Host and service views

The global view is most easily accessed via the Monitor menu.

Alternatively, you can use the Overview and Views snap-ins. In the Overview each number is clickable and takes you to a global view that individually lists the hosts or services counted in each case.

All global views are accessible in the Views snap-in — grouped according to topics. In addition you will find a few entries that are not actually Views — such as for example, the dashboards, which are sorted under the Overview topic. Dashboards can however contain views.

From a global view, with a further step you can access the details for a particular host or service. The names of hosts and services and the individual cells in other columns are clickable …

… and bring you to a more detailed (context-dependent) view.

In views that relate to a specific host or service, you will find all sorts of entries in the menu Host or respectively Service, which will then take you to other views with the same context — i.e. for the same host or service.

You can also access host- and service-specific views by searching within the Monitor menu:

Another way to the detailed view is via the Quicksearch snap-in in the sidebar:

Which view you receive will depend on the result of the search. If the search identifies a host explicitly, you will reach this host’s Services of Host …​ page directly. There you will again find the menu items for the other views of this same host. A click on the host’s name takes you to the Status of Host view.

The distinction between global, host-specific and service-specific views is also noticeable in the menu bar. Always present are the menus Commands, Add to, Export, Display and Help.

In global host views there is also the menu Hosts as well as in global service views Services — note the plural here! You can use this, for example, to call up the availability of all hosts or services of the current view.

Views of a single host, on the other hand, contain the menus Host and Services with all actions for the current host and all services for the current view.

Views of a single service likewise contain the Host menu, but additionally, now in the singular, the Service menu, with actions for this one service.

At the top left of every view there are six symbols with which you can influence the display, and trigger actions:

Checkmk shows time stamps as relative values in all of the views of states if these are less than 24 hours in the past or future — e.g., 16 hours. You can switch this in the menu via Display > Modify display options and choose between displaying only absolute, only relative or both values, or also the ‘Unix timestamp’. You can also specify the date format in this dialog.

The views can be sorted by clicking on a column’s heading. A column has three states that can be selected through a loop of multiple clicks:

Views are initially sorted ‘naturally’ according to how the view is defined. In service views the sorting is alphabetic by service name — with the exception of Check_MK services which are always at the top. The Check_MK service is responsible for managing the monitoring agent. There are also the Check_MK Discovery and Check_MK HW/SW Inventory services. In order to sort such a list by the current state of the services a click on State will suffice.

Sorting by the Perf-O-Meter column sometimes produces surprising results. This is due to the graphic display of the measured values being partly a percentage summary of the actual values. The sorting is however performed according to absolute values, and is always based on the first metric produced by a service.

Data displayed in a view can be exported in various formats using the Export menu:

In a larger monitoring environment displaying all views is no longer practical. When you are monitoring 50,000 services and select the All Services view, not only will the display require a very long time — it will also not be very useful.

In such situations, in order to protect the user from long waiting times and to avoid crippling the system with absurd quantities of data, views are limited to 1000 entries in their display. Exceeding this limit produces the following warning:

As you can see, the records being displayed are not necessarily the first 1000 corresponding to the selected sorting method! There is a technical reason for this: namely that the limit is applied to the data source in the connected site’s monitoring cores. This is very important, because if we accumulate one million data records from your environment spread around the world, then 99.9 % of the data will be deleted immediately. The sorting takes place over the interface only at the end, thus it happens after the limit. The data from all sites must, after all, be sorted together.

If you really want to see more than one thousand records, then you can reach the next level by clicking on Repeat query and allow more results. Here the limit is 5,000 records. If this limit has again been exceeded, with unlimited you can continue. Insofar this is a potentially risky action, you will require Administrator rights. You have been warned!

You can define both levels under Global settings > User interface:

Checkmk allows you to customize its standard views, and even to create new ones that you can incorporate into reports and dashboards. Thus you can define numerous different aspects for every view:

The edit mode for views can be accessed in two ways:

The views supplied as standard are a part of the software and as such cannot be changed, however Checkmk does recognize the concept of cloning. When a view is first customized a copy of that view is created automatically. This copy is added to your user profile.

