Help with Housekeeper Processes (100% busy for 3 hours every hour)

[kernbug]

At least its severity should be 'Information'

Nope. If we look at the current load on the disk subsystem, then we are more likely to find problems with performance, at 100%(many hours) load on the housekeeper.
Most performance problems are often caused by insufficient performance of the database (hardware and software settings) used.

[kloczek]

If so , why it's listed in the problems list. At least its severity should be 'Information'

Such alarm it is more like template issue (if it is with high severity).
Problem may be if HK process will be running to long. It always means that something in zabbix stack needs to be changed.
If for example HK spends to much time in history tables to delete oldest data it means that housekeeper process should be disabled after introduce partitioned history* tables and delete oldest data by dropping oldest history* tables partitions.
Maintaining history*/trends* tables content without partitioning is OK in some small/POC zabbix stacks only.

[lpossamai]

Which version do you use? Does upgrading to latest one help ?

Zabbix Server 3.4. It's been running for less than a month and the DB size is already 4GB. Imagine in 2 years!!!

[kloczek]

If you will be using for example exactly the same history period (let's say 2 weeks) or raw monitoring data. With 3 years trends data those trends data may take 5-8% of whole database size.
Problem is that probably you are still using OOTB templates which have on most of the items 90 days of history period so your database may reach still max in next two months ans if you will be not using partitioned history* tables you database still will be slowly growing even if ~number of monitoring points data will be ~const.

Part of the proper zabbix administration is taking care of proper database backend maintenance which is part of the general DB support tasks .. and/or has nothing to do with zabbix per se.

[lahiruabeysinghe]

If so , why it's listed in the problems list. At least its severity should be 'Information'

The trigger which relates to this Problem checks if the value is not just above 75%, but also if it averages that value over 30 mins. Housekeeping process running for 30 mins is the "Problem".