Notification when database gets corrupt

[jba]

Hey guys,

Is there a way to get a notification when de database gets corrupted.
I had a power outtage a few days ago, and the server rebooted after the power got back, no big deal.
But I checked the energy consumption today and I saw that I lost 1,5 year of energy data consumption.
Im a sucker for that data so I my heart screamed like a little girl.

Luckily I make backups every night, but now I lost 5 days worth of data.

Is it possible to get a notification at least? Now HA just renames the db as CORRUPT and starts a new db…
Would really appreciate a notification or some sort.

Thanks!

[koying (Chris B)]

That would be a feature request (that I’d happily +1)

Feature Requests

[ur7x]

@jba

Would really appreciate a notification of some sort.

Or an ability to set the DB to self prune.

My Database got to over 4GB when it got corrupted. And triggered a daily outage. I don’t need/use history so I deleted it and everything is fine now… But it Would be nice if would could set a max DB size and have the OS prune old entries.

[koying (Chris B)]

It does self-prune, but by time rather than by size.
If you don’t care, just set “purge_keep_days” to 1. You’ll still have long-term statistics (aggregated data), which use basically no space.

[jba]

My db just got corrupt 2 days ago again.
I switched to mariadb and will start clean.

I have no clue why my sqlite db gets corrupted the whole time.

Hope mariadb woning have these problems.

[prom00]

HA has stated that they prefer users to use the default sqlite over mariadb at the moment. Myself I had a corrupt message today as well. I managed to recover the corrupt db file and reuse it so I only lost todays information instead of all previous history.

I’ve created a feature request:

Notification when database is corrupt

It would be nice if it was possible to send notifications via the app and e-mail when a new database is intiated and the current one is copied and marked as ‘corrupt’. I’ve experienced this today. I was missing all my history and started investigating. After a while I found out the database got corrupt (but the system was still/again running). It would be nice to send an e-mail and a push message since the faster you can address this, the less data is written to the new (kind of) useless datab…

[jba]

I converted back to sqlite since.
But I just had a recent db corrupt again, last week.
Made a new topic for that.

Because HA team preferred sqlite over mariadb I decided a few months ago to go back.

[tasomaniac (Taso Dane)]

I just set up the following automation. Just manually tested by triggering a fake event. Did not test with real database corruption. Fingers crossed that my DB is never corrupt again. But if so, this may help.

You need to enable the following: System Log - Home Assistant

alias: Notify when database is corrupt
description: ""
triggers:
  - trigger: event
    event_type: system_log_event
    event_data:
      level: ERROR
conditions:
  - condition: template
    value_template: "{{ 'corrupt database' in trigger.event.data.message|lower() }}"
actions:
  - action: notify.mobile_app_pixel_6
    metadata: {}
    data:
      title: Database is corrupted
      message: "{{ trigger.event.data.message[0] }}"
      data:
        persistent: true
        channel: Alarm
        importance: high
        notification_icon: mdi:alert-octagram

[doninla (Don)]

Thanks - I had the same problem, didn’t know my db was corrupted and lost several days of data. Have you (or anyone following) verified that this automation works?