This copy can then be customized as desired. The original view is retained but is grayed-out — overlaid by your new version in effect. You can return to the standard view later by simply deleting your clone (achieved in the table of views, as you might expect, with ).

This concept has one further advantage: namely, that you can define whether the view should be changed for all users or just for yourself. You specify this in the General Properties of the view under Visibility with the Make this view available for other users checkbox. Not surprisingly, you can only select this checkbox if you have administrator permissions (or more correctly, this function has its own permission — Publish views). Additionally, single views can be locked in the role definitions.

What happens when a view has been customized and published by several users? Each user then has their own variant of the view. Which view will be visible for which user(s)? This can be determined with the following rules:

How can you create a real copy of a view, so that when done you can have both the supplied and your own views? This is defined by using Unique ID in the General Properties. Simply give your view a new name, so that it will no longer be identified as a clone of the supplied view, rather it will begin its own independent life.

This ID is the decisive keyword for opening views in the URL. The schema is very simple. Here for example is how the global view with its ID allhosts is opened:

/mysite/check_mk/view.py?view_name=allhosts

The concepts of cloning, customizing and visibility can be found at many other locations in Checkmk, namely in:

How and if a view will be shown in the Monitor menu of the navigation bar as well as in the Views snap-in, is defined under General Properties by the following characteristics:

The next block — View Properties — defines a view’s general appearance:

The Number of Columns option sets the default for the number of columns shown in the view, the Automatic page reload the default for the refresh counter. Both values can be modified by the user in the view itself via Display > Modify display options.

Under Basic Layout there are various styles for displaying the data in tables. Most views use Table — a normal table that can be sorted by columns — or Single dataset — which has the legend on the left and which is mostly used for single data sets. You can however also use Single dataset for views with more than one object. The All hosts view looks something like this when altered to Single dataset:

The Columns box defines which columns you wish to see. The possible selection of columns depends on the selected data source. The most columns are found in services, naturally, as all information for the particular host is available here. The list can be quite long here, and if you are uncertain which column is the right one, there is only one thing to do — try it out:

Depending on the column selected, there may be specific options for this. With the column Hosts: Hostname, for example, it is possible to color the host name according to its state using the Coloring option.

The Link field offers a selection of all views and dashboards. If a view is selected here, then the column’s respective cell is clickable and takes the user to the chosen view. This really only makes sense if the targeted view has a context. The best example is the All hosts view. The Hostname column is clickable here and takes the user to this host’s Services of Host. Alternatively, dashboards can be specified here — which obviously do not require a context.

Under Tooltip, on the other hand, you will find a list of all columns. Thus you can show further information for the host or service when the user moves the mouse cursor over the respective cell (the IP address in the following example):

Let’s imagine that you would like to display information on specific services in a table of hosts. The following shows a very good example of such a scenario: For each host, the current uptime, CPU load, memory usage, and execution time of the Checkmk agent are output:

Here a table of hosts has been created, in which for each host the Perf-O-Meter service column from each of four different services is shown. You can also see that for one of the three hosts the CPU, Memory and Uptime services do not exist and so the columns for those services are consequently empty.

To implement a configuration to this view, add columns of the Joined column type to the view. Here then under Column the columns for services will appear, from which you select Perf-O-Meter in this example. With the Title entry further down you can then define the heading of the column as required if the default heading does not fit.

Under of Service, first select either Explicit match or Regular expression match. For the former, enter the exact name (upper and lower case-sensitive). This will determine the correct service associated with the host.

If you don’t know exactly what the service is called, or its name is slightly different on each host, specify a regular expression that matches to exactly one service per host. This is especially useful if you want to look at a group of hosts where there is a comparable group of services, only one of which can be present on a host at any given time. In the example below, this is the CPU load or the Processor queue, which can only exist on either Linux or Windows systems.

