at least by monitoring the hosting company edge router and a couple hops downstream.

chip

So far, in the past two days, pingdom only indicates one brief (<1 min) interruption. The related traceroute that they supply seems to point to two separate routers, but not to anything near TotalChoice. It's also different routers in each case, since pingdom performs its service by checking in from various locations throughout the US and Europe.

I'll continue to watch.

Old_Joe

Reply to @chip: FWIW, happened to get up at 4 AM (Pacific time) and since computer was sitting here already turned on took a quick look at MFO. Found response times to be really quick, as opposed to other recent quick looks during the daytime. That sounds like a traffic issue somewhere.

BTW, by coincidence both Catch and I are in the middle of major remodeling/maintenance construction on our homes, so even though I leave the computer on most of the time my actual time spent reading/commenting on MFO is much less than normal.

Accipiter

Aside from outage issues, I wonder if people who have slow response times occasionally, have auto-updating of windows and other programs going on at the same time in the background.

Old_Joe

Reply to @Accipiter: Yes, that's for sure. I have about ten windows open right now, and some of them are in fact constantly updating. I usually keep MFO up all the time, which is why it shows me on-line on the main page, but I'm not really paying attention most of the time, especially lately what with major construction going on here. Later... thanks & hope all is well at your end.

OJ

Accipiter

I'm talking more about behind the scenes updates to the OS e.g. windows 7 updates updates the OS automatically depending on the settings and this can be an enormous amount of data straming through which will slow down performance. Other OS's can do this as well. But many browser windows open can also effect performance if they are streaming videos or audio etc.

hank

Linked is recent NY Times article on major ongoing cyber attacks. Probably nothing you're not already aware of. Few problems here. Occasionally slow to load. I use a 3G cellular connection & I-Pad (non-embedded link) FWIW

http://www.nytimes.com/2013/03/27/technology/internet/online-dispute-becomes-internet-snarling-attack.html?pagewanted=all&_r=0