Call for discussion on ZABBIX 1.8

[ptader]

+1 for a maintenance module (with scheduling please) for groups, host, media all of Zabbix.

For example, going through a site wide downtime. On a lesser scale, your taking down a node for an hour or two (patches, relocating, etc) you initially mark it "not Monitored" but forget to change it when the outage is complete. The ability to mark it down for an hour would help prevent this problem.

-pt

[BHG_2008]

Maintenance Window

One of the competitors, Orion, has the "maintenance window" feature.
It would be a most welcome edition to the ZABBIX arsenal of features.
Again, an APM would be most useful.

[multazim]

APM would make Zabbix stand out

[BHG_2008]

Calculated Items

Since at least 2007 many people have asked this feature...
ex :

http://www.zabbix.com/forum/showthread.php?t=8679&highlight=calculated+item

http://www.zabbix.com/forum/showthread.php?t=1368&highlight=calculated+item

http://www.zabbix.com/forum/showthread.php?t=9071&highlight=math+item

http://www.zabbix.com/forum/showthread.php?t=5951&highlight=math+items

For example i just want to graph the used memory... today noway to have an item like this : vm.memory.size[total]-vm.memory.size[available]

Now, we can just do math... on triggers... the only math permitted on an item is a multiplier...

There's many example where you want to have an item calculated...

please add such functionnality and i will convert my IT to zabbix!

Second motion! This would increase flexibility immensely.

[otheus]

Suggested improvements to Zabbix (1.8 or even sooner!)

• Improved charts:
  • Better handling of Y-axis when more than 2 items on a single chart
  • Log-scale charts
• Built-in DNS check via "Simple Checks" to see if a DNS server responds to a simple request.
• zabbix_sender modules for Perl, and maybe Python and Java. (instead of an API).
• zabbix_sender program reads from standard input.
• logCount check (for agent) which operates like the log item, but returns the number of matching lines instead of the text themselves.
• More aggregate functions, such as "abs" and "diff" (between items in a 2-member group) or maxabs and maxdiff (between highest and lowest in an N-member group)
• "Virtual" Hosts -- not directly monitored, but may contain aggregate and internal Zabbix items and triggers of any sort.
• Mass acknowledgement -- one summary text for several different alarms that has one underlying cause (but not necessarily dependent on each other).

As I find time, I will try to provide patches for these.

Thanks for the great work, Zabbix guys!

[froztbyte]

Not entirely sure if this has been raised yet since I haven't finished reading the whole thread, but the following functionality is what I'd like to see in a new release:

* An ability to only utilize a single contact method for a user when sending a message out to a user group (think webmasters.cellphone). This is especially useful if you have various sort of Disaster alerts which a given user needs to receive all of, but only specific alerts at specific times.
* Sounds available on the dashboard.

[mbm]

Suggested improvements to Zabbix (1.8 or even sooner!)
...

• zabbix_sender program reads from standard input.

...

As I find time, I will try to provide patches for these.

Thanks for the great work, Zabbix guys!

This particular one can be done in about 10 lines.

In zabbix_sender.c:
int main(int argc, char **argv)

FILE *in; => FILE *in = NULL;

if( INPUT_FILE ) (still in main)

Change:
if (NULL == (in = fopen(INPUT_FILE, "r"))

To:
if (0 == strcmp(INPUT_FILE, "-"))
{
if(NULL == (in = fdopen(0, "r"))
{
zabbix_log(LOG_LEVEL_WARNING, "Cannot reopen stdin [%s]", strerror(errno));
return FAIL;
}
}
if ( NULL == in && NULL == (in = fopen(INPUT_FILE, "r")))

A word of caution. If you use the ZABBIX sender and the input file, don't believe the usage message or the output from --help, they both lie.

[Kerrygeek]

I mentioned one of these in another part of the forum but here are my 2 requests:

1. Alert sounds on the dashboard screen so you can see/hear everything in one place.

2. Under the Configuration/actions, make dropdowns to look at the actions by groups like in the other parts of configuration. When you get a lot of actions in there it would make them easier to find.

My Zabbix server is in production, running parallel with my old system for a few more days but it looks like I can shut down the old system by next week. Zabbix seems to be working great for us!

Thanks,
Kerry

[froztbyte]

After some browsing around it doesn't seem like this functionality is available in the current capabilities of Zabbix, so I'll add it here as a feature request:

The ability to make a trigger event persistently stay in the dashboard until it gets acknowledged, instead of disappearing from the events when the problem gets resolved and stays resolved.

Update: also the ability to specify user preferences for the dashboard. If a user normally views a master node with subnodes then logs out, that should be the view when he logs back in.

[froztbyte]

Perhaps also the ability to save specific filters when viewing triggers? This is helpful if you've got multiple host groups where you want to see different levels/combinations of information. Then you can just select the appropriate trigger in much the same way as the "Favourite Graphs/Screens/Maps" areas in the dashboard work.

