Call for discussion on ZABBIX 1.8

[sbadger]

Text feilds in the screens

I would like to be able to put text labels in the screens to be able to title the screen. This will allow you create a label for the screen which would be handy when using them in slide shows.

[exkg]

Sugestions for 1.8

Something like a odometer ro show:

- hosts and group status;
- SLAs status;
- Triggers status;
- availability reports;
- etc etc



[]s,
Luciano

[troffasky]

SVG for Maps

The current bitmapped Map screens should be replaced with a vector format, preferably SVG. This page has a lot of examples of what can be done with SVG:

http://www.carto.net/papers/svg/samples/

[lamont]

This is both 1.8 suggestions and 1.6 feedback on graphs.

The feature in 1.6 to highly a map area on the graph to zoom is outanding and saves me a lot of time.

Unfortunately, the new sliders at the bottom of the graph and the popup menu in the lower left is nearly uniformly hated by everyone at my workplace and viewed as a major step backwards in Zabbix UI. That thing needs to get fixed or reverted.

Sorry, you win one and you lose one.

Having the ability to modify the y-axis scale of any graph I'm viewing would also be a huge improvement for 1.8

It'd also be good to allow user-private graphs and screens, and make them shareable. The zabbix admins at work do not have the time to make all the graphs that we need to, and that work needs to get decentralized out to the broader user community.

[troffasky]

Unfortunately, the new sliders at the bottom of the graph and the popup menu in the lower left is nearly uniformly hated by everyone at my workplace and viewed as a major step backwards in Zabbix UI. That thing needs to get fixed or reverted.

There's enough empty space in the new graph slider to fit the old controls in, so both could be had :-)

[phoemix]

Hello,

currently adding SNMP devices is not so comfortable, it's a lot of work. for myself, i've got my own template-generator to overcome this issue, but this is a workaround, not a solution. so, i'd like to vote for highly improved SNMP handling. Zabbix should take note of snmp index, should know what oids are to monitor on a switch, make automatic triggers, traffic graphs (both pps and bps), and so on.

also, having item references in graph and trigger titles/descriptions/anywhere would be very nice. currently if the description of an interface, or any SNMP-monitored thingie changes, all the triggers, graphs, what are using that thingie has to be renamed by hand, which is just a pain.

maintenance planning would also be nice to have. defining maintenance windows in which the given category of devices shouldn't fire triggers, because there's some work going on, is also very handy.

even more, some kind of indication how much zabbix is lagging behind with polling items would be nice, that'd indicate how much is the monitoring system is overloaded, and needs actual tuning.

so far, I have these ideas, thank you for listening

A few more ideas:

zabbix should be able to interpret item values, especially for SNMP devices. MIBs have the information for this. Also it'd be nice, if this could be specified in the template for the item, like "<values><value data="3">Signal Lost</value>...</values>".
this would help a lot at the notifications, because administrators would know what's the error value of the item.

being able to define the hostup status by hand for a host would be very nice. maybe being able to specify a severity for the host would also be very useful (a desktop is down: nobody cares. a core router is down: heads up!)

[DoubleP]

Latest Data Enhancement

I would like to be able to go to latest data and have the ability to "check" multiple items in latest data. Then be able to generate a temporary screen with the selected latest data items.

[Alexei]

Thanks for your active participation in the discussion!

In order to make all feature requests logged and well maintained, we introduced a new project in Zabbix Support System at https://support.zabbix.com/browse/ZBXNEXT

Please spend some time registering your ideas there. Do not forget to vote for your favourites!

[bek99]

kill -HUP to re-read configs

I don't know if this was posted, but perhaps a kill -HUP to re-read configs? it might require a master zabbix process for instance or just using killall (pkill for solaris) -HUP. start/stop can be a bit of a pain.

[Farley]

+1 for API... but...

Don't know if I'm too late chiming in here...

But, I'd say +10000 for API from me... but in the referred link discussing API features...

http://www.zabbix.com/forum/showthread.php?t=10251

I see no mention of features that I would use immediately. The type of API I am referring to/interested in is basically simple read-only export of important data. No calculation, no control, no "writing or modification" of data.

