Call for discussion on ZABBIX 1.8

[wlshanquan]

Add device's syslog available on WEB page.

Add device's syslog available on WEB page.
It will be very usefull!
A system is composed of more than one ITEM，Tied More than one ITEM for monitoring, monitor the IT item from the Point of view of Business.

[akatsuki]

- Parameters allowed meta character to keys.

- Free words or HTML codes inserting on screen and Map.

- Csv data downloading

- Changing alert|waring sound

- SNMP Trap to Media type

Many thanks for great tool developing.
regards.

[ronnyromero]

GoogleMaps usage for the screen maps
packaged version for centos and ubuntu (with all the required dependencies link)

+ dont forget v10 IPFIX (netflow)...like scrutinizer and netflowtracker

[krimson]

- Item calculation. E.g. item3 = item2 / item1 * 100

[Fatima R R Tonon]

Re: IT SERVICES for an specific period

I have modified IT services code to display SLA for last N days (instead of hardcoded 7 days) and it works. It is currently an "ugly hack" that needs some polishing before being published but yes, it seems quite easy to produce an SLA for a specific period of time.

Yes, we implemented a patch that was released in the Zabbix forum for v.1.4.4 of Zabbix and adapt to version 1.6.1. But our customer asks reports of availability between periods of 1 or several days.

[Calimero]

WEB Monitoring

Here are a a couple of requests regarding Web Monitoring:

1/ Currently, zabbix can check for each step of a scenario what HTTP response code it got (typically: 200) and check whether a word/regexp is found in content. If HTTP code is OK and word is found ==> zabbix considers step successful.
I'd like to have an extra regexp that must not be found. Typically if words like 'error' or 'connection failure' are found ==> consider step failed.


2/ Being able to give Zabbix the HTTP Host: header (www.mycompany.com), Path (/index.php) and IP/port (192.168.10.10:80) instead of just a full URL.


If you're using CDNs like Akamaï. All/most your real URLs will point to akamaï's servers.

www.mycompany.com ==> somethingXXX.akamai.com

So when you tell Zabbix to check "www.mycompany.com" you're actually checking whether akamai are doing their job. Which is fine (you check what people really see).

At the same time, you may want to check your (source) servers. Akamai has a feature that will (try to) hide failures: if your server (or pool) doesn't answer, Akamai will serve the previous cached page. Would be useful to find out that your servers have crashed anyway.

To check your own servers you have to cheat and pretend www.mycompany.com points to A.B.C.D (real IP of your server, or virtual IP of the pool on your load balancer).
You could cheat with /etc/hosts on zabbix server, but that way you'd only be able to check either Akamai or your hosts. Not both at the same time.

Another option is to have different URLs: www.mycompany.com for akamai, www-original.mycompany.com for your servers.
Not really easy to maintain. Can become a real pain if your site has lots of URL rewriting that will always take you to back "www.mycompany.com".

So far, I'm using a custom script (run as a UserParameter on my zabbix server's agent). Works but direct integration in Zabbix would be better.

I'll try to look at the code. libcurl can easily be configured to "cheat". Most work would probably come from the UI, DB schemas changes to store extra params, ...

[flexguy]

really big things

For easier integration an API would be great. Better reporting facilities will improve acceptance by the management.

Some new things?
- Asset tracking
- Software deployment
- Remote access integrated (VNC)
- Additional graphical elements on maps (not related to any host, for instance clouds, arrows, lines, borders)
- Some information from a host profile made visible in dashboard, graphs, maps, screens

[cstackpole]

The Call for Discussion on 1.8 should be a sub-forum rather than a thread, because I see a lot of interesting ideas being raised that result in tangential discussions with further interesting ideas in them; these could easily get lost in a single thread. With a sub-forum, each idea could be its own thread, and related ideas could be kept together.

I second this!

[mlange]

I really would like to see this feature in 1.8:

http://www.zabbix.com/forum/showthread.php?t=10701&highlight=duplicate+triggers

[Neurotox]

1.8 requeste

I'm quite new with Zabbix so maybe some of my suggestion are already include, sorry if it's the case.

General idea: Easynest of use (speed). Zabbix is a great product but many step are time consuming. For example adding a switch, creating screen and graph for every server, ...

Specific idea:
-SNMP (+1 for better SNMP support (Cacti do a great job with SNMP))
--ifDesc and other value from snmp in graph title
--SNMP discovery instead of doing template, something like you look at .iso.org.dod.internet.mgmt.mib-2.interfaces.ifTable.ifEntry.ifIndex or other OID and you can chose which interface you want to monitor (and chose to create graph or not)

-Screen
--Screen template
--Creating screen in text mode
---Example: you want to create a screen with 48 graph. Right now, you have to chose "change", then the page reload, you click select, you chose the graph, you click "save", then the page reload... At every time the page reload it also reloading all graph creating unwanted load on the server and it a time consuming process.
Instead if we could do it in text mode by entering something like: "graph:host:graph_name". It will probably take something like 60 secondes instead of many minutes. Or at least the ability to not show the graph when you configure the screen.

