Launch of the server monitoring and response service!
By Ryan on Thursday 16 July 2009, 11:33 - Hosting - Permalink
This service will let you to monitor all of the statistics of your server that we already make available to you, and to automatically perform a certain action if you wish.
We have (for now) implemented 3 types of alerts that you can set up:
Exceeding a threshold
In this case, your alert will be activated and your action list will be executed when the parameter that you are monitoring exceeds a certain value (x) over a certain amount of time (t1). The alert will stop once it falls below the limit for another predefined time period (t2).
Dropping under a threshold
In this case, your alert will be activated and your action list will be executed when what you are monitoring falls under a certain value over a given amount of time. The alert will stop once it returns above the limit for another predefined time period.
Inactivity of one of your parameters
Finally, our alert will be activated and your action list will be executed when the parameter that you are monitoring seem to no longer be active. The alert will stop once the probe detects that there is a return to activity over a chosen time period.
When an alert has been given, you have the possibility of executing an action list such as:
- Emailing you a warning. You can choose the email address that you want to receive this alert, and when you want to be warned.
- Send you an alert by instant message. This way, you can be alerted by ICQ, AIM or even IRC (MSN will soon be available as well)
- automatically add a Flex Share to your server. This way, you can optimize the quality of your service by having your server's power increase automatically with the load. Imagine, for example, that you are enjoying a peaceful vacation, sunbathing on a nice beach somewhere and...(ok, I need a vacation haha...).
Simply imagine that you are unavailable, and do not have an internet connection right at hand, when a nice journalist publishes an article about your new website. Suddenly, the 2 shares that you estimated for your service are no longer enough to handle the load: our new system will now act on your behalf, and so you can now have the time to calmly analyze the situation in detail. Adding shares to your server and increasing its power will happen automatically and without having to restart your server.
The interface is currently available for everyone to use. It may be found on your server's management page, right under the performance monitor graph. We will be adding additional actions to this list (such as the possibility of rebooting or stopping the server) in the near future.













Comments
Looks interesting. I think it might be useful as well to have the option of configuring it to visit some special http url set by the user as action. This would generally be more simple for the user to implement some useful action directly on the server to handle the situation automatically.
Kaetemi: that's a good idea...we'll see how to handle that quickly
This is a great service! Thank-you!
A few notes:
1. The CPU alert, at least, appears in French.
2. It would be nice if you could edit alerts. (e.g. change trigger point from 90% to 80%.)
3. If the target is "disk", the options for "name of the target" are in French.
4. Can the From email address be something like "probe" or "alert"? At the moment it's "hosting"--I think it's should be something more explicit to facilitate filtering. (And it definitely should if non-alert messages will ever be sent from "hosting@support.gandi.net".)
Ummm..... is there a link somewhere?
Michael S.: thxs for the bug report...we will fix that asap and change the sender
Justin Noel: the block just below the graph on your server page
Please consider supporting IM via XMPP (Jabber, Google Talk, etc) as well.
@Justin - I missed it first time too, On the page 'Servers', at the bottom is a list of all the virtual servers you have. You can click the name of the server to see more information about it, like the resource usage graphs.
I too would like http response monitor as Kaetemi, and Michael S covered other things I noticed.
On the servers page, (https://www.gandi.net/admin/hosting...), the Disk size is reported in Go, could be GB for English language (not Gb! GB = Gigabyte, Gb= Gigabit)
Email alert sent to me was in French, fine for me but the account is English so perhaps other customers won't understand.
I would like a monitor to tell me when I reached a threshold of disk usage, rather than load if that's possible (though there are other ways to check this that spring to mind, so I'm not overly concerned)
I would also like an SMS message option.
Thanks for creating the monitor system, having just moved half our web services to a Gandi virtual server, one of the things on my todo list is organise a monitoring service of some description and you've just done that for me
Josh: yes of course XMPP will be soon in the list (I already have fixed the french news but forgor the english one)
Steve Root: Will have a look at the Go/Bg/GB...The mail language problem will be fixed on monday or tuesday. For disk usage, except for Gandi AI, we don't have access to your filesystem, so it's not really possible, Leland is looking at the SMS platform, we will probably add it too.
Great feature. Sadly, no alerts ; -).
If I can make one suggestion, as your hosting service continues to grow, it would be to move your server operations from the web _site_ to a dedicated web _application_. This would allow you to improve on the UX from scratch and not be tied to a layout and graphical style that is arguably less than ideal for its purpose. Part of this suggested move is basic design psychology; form follows function.
handy...
+1 for http (and other tcp level) probes!
Thanks for the monitoring!