[Looking for Charlie's main web site?]

CF911: Want to monitor ColdFusion "out of process" (from outside the instance itself)? Many ways.

I just blogged about how the hidden gem "enable monitoring server" option in CF 9.0.1 does NOT cause the CF Server Monitor to somehow magically run "out of process". See more on that.

Yet people will reasonably want to be able to have some mechanism that "watches" CF "from the outside", to know when it's gone down. How can you do that? That's what I'll point out in this entry.

And beyond talking about what goes along with the CF Enterprise Server Monitor, I'll also point out options for those who are NOT running CF 8, 9, or 10 Enterprise and therefore do not have the Enterprise Server Monitor. This also includes those CF 6 or 7. There are solutions for you, and also for those running Railo, BD, or indeed any Java server. More on all that in a moment.

This is part 4 of an unexpected series of entries today on the CF Enterprise Server Monitor. :-) I got on a roll, and each seemed deserving of its own topic. See the "Related Blog Entries" below this entry for links to those.

What the CF Server Monitor is, and is not

To be clear, The CF Enterprise Server Monitor (and indeed, FusionReactor and SeeFusion in their basic configuration) is "just" a web interface (Flex-based) that talks to an embedded flex gateway component running within the CF instance/address space to get information about how the instance is doing. If that instance goes down, then the web interface will have nothing to talk to any more, and the "monitor" will no longer be of value.

But there are alternatives to watch the instance (indeed, multiple instances) from the outside

First: The CF Enterprise Multiserver monitor

First up, let me talk about what's built into CF 8/9/10 Enterprise, to go along with the CF Server Monitor. Many never notice it, or they misunderstand it: the Multiserver Monitor. This is different from (but closely related to) the Server Monitor.

It is a single Flex-based interface that watches whatever other CF Enterprise server monitor instances you tell it to watch. It does run "out of process" in that it runs on your desktop (as a Flex-based web page) and can show you if a monitored server has gone down or become unresponsive. It also shows a few key stats about each monitored instance in the one interface, while such monitored instances are up. If you want to see the full details of what's going on, you then can ask it to open the full CF Server Monitor (as another web page).

I talk about it more in an article I did back in 2008, part 4 of my 4-part series of articles on the Server Monitor.

But briefly, it's launched from the same "Server Monitor" page in the CF Admin. I suspect that many mistakenly assume that it's only for monitoring "instances in a multiserver deployment of CF". It can do that, but it can be configured to watch any other CF Enterprise instance (whether on the same machine or on another, and whether that's running in a Server, Multiserver, or indeed J2EE form of deployment.)

For more on how to configure the Multiserver Monitor itself, as well as how to configure a server to allow you to "watch" it (you need to tweak an XML file on the server to be monitored), and more, see that article above.

I should note, as well, that CF 9 (and 10) now offer a Server Manager, also launched from the same "server monitor" page in the CF Admin, and it does much the same as the Multiserver Monitor, and a little more (as well as a little less). You may want to look at both to decide for yourself. See the CF documentation for more on the Server Manager.

What if you don't have CF 8/9/10 Enterprise?

Since the CF Server Monitor and Multiserver Monitor come only with CF 8/9/10 Enterprise (and the MS Monitor can only monitor such an instance), what do you do if you're on CF Standard, or CF 6 or 7? Or indeed Railo, BlueDragon, or Open BlueDragon? There are solutions for you.

First up, as I talked about in an entry back in 2007, CF8 monitor doesn't run on CF8 Standard, or any 6 or 7. What to do?, there are indeed alternative CF Server monitors, FusionReactor and SeeFusion, which both work on CF Standard as well as CF Enterprise.

FusionReactor even works with Railo, BD, OpenBD, or indeed any Java server. And by that last point, that means such things as LiveCycle and Flex Data Services, as well as generic tools that run on Java like JIRA and Confluence. It also means the Solr Server that comes with CF 9 and 10!

And of course both FR and SF also work just fine on a CF 8/9/10 Enterprise server, even if you have the CF Server Monitor running. They can all run with very little overhead, though of course see the entry I did back in 2007 on potential overhead concerns with the CF Server Monitor.

FusionReactor and SeeFusion each also offer an Enterprise Dashboard feature

But the point for this entry is that both FusionReactor and SeeFusion each of offer an Enterprise Dashboard feature which both offer a single interface to watch multiple monitored instances (on one or many machines), similar to the CF Enterprise Multiserver monitor.

You do need to have FusionReactor installed on any instance to be watched by its dashboard, and you need SeeFusion installed on any instance to be watched by its dashboard. And you do need the Enterprise edition of each for them to be watched by their respective Enterprise dashboards. But to be clear, the CF license does not need to be Enterprise (and again, FR can watch Railo, BD, and other Java servers.)

There is a page on the FR site devoted to FusionReactor Enterprise Dashboard. There's no specific page on the SeeFusion site about its Enterprise dashboard.

