We're running a Zabbix proof of concept with a single server. Poller processes that use ODBC will increase steadily in memory usage until the server runs out.

UnixODBC version 2.3.7 (upgraded from 2.3.1 that was part of Centos 7).

IBM i Access ODBC driver 1.1.0.11 (latest).

Number of db.odbc items is almost steady, so it looks like every odbc call eats up some memory that isn't released.

I tried to open a consulting request, but they ask that I open a bug report because they suspect a memory leak in the zabbix code. What can I do to help sort this out?