VmBix 1.2.1 released

[tatapoum]

What vmbix does is that it consolidates real-time values other the specified interval. It's because if you query your vCenter for values other a larger interval, it returns historical values, aggregated at least every 5mn I think. So they are updated every 5mn. As Zabbix could need real-time data, we need to use the real-time interval. For VMware, the real-time interval is always 20s. So if you set a 60s interval in your vmbix query, vmbix will aggregate 3 values. The way it processes them depends on the performance counter stats type (delta, rate, etc) and the rollup type (average, min, max, etc). For example, for cpu.ready, the vm.counter.list methods returns :
12 cpu.ready ... (delta, summation)
which means that Vmbix will sum the collected values other the specified interval.

That's why you need to set the same check interval in the Zabbix item.

That being said, you should have approximately the same results in Zabbix and VMWare... I just see a slight difference in my setup.

Could you try to collect these perf metrics every 20s to verify if you obtain the same results than in the vCenter ?


The esx.memory and ex.cpu.load methods don't use VMware Performance Metrics but QuickStats :

https://www.vmware.com/support/developer/converter-sdk/conv51_apireference/vim.host.Summary.QuickStats.html

So the values are not the same than the memory or cpu counters in the Performance Tab of the vCenter view. It corresponds to the Summary tab values.
Again, I have the same results in my setup.

[Zabbix_User_1138]

I tried changing vmbix[esx.counter,{HOST.HOST},cpu.ready,,60] to vmbix[esx.counter,{HOST.HOST},cpu.ready,,20] and also changing the item interval from 60 sec to 20 sec, those changes made the item not update. Zabbix didn't flag the item as "not working" and I didn't see anything in the Zabbix or VmBix logs.
I then tried going from 20 to 30 sec on both settings and then the item started updating again, and now the values generated match those seen via vCenter.

I've ditched all the "esx.vms.memory" based metrics from the template and have switched to using the "esx.counter" based ones as they produce identical numbers in Zabbix when compared to vCenter.

[Zabbix_User_1138]

Discovered that 2 items, vmbix[esx.counter,{HOST.HOST},cpu.ready,,30] and vmbix[esx.counter,{HOST.HOST},cpu.costop,,30], out of 20 "vmbix[esx.counter" based items have stopped producing data during the weekend.
They didn't stop at the same time, but about 34 hours apart.
Strange thing is that "Last Check" for both items just contains a dash "-".
I had to look at the graph's for each item to see the latest reported data.

Have anyone else observed similar behavior ?

[tatapoum]

I've never seen this.

Do you get something when you query vmbix using zabbix_get ?

[Zabbix_User_1138]

I did a number of tests with different sampling intervals using zabbix_get.
I changed the interval on esx.counter["ESXiHOST",cpu.ready,,30] from 30 to 40, 35, 31, 32, 33, 34 and 60.
Only when using 60 as the interval did it return data every time.
Seems that the lower the interval, the more often it didn't return data.

Strange thing is, that now, after changing the Zabbix item to vmbix[esx.counter,{HOST.HOST},cpu.ready,,60] and the Zabbix item update-intervel to 60 sec. it produces near identical values in Zabbix, when compared to the values in vCenter.
Earlier on with the exact same intervals, the values in Zabbix was 2x those seen in vCenter.
I don't get it ?
For the record, I haven't changed anything anywhere, and neither Zabbix, VmBix, vCenter or the ESXi host has been restarted since I discovered the values, when using the 60 sec intervals, where 2x up in Zabbix.

I'm a little worried about when/if the 60 sec interval will stop working for the two esx.counter metrics in question.

[tatapoum]

As the minimum interval of the Perf counter API is 20s, you should use a multiple of 20s to get accurate results.
By the way, my setup always returns something with lower intervals. Maybe your vCenter is very loaded ?

[Zabbix_User_1138]

The vCenter is actually a vCenter Appliance (ver 5.5) used only for test purposes.
Only one ESXi host, with about 10 running VM's, is connected to the vCenter at the moment.

[gavind]

Would be also nice to know if someone can confirm about the interval.

[tbrew]

I was wondering if anyone had some advice as to why I cant get it run correctly. When I issue "vmbix" it runs correctly but when testing it I get the following:

[root@zabbix local]# vmbix -P 12051 -u user -p pass -s https://172.16.x.x/sdk
Exception in thread "main" java.lang.NullPointerException
at net.djarvur.vmbix.VmBix.main(Unknown Source)
Shutting down
Error disconnecting: null
Shutted down


It is pretty generic and I can not seem to find the reason behind it. Any ideas?

I am running CentOS 7

[Zabbix_User_1138]

Standalone ESXi host VmBix monitoring

I can't seem to get standalone ESXi monitoring to work.
None of the esx.counter metrics work, and other metrics like esx.status and esx.connection produce incorrect data.

When I grep for "vmbix" in /var/log/messages, I get "No host named < 'Hostname' > found"
In Zabbix host config I've tried every possible hostname I could think of: the host mgmt IP, localhost, 127.0.0.1, the name of the host, fqdn, even the ESXi UUID with both useuuid set to true and false in the vmbix.conf file.

Have anyone had success with monitoring a standalone ESXi host, i.e an ESXi host not connected to a vCenter ?

[tatapoum]

Hi,
vmbix was designed to work with a vCenter. Part of VMWare SDK is exposed by a vCenter, not ESXis.

[tbrew]

Hey tatapoum,

Do you have any ideas regarding my last post about the null exception error?

[Zabbix_User_1138]

VmBix Loadable Module Issue

I'm about to deploy my new VmBix based templates to monitor multiple vCenter servers (have started multiple vmbixd daemon's listing on different ports, using seperate vmbix.conf files)
Now I'm trying to monitor our 2nd vCenter, specifying the VmBix server IP and the specific port for the corresponding vmbixd service in the Zabbix host "Agent interfaces" section. Problem is, its not working, its connecting the the default VmBix loadable module port (12050) and that way connecting to the wrong vCenter server, (the test vcenter in this case)

To me it seems like the loadable module is ignoring the specified port and is using the default port, as if it was hardcoded in the vmbix.so module.

Forgot to add that I have tested the non-default-port vmbixd daemon from the Zabbix server with this string:

zabbix_get -s <IP> -p 12010 -k about[*]

and it works fine.

[tatapoum]

tbrew, Zabbix_User_1138, I'll try to look at it when I have time.

[tatapoum]

tbrew, I don't have much time to check but it may be related to a self-signed certificate issue. Could you try a :
curl -v -XGET https://172.16.x.y/sdk
and check if there is a certificate warning ?