What if you're not sitting there in front of these multi-instance dashboards?

A reasonable question might be: all these dashboards to watch multiple servers at once is nice, but it won't help if you're not looking at them when things go amiss. And that's true.

Now, before I talk more about these multi-instance dashboards, let me clarify that all 3 CF monitoring tools have means to notify you by email when an alert condition arises, as detected within the running instance. In the CF Server Monitor, they're called "Alerts" (and I discuss them in part 3 of my 4-part articles series.) In FusionReactor, they're called "crash protection notification", and that's discussed more here and here. And in SeeFusion, they are called Active Monitoring Rules (there's no page dedicated to them that I can point you to.) These alerts are available in the Standard edition of both FusionReactor and SeeFusion.

But if those individual instances become unresponsive or crash, then those alerting mechanisms will not notify you, either. (That said, they may well alert you when you can't get into the monitors, but the instance is still running. They can also alert you with info just before the instance goes down, and may have valuable info. So these alerts are indeed very worthwhile. I should write more about them separately some day.)

Back to the point of the multi-instance dashboards, where the one interface is watching the others, that's a different story.

The CF Enterprise Multiserver Monitor is just an interface: it has no means to email you that a monitored server has gone down. Neither does SeeFusion's dashboard, as far as I can tell.

FusionReactor offers several enhancements in its Enterprise Dashboard

But FusionReactor's Enterprise Dashboard has several enhancements over the others. I didn't set out in this entry to make this point. It just is worth noting.

First, it can in fact be configured to email you when a monitored instance becomes unresponsive (goes down) and comes back up. That can go to one or more people, and of course it could go to an address that could trigger an SMS or other text message. (More on mobile device notification in a moment.)

Second, the FR Enterprise Dashboard configuration page can also be configured to run a script when a monitored server goes down or comes back up. For more on that, see "Using FusionReactor Enterprise Scripting" in the FR documentation.

Third, FusionReactor offers an available AIR version of its dashboard, which can pop up an alert from your desktop status tray when a monitored server has problems.

Fourth and finally, and a delight to many I'm sure, there are available mobile app versions of the FusionReactor Enterprise Dashboard (IOS and Android).

So, whether you're coding away, or playing Angry Birds, whether you want a mobile interface or a text message, or good 'ol email. You can get notification about your monitored servers.

So you can, indeed, monitor CF from "outside the process"

Again, the main point of this entry was to show how in fact you can monitor CF (and other CFML engines) from "outside the process". Those on CF 8-10 Enterprise have the MultiServer monitor and Server manager. Those on CF 6-10 Enterprise or Standard have FusionReactor and SeeFusion and their Enterprise Dashboards. FusionReactor also supports other CFML engines and indeed any Java server/app (whether those you have associated with CF, including Solr, LiveCycle, FDS, and more, or really any Java server like Tomcat, JBoss, Glassfish, WebSphere, etc.)

And all 3 tools offer email-based alerts from within the process, which certainly has its own value.

Still other non-CF solutions

Finally, someone may want to ensure that I point out that there are, of course, many other monitoring tools that can "watch a CF server" from "outside the process". I did focus here just on those that are devoted to CF itself. But there are many different kinds of other tools you may want to consider.

For instance, there are free (and paid) tools and services which will send requests to your server on a periodic basis and will then email you when they are down (and may track that in an interface). I talk about such tools in a category of my CF411 site: Web Site Uptime Monitoring Tools. Some may recognize some prominent ones like Pingdom, Watchmouse. There are many more I list.

Second would be the category of system monitoring tools, like Nagios, ServerDensity, Spiceworks, ManageEngine, and others. I have a category for such tools, System Monitoring Tools, again listing many more, free and commercial.

Third would be Java Application Monitoring Tools like NewRelic, JaMon, Glassbox, Orion, and others. And I have another separate category for monitoring a JVM itself, but again the focus here is "out of process" monitoring.

Indeed, I have several other categories of monitoring tools for other aspects of a web environment. Check out the top-level category that those and the others abvoe are in, Monitoring Tools/Services, to include also database monitoring tools, other (in-process) ColdFusion monitoring tools, SAN/NAS monitoring tools, web analytics tools, and more.

Bottom line: there are many ways to watch a CF instance from "outside the process". Go nuts, kids!

And let me know what you think.

Dont people have multiple instances of CF on seperate servers that use probe.cfm to health check each other and send an email if an instance crawls or dies?
# Posted By Mark Ireland | 2/24/12 7:31 PM
They may, sure. And I suppose I could/should have added that as an option at the end there. Thanks, Mark. So, any other feedback? :-) The real focus was on more traditional monitors. I only threw in that last section as a nod to other possible options.
# Posted By charlie arehart | 2/24/12 11:05 PM
BlogCFC was created by Raymond Camden. This blog is running version 5.005. (Want to validate the html in this page?)

Managed Hosting Services provided by
Managed Dedicated Hosting