For example, the triggers (that have triggered, the trigger log) could be just as easily a RSS feed, which I could check. And/or it could be put alternatively into JSON for easy consumption/parsing by any language (I'm looking for this to use in javascript for a nice simple web 2.0 monitoring interface, and for consuming this in Objective-C for a iPhone zabbix app).

In addition to an export of triggers, I would be interested in a API/mechanism to grab the data in a simply consumable format for viewing with a device. Preferably _very_ well designed, so I can give it any number of parameters in a simple request format, either in a URL scheme, or in a XML request, or something. Examples of data chunks I would like to be able to retrieve as a web service would be...

"CPU Load Percentage of all servers in group x"
or
"Free disk space on all servers"
or
"Last 10 1-minute CPU Load values on server x"

Of course, to accomplish being able to return that data intelligently, it would need to be a flexible format to return this data (xml/json/etc). So, if I asked for the load percentage of all servers in group x, it would return a array of servers, and in each individual server array would contain a key->val of my key name (cpuload) and value. Alternatively/additionally, to be able to accomplish the last item, it would mean that the xml would need an additional property to each value which is the timestamp of the retrieve values. So for example to accomplish this last item I would need the xml (in very simple/non-realistic xml) as follows...

<zabbix_api q="Last 10 1-minute CPU Load values on server x">
<server hostname="x" ip_address="x.x.x.x" ...>
<item name="cpuload-1" timestamp="1234567890">10.4</item>
<item name="cpuload-1" timestamp="1234567891">11.2</item>
<item name="cpuload-1" timestamp="1234567892">8.3</item>
<item name="cpuload-1" timestamp="1234567893">3.2</item>
<item name="cpuload-1" timestamp="1234567894">9.9</item>
<item name="cpuload-1" timestamp="1234567895">15.2</item>
<item name="cpuload-1" timestamp="1234567896">12.2</item>
<item name="cpuload-1" timestamp="1234567897">10.0</item>
<item name="cpuload-1" timestamp="1234567898">33.0</item>
<item name="cpuload-1" timestamp="1234567899">41.0</item>
</server>
</zabbix_api>

That would allow all of us the ability to extend and abstract zabbix into a million different ways. Like I said, I want a personal control panel running on my machine which remotely grabs data from all my zabbix servers (multiple clients with zabbix servers) and aggregates them intelligently on my computer. And I want to make a iPhone app for Zabbix, which doesn't really make sense without a service to consume.

Cheers!

[Garo]

Full hostname

Allow agent to use fully qualified hostname when it talks to server (in unix use 'hostname -f' instead of 'hostname')

There's now a problem when auto discovery registers a new agent because it uses its fully qualified hostname (like 'machine.domain.com' instead of 'machine') but the agent itself tries to talk with the server with its plain hostname without the domain.

Agent using fully qualified hostname would fix this problem.

- Garo

[dagun]

Add a new function--auto network topology

If you can add a new function --"auto network topology" to the zabbix.Now I am working it ,but I have not success.I think it is a good idea.
Come on ,everyone.Do it.

[bek99]

if they weren't suggested.

1. history of who added/changed an item/trigger (running history even better, perhaps even command history of previous checks to see their evolution) for zabbix server

2. maybe in the logs for queries on the servers, attach an item_id # in the log files so full debug doesn't need to be turned on, but can see what checks are doing what.. it might help in trying to track down sigpipe errors..

[Andreas Bollhalder]

if they weren't suggested.

1. history of who added/changed an item/trigger (running history even better, perhaps even command history of previous checks to see their evolution) for zabbix server

You remember me of an old idea. This and other information (acknowledges, manual annotations etc.) could be made visible in the graphs with markers and a legend. So you can see on the time line what has changed/happened.

Andreas

[russhenry715]

i have been using zabbix for a few months and have been fixing alot of issues before our end users see the problems, there are a few things i would to see added

* Ability to change up the top menu dynamic -> able to create new menus and sub menus and have those items either open a local php page or a external link

*A better device view page (www.netmon.ca) currently if you are monitoring alot of items on a page it gets messy. I would like to click on a device and see all the information for that device

*Automatic updates for zabbix and agents

*Builtin reports that can by dynamic created by the users

*Dynamic Network Map


We have tried alot of network monitors and by far this is the best one out there