i thought i had this reported already, but can't find anything. hopefully i'm not creating a duplicate here

the current escalation configuration is... complicated. one has to match step duration with step numbers, take into account operation overrides. it is quite error prone. changing the default escalation interval messes everything up.

it is proposed to hide all the low-level detail from the users, instead exposing to them simple controls that would still offer the same level of flexibility.

the escalation interval would be changed to 60. this would be the lowest granularity of escalation steps - going lower is possible, but then alerter frequency should be taken into account (30 seconds by default).
users would get options to specify the time when an operation should happen. for each operation, they would be able to also set recurrence/repeating, setting interval and either the amount of messages, or the length of time, for how long such repeated messages should be sent. zabbix would calculate how the time and recurrence matches to the internal steps - user would not have to do that themselves.
this would probably allow to get rid of the ability to override interval on the operation level, it would serve no purpose anymore.

extra things to take into account :
a) migrating existing actions, especially the ones that use per-operation overriding;
b) with the interval of 60 seconds, check for how long an escalation could run (step value overflow) on all supported databases. or maybe get rid of steps/their ids, just check escalations and see what and when should be performed ?

there are likely other potential problems. it is suggested to take great care not to have functional regressions.