Greetings, colleagues!
I noticed strange housekeeper behaviour in out Zabbix 5.0 and it'd be great if someone can help me to sort it out.
First of all, my config.
Zabbix 5.0, on Ubuntu 18.04, PostgreSQL 10.10.
MaxHousekeeperFrequency is not defined (apparently, =1 by default), MaxHousekeeperDelete =0 (previously was not defined, SPOILER: nothing changed).
Global housekeeper settings in GUI are 4w for history and 155w for trends, Override item period checkbox is enabled.
Item settings are 90d for history and 365d for trends.
What do we expect here?
As for me -- 4w for history (because it should be overriden by global settings).
Okay, 90d if this hierarchy works weird.
But, what I actually get.
SQL query, shows the oldest record for this host and this item -- it has 1y 89d of age now (and it's updated, so after each housekeeper iteration exactly the same history period remains). -- it should be 4w ago, doesn't it?
SQL query also showed, that we got exactly 2 record per minute for this item (delta 30 seconds) -- as configured, that's OK.
SQL query also showed, that we have 1.3mln records in this table (which correct if it's 2 records per minute during 1y and 89d).
Nevertheless, housekeeper regularly runs in an hour after last iteration, working for ~3.5hrs and reports, that he succesfuly deleted ~1.6mln records from history (each time same).
Could anybody help, why I have this weird storage history for this item?
This is not the only problem item, but just as an example.