Something in .storage hosed all of my automations


#1

I just spent several hours troubleshooting every possible corner of my hass install to fix broken automations.

Seemingly out of nowhere ALL of my automations are failing to trigger.

I’m not sure exactly what happened or exactly where the culprit is, but it is certainly something in .storage and it is not core.entity_registry.

If I manually trigger an automation from the device entity page the automation works as expected, if i make a service call to automation.trigger, they still don’t work. Every possible type of trigger is failing on all of my automatons. I finally tried moving .storage so it could be recreated and suddenly my automations are all working again. I moved core.entity_registry back to get my entity names back.

Any idea what happened here? What in .storage could cause this behavior?

More importantly, I don’t want any part of my configuration to be derived code that I don’t understand. How much of the responsibilities of .storage can be moved directly into YAML and how do I do so? I’ve been using HASS since like .29 and I have NEVER had an issue like this, .storage seems like a step in the wrong direction to me.


#2

I don’t trust configs I don’t understand, and I don’t understand most of what is in .storage. So count me in on this discussion.

Sorry I can’t help you with the particular problem.


#3

Did you happen to just do an update to a current version when that happened?

there was a breaking change in v84 that changed the way automation enabled states were stored so if you don’t have the initial_state “option” set then the automations could all get restored in the disabled state.

I thought that in the past the initial_state defaulted to true but reading the new docs it doesn’t look like that is the case anymore.

And count my vote for the .storage (and entity registry) concept making things more difficult to troubleshoot and fix when things inevitably don’t work.


#4

I have been running 84.6 for a while and my automations have been working correctly until yesterday. I edit all my configs in Gitlab and deploy them to my hass environment in Docker once the config passes validations. All I can think of is that deploying a new config somehow mangled .storage. This is something I’ve worried a bit about but hasn’t been a problem so far. I didn’t realize that initial state was even a thing for automations so I’ll definitely keep that in mind in the future.

But I’m still very frustrated by the move to .storage. Prior to this my entire home automation configuration and infrastructure (Docker) was essentially stateless and lives in Git. .storage breaks this model by allowing changes during runtime.


#5

As I’m digging into this more, I’m realizing that my automations are working fine but the triggers never record in logbook. I do not have any logbook exclusions configured.

I should expect to see automation triggers recorded in logbook, correct?


#6

same problem in my automation
platform: time
seconds: ‘/5’
this action doesn’t work any more plz help me


#7

Read the breaking changes. You have to use the time_pattern platform now.


#8

For what its worth, I ended up finding out that my .storage was likely hosed due to a bad stick of memory in the system running hass.

Once I pulled the bad memory, my automation triggers in logbook also came back… Weird.