[BlackRabbit]

Add an option to remove the legend from a graph.
(on some graphs this legend is ~the same as the graph's title, so they take up unnecessary space)


Export/clone screens.
eg: I have 30+ identical switches, for each I want a detailed screen showing ports-stats etc. Now (1.6.1) I have to create a new screen for each switch, with each screen containing 30-50 graphs.

[niranjan]

Zabbix 1.8 - Reports

Hi Alexei

After a long time I came back to zabbix and its gr8 to see the work is going on v1.8 but still I dont see reporting in the roadmap.

What a tool will be in use if it cannot produce reports. I cannot ask my management to open up their laptops and browse through the realtime or historical online reports they may even not be able to understand.

I know its too late as in a week's time 1.8 will be released but there should be work on reporting part (pdf generation or any such tool). this can also be released as an addon or patch.

Well overall awaiting to see v1.8

regards
Niranjan

[MarkusL]

Hi all!

1.) Extended language-support
At the moment you can choose different languages for the gui.
But items and triggers are all the same.

It would be fine, if items and triggers could also be written in different languages. Therefor it would be great to copy all item / trigger / action-names from one language to another one and than translate it.
The language-setting of the user defines in which language he get´s his COMPLETE zabbix-frontend and his actions (SMS, mails, etc.).


2.) Applications
We often ran into trouble when two or more templates have the same application-name (f.e. 000-Information, 010-Performance, 020-Monitoring, etc.). We were sometimes not able to put a new template to a host with existing templates where application-names overlap on the templates.
When we installed zabbix our consultant (also here in the forum) noticed this effect, too. I hope he has in mind the exact problem. As I can remember it was a database-related problem.


3.) Dashboard
It would be fine, if you could setup if triggers that have been acknowledged were not shown at the dashboard.
Therefor it would be great to have this setting in "user group", "user".
My wish is to have the following settings:
- do not show acknowledged triggers at all
- do not show them for X days
Another way would be to define a filter at the dashboard,...

Example: a machine has poor memory. I order memory and install it. Takes about 5 days from reporting to have installed and finished. Within these days I always see the fired trigger "memory to poor". By setting the "view-acknowledged-triggers" to "not for the next 14 days", I have a clean dashboard.


4.) Dashboard
Another thing we would like to have is the possibility to filter the dashboard by severity.
First thing is to have the possibility to blend out f.e. Not classified in the System status-table.
Second thing is to blend out f.e. "Not classified"-issues in the issue-list.

Why: we use "Not classified" ONLY for "nodata"-Triggers. Most of these triggers are developing-related triggers, f.e. a mistake in a batchjob or something. For our helpdesk it is really anoying to see all those triggers fired. It´s only important for our developing.


5.) Massupdate
it would be fine, if we could use regexp to massupdate items and triggers (f.e. names, keys, trigger-expressions). At the moment we have to export it, modify by hand in notepad++ or something and reimport it.


6.) Documentation
We would love to see some more detailled documentation. Especially trigger-expressions.


7.) Host Configuration / Profile
We would like to have a free editable tab for the profile. At the moment zabbix gives hard-coded possibilities like "Device Type", "Name", "OS", etc. Maybe it is possible to define them by ourself.
We would love to integrate more documentation of our systems in zabbix. Some required fields don´t match any given fields, so we have to store the data in the "Notes"-field,...

Another thing is, that we want to fill these fields by other applications. F.e. some scripts that post information of a system to the profile-fields. Therefor it would be great to have sql-access to the profile-fields (in a clean, documented and supported zabbix-environtment, thats the point; with some "tricks", it´s already possible,...)


8.) Template / Trigger / Items
We would love to see a comment-field in Templates, Triggers and Items (and maybe in all "sections"). We are dealing with many templates. In some templates you have to know something before connecting it to a host. It would really be great to have a comment-field where I can place some handling-instructions.



Kind regards,

Markus.

[sarathyme]

Node Sync only for specified Hosts (distributed Setup)

If it is not too much to ask, please do Node Sync only for specified Hosts (distributed setup)

Currently I believe it does it for all hosts of a child node, may be if I can have fine grained control on what hosts to be replicated on master node that would be good to have (for me at least)

[Wolfgang]

2.) Applications
We often ran into trouble when two or more templates have the same application-name (f.e. 000-Information, 010-Performance, 020-Monitoring, etc.). We were sometimes not able to put a new template to a host with existing templates where application-names overlap on the templates.

I 2nd this request. From my understanding these "conflicts" are basically by design as it is now.
However it makes much sense to group items/triggers in templates so the show up together in the same application on the host they are linked to. This is/would be the intutive way to use them.

Wolfgang

[Silvery]

I would like to have more flexibility with escalations.

It would be nice if you could define multiple independent operations for one Action. Because it's hard to build operations as desired when the steps and periods of the different operations affect each other...