By the way, there is no need to worry if a regular expression does in fact match more than one service. In such a case, the first match will always be displayed in the column.

In both variants, such a display is of course only useful if the view shows a list of similar hosts, which also all have the selected services or have similar services. This is also the reason why Checkmk provides only a few such views: Which columns are useful here depends entirely on the type of hosts selected. Linux servers are obviously interested in completely different information compared to UPSs, for example.

The sorting within a view is configured in the fourth box. It’s only a matter of the predefined sorting method. Users can — as described above — determine the sorting order themselves with a click on the column heading. In the view’s configuration however you have more possibilities — you can define a multi-step sorting order, e.g. first by service state, and for the same state by service name. The order so determined is retained as a subordinate sorting when the user resorts in a specific column.

Through grouping you can divide a table into several segments — in which each segment’s data is related in some way. The best example of this is the Service problems view, which is easily reached via the Overview snap-in.

As you can see, this table is grouped with service State (first all CRIT, then UNKNOWN, then WARN) - each with its own group header:

The grouping in a view is configured similarly to that with the columns. Simply define which column the grouping should relate to. It is usually only one, but can be more. All records with the same value for all selected columns will then be displayed in a group — and the column heading will be shown as the group title.

It is important that you also sort the records by priority according to the group’s selected characteristic! Otherwise it can be possible that the same group makes multiple appearances (which may at times be desirable).

Incidentally, a resorting by column performed by a user has no effect on the grouping — in such a case only the group’s sequence is determined and the records sorted within the group. The groups themselves are unchanged.

An important aspect of views is the data selection. Which hosts or services should be displayed in a table? Checkmk uses the Filter concept for this purpose. Here are a couple of examples of host filters:

Every filter can be defined with search terms or other criteria by a user, thus reducing the list of results to those records meeting the criteria. In this way the filters are AND-linked. The filter criteria actually used for a view are assembled from three sources:

The filters you assemble by editing in the view’s Context / Search Filters box have two functions. Firstly, by clicking on you can decide which filter will be available to a user. Secondly, you can predefine filters using criteria, thus limiting the data to be displayed in the view (point 1 above).

When you create or edit a view by using context — instead of the filter for the relevant object only an optional entry field appears. In this an exact comparison always applies (upper and lower case sensitive). As an example we can take the host view, which displays all services of a specified host. The host name can be specified by adding a context to the view. You can also build a display in which the host to be displayed is effectively hard-coded directly in the view:

In this way you can call the view without context, or if desired, simply add it to the Views snap-in without problem.

The Host search and Service search (and other) views supplied as standard behave in a special way in relation to the filters. When you you select one of these views, it opens with a filter form, and then only shows hosts and services when this filter is activated.

Why is this so? Because it would simply be very impractical if you first had to go to All services, and then be forced to wait until several thousand services are displayed before you could filter the result with a search entry. This behavior is regulated by the Show data only on search option:

The data source is what you might call a table or database view under databases. Checkmk does not use SQL databases, but it is similarly-structured internally. In most cases you will be correct with All services or All hosts. There are however a few data sources that should be listed briefly here:

Here it can be decided whether your new view should have a context or if it will be a global view. The selections available to you depend on the data source. The most common context by far is host. The image above appears after selecting the All services data source.

Selecting Show information of a single Host defines that the new view describes one specific host. You will thus have created the basis for a view that is not globally-visible, but instead visible via a link:

There are two options for the service context type. If you select only Show information of a single Service, you can build a view that displays all services with the same name on different hosts. If it should be a specific service for a single host, then select in addition Show information of a single Host.

Why do some cells have a colored background? This points you to values lying outside the majority. This is actually not so meaningful for measured data, but there are, for instance, users with a specially-constructed matrix who can tell at a glance if an incorrect contact group has been entered for certain hosts or services.