icmping and Hypervisor ping reporting unavailable but they are available.

[javiers]

Hi guys:

As described my problem is:

Zabbix 6.0.11 is reporting my hypervisors is down but it isnt, it collects perfectly other metrics.

The hypervisor is monitored by a remote proxy (6.0.11), fresh install (ubuntu 22.04), fping permissions and path in the proxy are correct.
From the proxy i can ping the hypervisor and its VMs.

Im kinda lost since i have 4 others proxies which i installed the same way and work perfectly.


Any suggestions would be apreciated.

Thanks

[Markku]

What do zabbix_server.log and zabbix_proxy.log say?

You can run tcpdump on the proxy to verify if the pings are really leaving the server:

sudo apt install tcpdump
sudo tcpdump -n icmp and host ip-of-hypervisor

Also check the same tcpdump when pinging the hypervisor manually.

Markku

[javiers]

Hi Markku, logs seem correct as they not say anything about icmp ping, they just say started icmp pinger which i think is normal behaivour,

Regarding tcp dump also seems fine while pinging hypervisor from proxy (check screenshots).

On proxy tab on frontend shows porxy on.

So i dont know what could be failing since on others installations i've made just works like a charm.

Again thanks for your time, if you have another suggestion or you see something strange in the screenshots letme know

Thanks.

[Markku]

So do you mean that your own manual pings are shown in tcpdump, but not the pings that the Zabbix proxy should do?

Markku

[javiers]

Yes, thanks Markku for your fast reply, manual pings are shown but nothing else, if i stop manual pinging tcpdump does not capture anything.

So it seems like the proxy is not preforming the pings even though icmp pingger is started.

Permissions for fping as well as the path seem to be correct in the zabbix proxy conf so i dont know what could it be.

[Markku]

How is the host interface configured for Vives 91 in Zabbix?

Markku

[Markku]

Also a comment: on a normally-installed Zabbix implementation on Ubuntu you shouldn't need to check for pinger processes or fping permissions or its location It just works. So your comments raise a question that do you have a specific reason to believe that your fping is somehow faulty? Just checking

Markku

[javiers]

Since the icmp wasnt working after fresh install, i did some research and people on similar posts were asking to check for fping permissions and its location so i just assumed something had to be wrong.

In zabbix host interface is configured the same way i have all my other hypervisors which work perfectly:

[Markku]

Let me first open up the ICMP ping case a bit more.

The VMware items don't use the configured host interface at all. They use the macros that specify what is the URL to the host or vCenter. Right?

In contrast, the ICMP ping item uses the host interface configuration to find out what is the IP address or DNS name to ping.

If the host interface IP address/DNS name is missing or is incorrect, your VMware items will work but ICMP ping items won't.

Now, I asked you how the host interface was configured, and you showed me a screen where there is no IP address or DNS name at all. We already know (based on your earlier posts) that the IP address of the host is 192.168.1.91. Did you remove it from the screenshot or is your host incorrectly configured?

In zabbix host interface is configured the same way i have all my other hypervisors which work perfectly:

I'm confused. In your initial post you showed a screenshot that showed two hosts having the same problems, but now you say Vives 91 is the one having problems. What is the real case?

In your first post you also said:

Im kinda lost since i have 4 others proxies which i installed the same way and work perfectly.

Are you saying that all hosts on all other proxies work fine, but none of the hosts on this one proxy work at all?

Markku

[javiers]

Let me first open up the ICMP ping case a bit more.

The VMware items don't use the configured host interface at all. They use the macros that specify what is the URL to the host or vCenter. Right?

In contrast, the ICMP ping item uses the host interface configuration to find out what is the IP address or DNS name to ping.

If the host interface IP address/DNS name is missing or is incorrect, your VMware items will work but ICMP ping items won't.

Now, I asked you how the host interface was configured, and you showed me a screen where there is no IP address or DNS name at all. We already know (based on your earlier posts) that the IP address of the host is 192.168.1.91. Did you remove it from the screenshot or is your host incorrectly configured?

Yes i did remove the ip address, but yes, it points to my hypervisor 192.168.1.91 which runs esxi 6.7

I'm confused. In your initial post you showed a screenshot that showed two hosts having the same problems, but now you say Vives 91 is the one having problems. What is the real case?

Yes it is happenig on both hosts monitored by this proxy, i only showed Vives 91 in recent posts in order to reduce duplicated information.

In your first post you also said:

Are you saying that all hosts on all other proxies work fine, but none of the hosts on this one proxy work at all?

Exactly, all hosts behind my others proxies work just fine, but these 2 hosts fail.


Javier

[Markku]

Based on your comments your proxy doesn't work.

Is it active or passive?

Is it communicating with the server? zabbix_server.log + zabbix_proxy.log

Edit: doh, you showed in the first post that Vives 93 was collecting other data.

Markku

[javiers]

It is Active, other metrics report just fine, on proxies tab shows Active.

[Markku]

Right. So basically the proxy does something correctly.

Can you try increasing the proxy logging level to see what it does when it should ping?

If needed to get clearer view of the log, you may want to disable other hosts and leave just one host active (on that proxy, that is), and even disable all other items on that one host, so that you won't be overwhelmed by the amount of log. I see the ICMP ping item interval in the default template is one minute.

Markku

[javiers]

Hi markku, after using debug level 5 i could find out that the ip which was beeing pinged was wrong, maybe because i misstyped it the first time i added the host to zabbix and it didn't update for whatever reason, so i removed the host and readded the host with the correct ip.

Sadly, even on the new host, frontend still shows down, the only part that has started working now is the tcp dump you suggested a couple of replies ago, ill add the tcp dump screenshot as well as the log from the proxy about the icmp ping.

[Markku]

Some progress that is! For me it looks like the pinger got the responses ok (three pings, one second apart as described in https://www.zabbix.com/documentation...cks#icmp-pings.

In your screenshot there were three normal items: icmpping, icmppingloss and icmppingsec. But the fourth is icmpping[]. What is that item?

Markku