Uptime Stats Correct?

[Deverill]

I'm just wondering how good the uptime statistics are and if they are correct.

 

I looked at the stats for 9-10 and they say:

 

From: Sep 10 2007 12:00 AM (US/Eastern)

To: Sep 11 2007 12:00 AM (US/Eastern)

...

43. server101.totalchoicehosting.com (server101.totalchoicehosting.com)

HTTP Test (Port 80)

- uptime: 99.142%

- downtime: 10min

- outage: 1

 

However in this thread it seems that server101 was reported down at 3:14 AM and at 6:25 AM Vivek said they were still working on that server.

 

Am I reading it wrong?

[Head Guru]

We dont control the uptime stats, those are provided directly from hyperspin.com

 

The server's httpd service was only down for 10 minutes. This most likely was due to a routing issue at the datacenter. The server was reachable for certain people on certain inbound connections, but not on others.

 

Bill

[Madmanmcp]

However in this thread it seems that server101 was reported down at 3:14 AM and at 6:25 AM Vivek said they were still working on that server.

 

Am I reading it wrong?

 

No, you are reading what the monitoring service has recorded, and they did not detect it as being down. Servers are equiped with 2 connections (nics) connected to the network and they are called the "frontend" and the "backend" and each are assigned different IP's. Normal Internet traffic uses the frontend IP and all the administrative work is done on the backend IP, monitoring is part of the administrative work.

 

Now the thread you point to says there were network problems at the datacenter and what could have happened is a switch or router or some other network device went down and it was a device that handled all the "frontend" traffic, the device for the backend traffic was still working. So the server was still "up" and working and the monitoring service still saw it as up and thats what it reported. Technically it is correct, it was up...but it was not visible to some people.

 

The monitoring service reports what it sees and does not know that the frontend was down which affected your server. I am sure the datacenter had monitors for their network device and there are reports saying that device and others were down but it could not put that in the report you are reading for your server.

 

Hope that helps you understand.

[Deverill]

Thanks for the explanation, Madman. BTW, it's not my server, just something I noticed... the 100% seemed a bit high based on forum discussions.