WEB-interface

[bmc]

hi all. I wonder if anyone do really think about rewriting standard zabbix web-interface to something more usefull, like DRUPAL or something.

Using some CMS would give a lot of pretty things like unified HTML-forms API, SQL-injection checking of data, etc. Do anyone have any suggestions or thoughts?

[elkor]

It certainly could be useful, although I've never been a fan of full-blown CMS systems (which I think is a bit of a misnomer anyway) as they tend to be a bit "busy" for my tastes.

I would really like to see some type of forum integration, ala phpbb or the like behind the triggers/hosts comments though. Integrated logins and persistant threads for things like this would simply be awesome.

Unfortunately, as much as I would really like to see this, I do not have the time to impliment it

so, hey, if you have the time I say go for it! there's plenty of room in the frontends directory for more choices

I know I'll certainly test it out if somebody took the time to do it.

[tvman]

I made my own display screens

I made my own display screens to fit my needs. As I understand If you've got enough brain power and time you can make it. But don't ask for support

I use jpgraph lib for my graphs and diagrams.
I showed my work to Alexey. He as a project manager desides what and how should be included in the official release.
Also in most cases it doesn't make sence to enforce security for this kind of systems because normaly it's MIS team internal stuff available from management network only. If you want to post some graphs or maps to the public web it could be just a periodicaly generated picture.

[bmc]

ok

ok, I see.

>Also in most cases it doesn't make sence to enforce security for this kind of systems

Security isn`t the purpose, there are a lot of features come with CMS "out of box". Using CMS could decrease complexity of web-interface and much more. Anyway - i got your point, thanks.

[tvman]

No limits

In open source community there is no limit. If you feel that something would be nice or useful why wait while somebody would implement that. Make it happen and than share it. Or don't share

[tvman]

Drupal CMS

By the way. I looked at Drupal. Nothing more useful then zabbix has currently. If you would look at zabbix frontend sourcecode you would see interface elements like forms, borders, footers. Stuff which could be taken from CMS and applyed to frontend. Any way you have to implement actual functionality as original code which couldn't be included in CMS and ready to go. I agree that there is the way to improve zabbix frontend.for example when thre are 200 host it's not easy to navigate though. But it's hard to support more than one type of interface for support team. The apearence is not an issue. The issue is the interface procedures. And it's definition is very complicated task. I would say that we should give this ideas to Alexei and his team and thay will take care about programming part.

[bmc]

>By the way. I looked at Drupal. Nothing more useful then zabbix has currently.

Actually it is more usefull. There are several features:

1) AAA feature. You can easily create users, groups, grant them any combination of permitions you`d like. And any element (like queue, graph, host, etc.) define it`s own permittion list (for example "create", "edit", "view", "delete", and so on).

2) Using HTML form. In drupal you create one form for editing/creating element. You can validate input data (in special function) and page will print out error and ask for correction of data. It`s created by 2 functions 1 - element_form() and element_validate(), that`s it, no overhead.

3) Separation of code and presentation. There is no HTML code through the module. All HTML created in the end, right before you give HTML to the client. BTW - you can use skins

4) Structure. All elements defined as one module. In this you define functions you want to be allowed with it. You create element`s mother-element, and so on. There is no mess in code.

5) Menus. As a fact of good structure, you can create menus for any element in any way you like. It`s simple and could be defined in one place.

All this features make possible to write clear, structured code. Right now interface isn`t intuitive. It`s not handy eather.

>The apearence is not an issue.The issue is the interface procedures. And it's definition is very complicated task.

True. My opinion is migrating to Drupal`s style of products will solve most of problems in architecture of frontend. Therefore there are a lot of code allready produced, so the thing is to reorganize the code.

[tvman]

network map tool

By the way who does know any drag-and-drop web based graphics editor or lib. Because I'm using a lot of network maps and it's creating or editing is a pain in the neck. I've channed this part to make links more thik and visible, added whole bunch of new icons. Unfortunately I'm out of picture because I did a lot of changes in the core code and cannot follow the product mainstream.

[elkor]

I've also made signifigant changes to the core code, upgrading to a new version is at best a day-long project of going through and applying my diffs from the last version so I understand what you mean.

That being said, if you take the time and plan out your graphics it can make the network map functionality much easier. For example, I added a few new entries to the icon label type definitions, based on these entries I adjust the alignment grid in the map configuration setting, what is displayed, and the font size. By sizing icons that fit within each grid square and aligning my background images to center and line up to the grid it can make new map creation *relatively* painless.

for example, the default grid is set up at 50px by 50px. I have icons created that fit within this boundry and have adjusted the code to put the hostname beneath in a smaller font with no status info. each individual host fits in the 50px square and the background lines up, so creating a map is as simple as putting hosts at 50,50/100,50/150,50/50,100/100,100/150,100. There is a similar setup for 75px icons.
It works well but did take time to plan and set up, a drag and drop editor for this would be fantastic but it sounds like a TON of work and I wouldn't even know where to begin implimenting something like that.

[James Wells]

Greetings,

Personally, I would be more in favor of moving the frontend to something like PHP-Smarty, than I would to any one particular CMS implementation.

The reasoning for this is simple, you would like to use Drupal, but I would like to use Xoops, Elkor would like to use something else and a friend of mine wants to make it a Zope module.

By using something like Smarty, it can remain a standalone package, that will run on almost any [LB]AMP server, while at the same time gaining the ability to be merged into almost any CMS you want with minimal work.

To be honest though, the biggest issue you will be faced with, if you want to take on some project like this, is that in the current frontend, the logic and the UI are mixed together. To port something like Zabbix to any CMS, requires almost a complete UI rewrite. By moving to PHP-Smarty, or any of the PHP template packages, you are forced to seperate the UI and logic, at which point it becomes almost trivial to integrate into most CMS packages.

[tvman]

I lile Smarty

To be honest I lile Smarty. I feels very natural when navigating through. And about maps I also have ajusted grid and icons and I've added facke type of objects kind fo "turn points" to make not just straight links between hosts. Which doesn't help when boss asks to strache or squize the map and take out or add some number of icons and links. I do it through UPDATE query directly but true screen editor would be mutch better. I know there are Java applets which allow to do similar job. It's just two layers, background and icons on the foreground. If somebody use it please let me know.