It looks like a typo in "src/libs/zbxdbcache/dbsync.c" file which causes to performance degradation in PgSQL queries.

It was caused by code optimisation in commit "0e36dceb9857d2fdbca8a116b0ef1afbf6c80986"

Line 2247 now contains:

" where t.triggerid in (select distinct t.triggerid"

but nested select should refer to tg.triggerid instead.

Please find attached:

1. top output showing postgres processes stuck in SELECT
2. active queries containing aforementioned select with possible typo which causes server performance degradation
3. explain analyze outputs with typo (original query) and without it (fixed query). The difference is 232588 ms vs 38 ms.
4. screenshot of the commit

Thanks!