Issue not alerting when proxy goes down?

[killmasta93]

Hi,
I was wondering if someone else has had this issue before, Currently I have zabbix proxy on Site B But when Site B goes down with the proxy i dont get an alert it just shows unknown. Currently running 5.0.2 Zabbix

I already tried these tutorials but didnt seem to work. not sure what i missed?

https://www.zabbix.com/forum/zabbix-troubleshooting-and-problems/48911-trigger-dependencies-zabbix-proxy-down

https://www.zabbix.com/forum/zabbix-help/398622-no-alert-when-zabbix-proxy-no-running

Thank you

[NilsA]

In my configuration my Zabbix server is monitoring each Zabbix proxy in the different networks. This is the workaround I have implemented for this problem.

Since the hosts of the remote network are monitored by the proxy, the server waits for information from the proxy. If the connection is down, no information can come in of course.
This makes sense, because the hosts might not be offline only because the connection to the remote network is down.

[killmasta93]

Thanks for the reply, the issue is that when the proxy goes down instead of alerting me it just shows hosts as unknown so not sure how you configured yours?

[NilsA]

I just apply the template "Template App Zabbix Proxy" and my own template for Zabbix agent hosts "Basic Monitoring Template" to it.

I added an item and a new trigger to the Zabbix Proxy application template to see how many processes are runnning of Zabbix Proxy (item key is "proc.num[zabbix_proxy]"). The trigger expression is "{proc.num[zabbix_proxy].last()}=0 or {proc.num[zabbix_proxy].nodata(3m)}=1".

My "Basic Monitoring Template" has a Zabbix Agent item with the key "agent.ping" and the trigger expression is "{agent.ping.nodata(2m)}=1".

So when a proxy loses connection or, for whatever reason, the service stops running, I will see it on my dashboard.
You might want to check out some other dashboard functions if you are still on the default one. I recommend at least one map with a "host group elements" element for each remote location and whatever else is most important for you.

[Hamardaban]

https://www.zabbix.com/documentation...types/internal

zabbix[proxy,<name>,<param>]

Information about Zabbix proxy.			Integer.	name: proxy name Valid values for param are: lastaccess - timestamp of last heart beat message received from proxy delay - how long collected values are unsent, calculated as "proxy delay" (difference between the current proxy time and the timestamp of the oldest unsent value on proxy) + ("current server time" - "proxy lastaccess") Example: ⇒ zabbix[proxy,"Germany",lastaccess] fuzzytime() trigger function can be used to check availability of proxies. This item is always p

[killmasta93]

I just apply the template "Template App Zabbix Proxy" and my own template for Zabbix agent hosts "Basic Monitoring Template" to it.

I added an item and a new trigger to the Zabbix Proxy application template to see how many processes are runnning of Zabbix Proxy (item key is "proc.num[zabbix_proxy]"). The trigger expression is "{proc.num[zabbix_proxy].last()}=0 or {proc.num[zabbix_proxy].nodata(3m)}=1".

My "Basic Monitoring Template" has a Zabbix Agent item with the key "agent.ping" and the trigger expression is "{agent.ping.nodata(2m)}=1".

So when a proxy loses connection or, for whatever reason, the service stops running, I will see it on my dashboard.
You might want to check out some other dashboard functions if you are still on the default one. I recommend at least one map with a "host group elements" element for each remote location and whatever else is most important for you.

Thanks for the reply is there a possibility you can attach the pictures? of the config? because i see other options like type units it be nice to see the picture of the config

[NilsA]

Here you go:

[jvedman]

After several failed attempts at getting this set up for myself, I thought I would share what I think is the correct solution from start to finish:

1. You have to create an item on a host. I suppose you could do that on your Zabbix server itself, but I created them on systems being monitored by each proxy I wanted alerts about:
Key Fields:
Name: [Customer] Zabbix Proxy Online
Type: Zabbix Internal
Key: zabbix[proxy,"proxyName",lastaccess]
Units: s [seconds]
Applications: General

2. After creating the item, create a trigger on the same host:
Key fields:
Severity: I set all of mine to high
Problem Expression: Use the Add and Select buttons to find the Item you just created
Use the Function drop-down to select fuzzytime
Last of (t): I set to 120 for most of mine, though I did get some false alerts for one location and used 180 there
Result: = 0
OK Event Generation: Recovery Expression:
Essentially the same as the Problem Expression except (make sure you pick fuzzytime again!):
Last of (t): 60
Result: = 1

3. Test by stopping the proxy service that you want to monitor.

After that, you should be able to easily create you action against that trigger.

I tried to attach images of the item and trigger, but I'm not sure if that worked or not.

Good luck!

[markfree]

This kind of item key doesn't seem ideal for Proxy availability check, but it works.
Collecting proxy data using zabbix internal key "zabbix[proxy,...,lastaccess]" results in a

timestamp of last heart beat message received

, not in proxy's last heartbeat age.

For example, "lastaccess" parameter collects this data:


If using an item's unit of seconds (s), those timestamp values are automatically converted to years, months and so on.


When using proxy key with "lastaccess" parameter, "fuzzytime()" function actually works and you are able to receive alerts when there's an unavailable proxy.
Something like this:

fuzzytime(/Zabbix Proxy/zabbix[proxy,"unique",delay],60)=1

But keep in mind that, custom event names and operational data might display a weird value.


If changing item key parameter to "delay", data is more consistent with actual proxy's heartbeat.
It seems that, when using "delay" parameter, values represent the real proxy's last seen time.


This way, the trigger should be changed like this:

last(/unique/zabbix[proxy,"{HOST.HOST}",delay],#1)>=60

And event values are more consistent as well.


I could not find any out-of-the-box template with some kind of "proxy.ping" item, though. Not sure why...