-Discovery for agent
--When you add a new host, it will be nice if it auto detec the number of CPU, drive, network interfaces, ... and let you choose which one to monitor, template are nice but only if you all have the same machine all around. For example, you don't really need to monitor the free space of a cd-rom on a Server.

-Graph
--Graph template linked to a item instead of a host
---The ability to create a graph in one click for and items. (if you have a template, you could just click on a button that will create graph for this item)

Thanks for Zabbix, it's a really nice product.

[richlv]

The Call for Discussion on 1.8 should be a sub-forum rather than a thread...

i'd go further and say that feature suggestions should be gathered & polished on the wiki.
ok, i went that far (i'll probably regret mentioning it here, but whatever ) and attacked http://www.zabbix.com/wiki/doku.php?id=contrib:ideas

so, the idea here is - gather feature ideas on the wiki. as mentioned there, describe each feature as much as possible, so that if a company/individual comes along and wants to fund/develop a feature from that list, they can just grab that spec & get to the work.

smaller descriptions go on that page, larger go on their own pages with single sentence & link on the main page.
linking to outside crucial information (including the forum) should be discouraged - if the idea originates in a forum thread, take that, polish & tuck on the wiki. if the idea is being discussed on a thread, it is ok (and should be done) to link to this thread - but any consensus or important ideas from this thread should migrate to the wiki page.

if you see a request you made somewhere on the forum, irc or whatever - yeas, it is quite likely i lifted it from there (on the other hand, there have been cases when a request on my list has appeared in forums later)

what information should be added there ?
so a very detailed spec could help. that includes current location in gui (this helps to understand the description; of possible, should include version where this location is valid - sometimes they change between versions).
if possible, scope should be evaluated. this includes amount of work in case funding is considered, and possibility for the change to go inside stable branch (larger changes go on trunk, smaller improvements on stable etc).
implemented stuff walks over to 'implemented' section.

during creation of this list i've had to fight some wiki problems.

1. numbered headings are a must. as i understand, this is possible with a plugin;
2. edit button at all heading levels would helpalot;
3. edit button moved to the top of the section. possible with a plugin;
4. i am unable to create new pages currently. this means "hold your horses", i have api page structure semi-written, once we can create new pages, i'll create it (same goes for cli interface).

how can you help ? you can move completed stuff i have missed to completed section (only fully implemented - if parts are there, it stays where it is & gets that mentioned), fix incorrect information (surely i have created such), fix language mistakes (me bad english).

while this is wiki and all, i'd appreciate not running all amok the article/section. i've spent quite some time fighting this, so i'll be protective for some time period ;D

in the end feature suggestion threads for next version could only mention numbers like in a joke about prison where guys sit around a table, one calls out a number, other laugh ;>

or it could be that such threads are simply converted into a vote for the listed features, and new features are indeed discussed in a new forum (but not only when preparing for a new version, but constantly).

of course, seeing the feature listed there should not get you excited. some require huge amounts of work, some might be incorrect/unimplementable. amount of work the ones already there easily could be for years to come...

[rafael]

The ability to monitor VM systems and resources.

[exkg]

Hi,


Do u know 'BBTray' (http://www.deadcat.net/viewfile.php?fileid=233) its a SysTray tool that open a URL (bb2.html) and cheks if exists some trouble in hosts ... its play a sound and popup a window to click and go to monitoring page.


I wish: ZBXTray
It can open 'tr_status.php' (with parameters), search for a problem and show it.



[]s,
Luciano

[gullevek]

1.8 Suggestions

I am using Zabbix now for less than a month, but in the long run I want it to be my only monitoring system. Right now I run munin & hotsanic & some self written perl scripts too.

I especially need visual graphs to see the daily, weekly and monthly trend. And still the Zabbix graphs are nice, but far from ideal as they do not work like rrdgraphs. So I hope this will be updated (perhaps not in 1.8 directly).

My other things are:
- SNMP community string per host and not per item/template
- additional values in the zabbix agent, because I think the currently collected are a bit sparse
* more detailed memory data (like munin does), at least free memory
* detailed IRQ stats
* CPU: collect iowait, irq, etc stats too
* add stopped process to the process collection
* netstats (active, passive, etc)
* swap io and disk io with bytes, etc for linux
* fork rate, entropy, context switches for linux
- add sum data flag, so you can have eg traffic per hour, day, week, etc. without this, the data is kind of useless
- automatic dedection of disks + partitions and setup of proper monitoring values
- some plugin/api to easy write special collectors, eg for bind, postgres, nfs server/client or other things in various languages (C, perl, python, ruby, shell script, ...)
- if [ALL] is used, use the color of the line to draw max/min to make it easier to distinguish if there are several lines in the graph
- reports for week/month of service stats and problems: what problem, how long was it a problem, etc

[gullevek]

-Graph
--Graph template linked to a item instead of a host
---The ability to create a graph in one click for and items. (if you have a template, you could just click on a button that will create graph for this item)

Each Item has a graph already. If you go to the Lastest data, you can open a graph for each item there.

But I think what would be cool to easy select several items, generate a graph and then say, store this graph for this host/template.