Third party uptime monitoring?


#1

So the uptime stats on the “System Status” page on the Web Control Panel are reportedly not accurate, is it better to have some kind of third party monitoring like WebSitePulse?


DreamServers.net :: DreamWhois.com :: DreamRegistrar.com


#2

Yes and no. WebSitePulse will be more accurate and tell you when your site is down, but so would a simple visit to your site.


MacManX.com
I don’t work here. I’m just your typical support forum volunteer.


#3

[quote]but so would a simple visit to your site

[/quote]

You mean every 5 mins? :smiley:


DreamServers.net :: DreamWhois.com :: DreamRegistrar.com


#4

(> but so would a simple visit to your site

You mean every 5 mins? :D)

He has a point there, macman :slight_smile:

IAmAtMyWitsEnd


#5

That he does.


MacManX.com
I don’t work here. I’m just your typical support forum volunteer.


#6

The reason that I suggest a third party monitoring of uptime is that:

a) it may actually bring more credibility to DreamHost. I’ve been with DreamHost for quite a long time, and always been happy with and trust their uptime stats (historically 99.97%; NOT 99.7%) until more recently. I think with the “problems” lately, an independent source will bring more faith esp. to new or potential clients.

b) for those users who have their own clients on board, this will definitely help relief some burden off our shoulders, e.g. in reporting and/or explaining uptime (or more relevantly, downtime).


DreamServers.net :: DreamWhois.com :: DreamRegistrar.com


#7

In light of the latest extended downtime I suffered this morning (re: server “Apok”), allow me to raise this issue again.

To be frank, when all my sites had been down for about 3 hours, and all I saw from the System Status page was “green” all the way (Up for 10 mons 14 days. 100% 100% 100% 99.9974%), do you think it’s reasonable to ask for some relatively more reliable uptime reporting, so that we would at least know what is happening?


DreamServers.net - What do you think?


#8

While we are on the topic of redoing system status, thought I should post some of the ideas I’ve been thinking of submitting to the suggestion area.

Make System Status it’s own tab or add it to the support area.

Add support for mysql in the status page or just create a general mysql page listing the status for all of the servers (because a user could be using different mysql servers).

MRTG graphs - provide snapshots of the bandwidth for each of the uplinks along with the server(s) you are on.

Add a special section for each server. e.g… Bob Users, Neo Users where admins/devs/techs can post a quick update if the server is down and not require a whole announcement which is usually received after the fact. This would be especially useful if the server needed a kernel upgrade the dev could write - “kernel upgrade on the server - downtime 2 minutes,” or “server hardware is toast, swapping servers.” It would probably reduce the load on support too.

happylittlethings.com


#9

Now it’s getting even weirder (or more weird):

Service: http
Status: Up for 10 mons 14 days.
1 Day: 100%
1 Week: 100%
1 Month: 100%
All Time: 99.9989%
Last Updated: 6 hours 7 mins ago.

Last Updated 6 hours 7 mins ago? That’s AFTER the crash. But how come the All Time uptime is even higher than yesterday BEFORE the 3-hour crash!!

Am I missing something here? Are we on the same page?


DreamServers.net - What do you think?


#10

The statistics listed there are tied to the physical hardware your server (apok in this case) is on so when the hardware was replaced the other day, the stats went up because the newer hardware has had better historical uptime than the old hardware. That is not made clear from the web panel and is probably not the best way to present the information either so it is understandable that you would be confused!

Regarding the server uptime statistics in general, they are based on our own monitoring system we use to detect downtime. The monitoring system does not watch individual websites but rather the server itself and the different services running on the server. It is very possible that everything on the server appears to be fine according to our monitoring system but an individual website is down due to a misconfiguration or a failure somewhere else in the system. In a case like that the outage would NOT be counted in the uptime statistics shown in the web panel.

Our current system uptime statistics are not ideal and are showing some age and are in need of some attention. They are a general approximation of your server’s hardware and software uptime, but they do not indicate uptime for your own website. It is not practical for us to keep track of that information for every single website so if you need very precise information you would be best served by an outside tracking service. We do have plans to improve our provided uptime statistics, but outside services have a more objective view of the situation and are often able to be more statistically precise as a result.

  • Dallas
  • DreamHost Honcho

#11

[quote]It is very possible that everything on the server appears to be fine according
to our monitoring system but an individual website is down due to a
misconfiguration or a failure somewhere else in the system.

[/quote]

The thing is Dallas that the Service Status display misrepresents the uptime figures as for the /domain/, not for the machine. I think you could avoid much dissatisfaction by simply correcting the misrep.

Good luck, and thanks for the mostly great service.


#12

We don’t monitor individual domains and do not have any plans to begin doing so. There are third party services that offer such functionality for anyone that needs it. In such cases it is generally better to have someone on an outside network doing the monitoring anyway!

  • Dallas
  • DreamHost Head Honcho/Founder

#13

[quote]We don’t monitor individual domains

[/quote]

Well yes, and that’s what makes the uptime stats a mis-rep - the stats are presented for each individual domain name. I’m suggesting you change this to present the stats for the servers, not domains, that in truth they do apply to.


#14

I use host-tracker.com, it’s a completely free service that monitors unlimited number of web sites. They will even check your website every minute and email you and/or text message you if it can’t connect.