SLES 15 sp3, PostgreSQL v13, Zabbix server compiled from a sources
Sprint 86 (Mar 2022), Sprint 87 (Apr 2022)
Steps to reproduce:
- Install a Zabbix Server v5.0 (LTS);
- Set up some services ("Configuration" -> "Services"), for example:
- "LDAP pseudo-cluster": has the status calculation algorithm as "Problem, if have problems";
- two child services: "ldap1 node" and "ldap2 node" with the status calculation algorithm as "Problem, if has a problem";
- Perform on-site upgrade onto Zabbix v6.0;
- After Zabbix server v6.0 is started and the database has been converted, look at the menu "Services" -> "Services" and check the appropriate services.
The calculation algorithm "Problem, if have problems" has been converted onto "Most critical of the child services", but "Problem, if has a problem" – onto "Most critical if have problems".
It causes to the wrong parent service calculations when the child is failed.
See the screenshots attached:
- "all children" in v5.0:
- the same service after conversion in v6.0:
- "one child" in v5.0:
- the same service in v6.0:
Just opposite conversion:
|Problem, if all children have problems||->||Most critical if all children have problems|
|Problem, if one child has a problem||->||Most critical of the child services|