ubuntu 20.04

Hello Tristis!

This is really odd for sure.
Can you share the template you are using to check this?
Did you apply any modifications for that trigger?
Are those hosts monitored by a Proxy?

Best regards,
Victor.

no proxy, all VMs at same subnet. Default template config except some history value.

zbx_export_templates.xml

Thanks.

Can you please check if the time of your Zabbix Server is the same for your hosts?
I see that you also changed the Macro {$AGENT.NODATA_TIMEOUT}. There's a default recomendation to keep it with a high value (like the default 30m)
The nodata() function may return false positives if there are time differences between Zabbix Server and Agent.

Best regards,
Victor.

yes, time same. All hosts use the same local ntp server.

this problem exist too with default value 30min.

Tristis Oris this is definitely is a time issue. Can you see this data gap on graphic? that's mean all data that comes to server is in past. Means if agent sent data for example in 12:13, but server time now is 12:15, this data will be recorded as data with 12:13 timestamp and server think it was in past. So never "actual" data will be displayed.

I would suggest check time of server with zabbix agent, with zabbix server and zabbix database.

got another trigger active, checked time on both hosts - just ~30sec difference.

with 5 or 30 min at trigger settings, 1-2 min should not be a problem. or not?

root@zabbix:~# timedatectl
               Local time: Wed 2022-03-30 00:03:03 MSK
           Universal time: Tue 2022-03-29 21:03:03 UTC
                 RTC time: Tue 2022-03-29 21:03:04

[root@kub-node-1 ~]# timedatectl
      Local time: Wed 2022-03-30 00:02:43 MSK
  Universal time: Tue 2022-03-29 21:02:43 UTC
        RTC time: Tue 2022-03-29 21:03:04

found one vm with wrong timezone. difference 6 hours and +2min. But it usualy fine, no triggers.

Tristis Oris Yeah, seems like its not an agent time issue. According to graphic gap is 7 minutes. I would suggest check zabbix databse. May be it is overloaded (not properly configured or server CPU fully utilized) and have long queue to write data.

not a cpu load.

use the postgre database. i'm not sure, but probably something like this:

postgres=# SELECT LOCALTIMESTAMP AT TIME ZONE 'localtime';
           timezone            
-------------------------------
 2022-03-30 00:26:31.257526+03

postgres=# SELECT NOW()::TIMESTAMP;
            now             
----------------------------
 2022-03-30 00:28:01.172474
(1 row)

and to be sure

SET TIMEZONE='Europe/Moscow';

time same as real local time.

and correct tz at /etc/postgresql/14/main/postgresql.conf

i think it should use server local time, due it no any additional ntp server settings.

but  on some hosts it show a ping loss.

but at same time any over metrics without interrution.

Since upgrading from 6.2 to 6.4, currently on server/agents v6.4.1, running on a Raspberry Pi-4 using an SSD, I've been seeing the same thing for just one of my agents, in my case it's a Windows 2022 server. Zero issues on any other systems, Linux/Windows/OPNsense, and this server is not particularly busy - I've tried enabling debug logging on the agent but it doesn't show any problems yet I am receiving plenty of data while the agent is apparently not responding. Reinstallation of the agent and various reboots make no difference.

i evade this problem with new trigger, which appear a bit later after 6.0.

@Tristis+Oris can you expand on what you mean After several agent and server updates I still have this problem, and it's still just this one server.

I have the same problem with one host but in 7.0beta3, which I am evaluating at the moment.

All active checks are performing as expected, data is collected as it should be (e.g. ICMP ping, NIC bytes in/out, etc) without any gaps.

However zabbix agent availablity shows unavailable on a regular basis and zabbix agent ping shows gaps.

Time is not an issue as all servers are synched. Changing the heartbeat time in the agent config does not change the results here.... very strange!

edit:
I switched my DB  (mariadb 10.11) engine to innodb, that seemed to solve the problem for some reason, since then (or the zabbix-server restart) the problems has gone for me....

edit 2:  just after posting the edit above, the problem re-appeared. active agent availability raising to  2 again

edit 3: agent2 is working flawlessly for a day now.