Trouble with custom interval / scheduling.

[nscruman]

Hi,

I am trying to setup some telnet style tests to happen at the beginning of the day before our production processes start up, and not test during production hours. I have distilled this down to the attached image - running a custom tcp performance check of google dns.

I have been copying and pasting from the scheduling documentation -

https://www.zabbix.com/documentation/3.2/manual/config/items/item/custom_intervals

I can confirm that some simple filters work. So 'm0-59' worked, but anytime I try to add an hour qualifier I get no data. So the documented example 'h9m10-40' gave no data. 'h9m10' gave no data. My desired endstate is 'wd1-5h8m0-30' to do a check every minute from 8:00am to 8:30am, Mon-Fri.

Has anyone had a similar problem and were you able to resolve this to run an item for a few minutes during a single hour during weekdays only? Is there an easy way to debug through the log what might be happening?

Any help would be hugely appreciated.

[kloczek]

https://www.zabbix.com/documentation/3.2/manual/config/items/itemtypes/simple_checks

net.tc.service.perf[] key is working only as "simple check". As all simple checks is probed only from zabbix server/proxy.

[nscruman]

That is certainly tough to determine from the documentation.

https://www.zabbix.com/documentation/3.2/manual/config/items/itemtypes/zabbix_agent

This clearly shows the tcp perf as being under the zabbix_agent type checks. I guess i can do this as some form of system.run or similar, but why would this work for some schedule filter types and not others?

To verify I will clone the item and change it to proc.cpu.util[,root] and see if that has the same error - that will eliminate this as being a check type problem and allow us to verify that it is a schedule problem.

[nscruman]

Confirmed that running a CPU agent check suffers the same problem - scheduling using the documented examples doesn't appear to gather data.