Upgrade from Zabbix 7.0alpha4 to 7.0alpha7 broke many Zabbix hosts

[kbcetn]

Hi all,

I was running Zabbix 7.0alpha4 and had everything monitoring swimmingly. I wanted to upgrade to 7.0alpha7 and did so on Ubuntu 22.04.3 so I did an apt upgrade zabbix-server-mysql zabbix-frontend-php zabbix-apache-conf zabbix-sql-scripts zabbix-agent. This required a system restart because the kernel got upgraded. So I restarted and found that after the upgrade, many hosts (30 of 42) are now reporting MySQL down in spite of the only thing changing was the Zabbix server and reboot. Is there something I'm doing incorrectly here? Is there an upgrade step I missed?

KB

[kbcetn]

FYI - I can telnet to ports 10050 and 10051 as appropriate to or from each of the agents to or from the server. So I know that the firewall wasn't impacted. I've tried removing and then re-adding hosts to see if that was the problem. Apparently, it wasn't. Agents are complaining about a message coming to them that's too large (Message size 17179869211 from 192.168.76.120 exceeds the maximum size 134217728 bytes. Message ignored. - 192.168.76.120 is the Zabbix server). The host this appears on has been re-installed. I'm unsure of what to do next. Any suggestions would be greatly appreciated. :-)

KB

[kbcetn]

Just for grins/giggles, I blew away the MySQL database for Zabbix (after backing it up, of course), then let Zabbix re-discover all the agents. That didn't fix anything either.

[kbcetn]

Had to install zabbix_get to figure out this:

zabbix_get -s x.x.x.x -k mysql.ping
zabbix_get [112768]: Check access restrictions in Zabbix agent configuration

Strange!

[kbcetn]

FYI - as mentioned previously, this was working just find on 7.0alpha4, but broke in 7.0alpha7. I suspect a bug introduced between those versions. Setting Server= and ServerActive= didn't make a difference when I tried 0.0.0.0/0 and tcpdump on the agent side said that the request was coming from the valid server IP. Running zabbix_agentd -t mysql.ping works fine (reports mysqld is alive) on the agent, but any attempt to run the same from the Zabbix server using zabbix_get (see above), fails as shown above. Any suggestions on what I should do next?

[Andy Afanasyev]

Got the same problem on 7.0.0alpha8. It's impossible to get information from agents older than 4.0.
Message size 12884901898 from MyServerIp exceeds the maximum size 134217728 bytes. Message ignored.
As I can understand I can't downgrade to previous minor versions? I need to export all hosts and templates, reinstall and than import in older version?
P.S. I'm running 7.0 in docker compose stack.

[Alexei]

Thanks for reporting this problem. I passed this information on to our dev team. I hope it will be resolved asap and delivered in the next alpha release.

[Andy Afanasyev]

Thank you. We'll be waiting for new version. For now I've installed old alpha4 inside docker container and "downgraded" database. There was some minor problems with new widgets but it's working.

P.S. I've created an issue yesterday: https://support.zabbix.com/browse/ZBX-23727

[kbcetn]

Alexei - thanks. Looking forward to that solution. :-)

[kbcetn]

Andy - thanks for filing that Jira. :-)

[kbcetn]

FYI all - my issue was fixed in Zabbix 7.0alpha9 :-)