Host unavailable (server-side)

[apterix]

Hey folks:

I know that is a common problem, but I really have some difficult to troubleshoot what is happening.

I have a Zabbix server called A with IP A.ip. I have two clients called B and C with IPs B.ip and C.ip. I have others 3 clients too (D, E and F).

All of them can connect between them. No iptables, no hardware firewall and no SELINUX.

If I try telnet ports 10050 or 10051 between server and client I can connect with no problem.

I am using Zabbix Server 3.4.3-1 and all Zabbix agents are 3.4.3-1. I use same Zabbix repository with last update. All systems are CentOS 7.

I am using DNS to connect all of them. They can solve DNS fast and correctly.

At Zabbix server side I can see this log about B and C clients:

 21937:20171102:162922.502 cannot send list of active checks to "A.IP": host [A.FQDN] not found
 21932:20171102:163238.639 Zabbix agent item "system.cpu.util[,idle]" on host "A.FQDN" failed: first network error, wait for 15 seconds
 21936:20171102:163257.915 Zabbix agent item "system.cpu.util[,steal]" on host "A.FQDN" failed: another network error, wait for 15 seconds
 21936:20171102:163316.925 Zabbix agent item "system.cpu.util[,user]" on host "A.FQDN" failed: another network error, wait for 15 seconds
 21936:20171102:163335.935 temporarily disabling Zabbix agent checks on host "A.FQDN": host unavailable
 21931:20171102:164107.126 Zabbix agent item "agent.ping" on host "B.FQDN" failed: first network error, wait for 15 seconds
 21936:20171102:164126.157 Zabbix agent item "proc.num[,,run]" on host "B.FQDN" failed: another network error, wait for 15 seconds
 21936:20171102:164145.165 Zabbix agent item "system.swap.size[,free]" on host "B.FQDN" failed: another network error, wait for 15 seconds
 21936:20171102:164204.173 temporarily disabling Zabbix agent checks on host "B.FQDN": host unavailable

I can ping client and open telnet on server side. The same in client side.

No errors on client side about Zabbix agent:

  5522:20171102:171631.893 **** Enabled features ****
  5522:20171102:171631.893 IPv6 support:          YES
  5522:20171102:171631.893 TLS support:           YES
  5522:20171102:171631.893 **************************
  5522:20171102:171631.893 using configuration file: /etc/zabbix/zabbix_agentd.conf
  5522:20171102:171631.893 agent #0 started [main process]
  5523:20171102:171631.894 agent #1 started [collector]
  5524:20171102:171631.894 agent #2 started[listener #1]
  5525:20171102:171631.894 agent #3 started[listener #2]
  5526:20171102:171631.894 agent #4 started[listener #3]

I used same hostname in zabbix agent conf, zabbix server gui registration and server (client) hostname.


About clients D, E and F: They are working, but they had same problem when I configured. What I did to start work is set "disable" and then "enable" in zabbix server side. Did not work for B and C clients. I waited 2 hours with no success.

On server side I can see ZBX status in red color saying "Get value from agent failed: ZBX_TCP_READ() timed out".

I already tried (with no reason to do that) a zabbix-server restart, with no evolution...

When I do telnet on client por (10050) I can write "agent.version" and receive correct agent version before get connection closed.

I already searched and read a lot of topics about this problem, but until now with no good answers or final solutions. Threads since 2009 published this problem in this forum...

[kaspars.mednis]

First it looks like you have a active agent misconfiguration here:

cannot send list of active checks to "172.106.17.202": host [A.FQDN] not found

if you are using active zabbix agent checks, the Hostname field on zabix_agentd.conf
must match exactly (case sensitive !) with Host Name on zabbix web interface

Passive checks goes from server to agent and are using Server parameter in agent config to specify from which IP addresses connection is accepted

Active checks goes from agent to server and are using ServerActive parameter to know IP adress where to send check results, and Hostname field to let Zabbix server know to which hosts those metrics are sent

i suggest to fix it first, than look further for connectivity problems

P.S

you can mix active/passive checks on the same agent, just specify all needed parameters.
regards,
Kaspars

[apterix]

First it looks like you have a active agent misconfiguration here:



if you are using active zabbix agent checks, the Hostname field on zabix_agentd.conf
must match exactly (case sensitive !) with Host Name on zabbix web interface

Passive checks goes from server to agent and are using Server parameter in agent config to specify from which IP addresses connection is accepted

Active checks goes from agent to server and are using ServerActive parameter to know IP adress where to send check results, and Hostname field to let Zabbix server know to which hosts those metrics are sent

i suggest to fix it first, than look further for connectivity problems

P.S

you can mix active/passive checks on the same agent, just specify all needed parameters.
regards,
Kaspars

Thank you for your time. They have exactly same hostname.

I am using mix active and passive checks. At agent side I configured:
Server=A.FQDN
ServerActive=A.FQDN
Hostname=B.FQDN

remembering:
A.FQDN = zabbix hostname fqdn
B.FQDN = client hostname

From server side I can do telnet and submit agent.version and get the agent version.

[kaspars.mednis]

If telnet is going flawlessly between server to agent 10050 and agent to server 10051, and all the server and hostname fields are correct
you can try to use the zabbix get utility from the server side

https://www.zabbix.com/documentation/3.4/manual/concepts/get

simpliest example is

zabbix_get -s IP_ADRESS_OF_YOUR_AGENT -k agent.ping

What is your Timeout in zabbix_server.conf ? the default 3 seconds may be not enough, you can try to increase it

The error message

On server side I can see ZBX status in red color saying "Get value from agent failed

is regarding passive checks from server to agent only !!!


Regards,
Kaspars