Hello everyone,
First post here (excuse my french) and I thought I’d share a new Automation I am testing. Disclaimer: I am getting help from ai to create it
The idea is to have an automation that runs in the background and reloads any integration that seems “unavailable” or “unknown”.
I’ve had a few “spin-off” from a couple of integrations without explainable reasons. So I thought I’d mitigate the problem by automating their reload.
In more details, the automation runs every five minutes. 
- If an entity is OK, it doesn’t do anything with said entity.
- If an entity is KO since at least 10 minutes, it tries to reload it once.
- If the reload is a success then nothing (the automation works as it should).
If it’s not a success then:
- It’ll add all the KO entities into a helper ( input_text.watchdog_global_exclusions_temp). This is to exclude them from upcoming global watchdog scans until the underlying problem is fixed.
- It’ll add them to a to do list ( todo.watchdog_issues ) for me to remember! Those entities have a problem and I should work on them.
- It’ll send me a push notification, telling me which entities have a problem. And by doing so asking me if I want to “clear” or “rearm”
Clear means that the entities stay in the helper and that upon the next checks they will be ignored so that the automation doesn’t do constant reloads. This is valid until these entities are erased from the helper.
Rearm means that the entities are deleted from the helper and that the automation will rescan them.
On the side of this automation, I have a “maintenance” dashboard that has the todo list and the rearm button
Here is the code for the automation:
input_text:
watchdog_global_exclusions_temp:
name: "Watchdog global — Exclusions temporaires"
max: 255
initial: ""
input_datetime:
watchdog_global_dernier_reload:
name: "Watchdog global — Dernier reload"
has_date: true
has_time: true
watchdog_global_derniere_alerte:
name: "Watchdog global — Dernière alerte"
has_date: true
has_time: true
watchdog_global_derniere_mise_en_pause:
name: "Watchdog global — Dernière mise en pause"
has_date: true
has_time: true
input_button:
watchdog_global_acquitter:
name: "Watchdog — Acquitter"
icon: mdi:check-circle-outline
watchdog_global_rearmer:
name: "Watchdog — Réarmer"
icon: mdi:reload
template:
- sensor:
- name: "Watchdog global — Entités KO"
unique_id: watchdog_global_entites_ko
icon: mdi:heart-broken
state: >
{% set ignored_domains = [
'automation', 'script', 'scene', 'group', 'sun', 'person', 'zone',
'button', 'event', 'update'
] %}
{% set ignored_entities = [
'sensor.watchdog_global_entites_ko',
'sensor.watchdog_global_integrations_ko',
'input_text.watchdog_global_exclusions_temp',
'input_datetime.watchdog_global_dernier_reload',
'input_datetime.watchdog_global_derniere_alerte',
'input_datetime.watchdog_global_derniere_mise_en_pause',
'input_button.watchdog_global_acquitter',
'input_button.watchdog_global_rearmer',
'sensor.onduleur_production_cumulee',
'sensor.mesure_du_courant_exporte'
] %}
{% set temp_excluded_raw = states('input_text.watchdog_global_exclusions_temp') %}
{% set temp_excluded = temp_excluded_raw.split('|') if temp_excluded_raw not in ['unknown','unavailable','none',''] else [] %}
{% set ns = namespace(items=[]) %}
{% for s in states %}
{% set domain = s.entity_id.split('.')[0] %}
{% if domain not in ignored_domains
and s.entity_id not in ignored_entities
and s.entity_id not in temp_excluded
and s.state in ['unknown', 'unavailable'] %}
{% set age_min = (as_timestamp(now()) - as_timestamp(s.last_changed)) / 60 %}
{% if age_min >= 10 %}
{% set ns.items = ns.items + [s.entity_id] %}
{% endif %}
{% endif %}
{% endfor %}
{{ ns.items | count }}
attributes:
entities: >
{% set ignored_domains = [
'automation', 'script', 'scene', 'group', 'sun', 'person', 'zone',
'button', 'event', 'update'
] %}
{% set ignored_entities = [
'sensor.watchdog_global_entites_ko',
'sensor.watchdog_global_integrations_ko',
'input_text.watchdog_global_exclusions_temp',
'input_datetime.watchdog_global_dernier_reload',
'input_datetime.watchdog_global_derniere_alerte',
'input_datetime.watchdog_global_derniere_mise_en_pause',
'input_button.watchdog_global_acquitter',
'input_button.watchdog_global_rearmer',
'sensor.onduleur_production_cumulee',
'sensor.mesure_du_courant_exporte'
] %}
{% set temp_excluded_raw = states('input_text.watchdog_global_exclusions_temp') %}
{% set temp_excluded = temp_excluded_raw.split('|') if temp_excluded_raw not in ['unknown','unavailable','none',''] else [] %}
{% set ns = namespace(items=[]) %}
{% for s in states %}
{% set domain = s.entity_id.split('.')[0] %}
{% if domain not in ignored_domains
and s.entity_id not in ignored_entities
and s.entity_id not in temp_excluded
and s.state in ['unknown', 'unavailable'] %}
{% set age_min = (as_timestamp(now()) - as_timestamp(s.last_changed)) / 60 %}
{% if age_min >= 10 %}
{% set ns.items = ns.items + [s.entity_id] %}
{% endif %}
{% endif %}
{% endfor %}
{{ ns.items }}
- name: "Watchdog global — Intégrations KO"
unique_id: watchdog_global_integrations_ko
icon: mdi:puzzle-remove
state: >
{% set bad_entities = state_attr('sensor.watchdog_global_entites_ko', 'entities') | default([], true) %}
{% set ns = namespace(ids=[]) %}
{% for e in bad_entities %}
{% set cid = config_entry_id(e) %}
{% if cid is not none %}
{% set ns.ids = ns.ids + [cid] %}
{% endif %}
{% endfor %}
{{ ns.ids | unique | list | count }}
attributes:
entry_ids: >
{% set bad_entities = state_attr('sensor.watchdog_global_entites_ko', 'entities') | default([], true) %}
{% set ns = namespace(ids=[]) %}
{% for e in bad_entities %}
{% set cid = config_entry_id(e) %}
{% if cid is not none %}
{% set ns.ids = ns.ids + [cid] %}
{% endif %}
{% endfor %}
{{ ns.ids | unique | list }}
automation:
- id: watchdog_global_reload_integrations
alias: "Watchdog global — Reload intégrations KO"
description: >
Surveille les entités unknown/unavailable depuis au moins 10 minutes,
tente un reload de leur intégration, puis alerte si le problème persiste.
Exclut Huawei Solar, géré par son watchdog dédié.
mode: single
trigger:
- platform: time_pattern
minutes: "/5"
variables:
reload_cooldown_minutes: 60
alert_cooldown_minutes: 60
bad_entities: "{{ state_attr('sensor.watchdog_global_entites_ko', 'entities') | default([], true) }}"
entry_ids: "{{ state_attr('sensor.watchdog_global_integrations_ko', 'entry_ids') | default([], true) }}"
last_reload: "{{ states('input_datetime.watchdog_global_dernier_reload') }}"
can_reload: >
{% if last_reload in ['unknown', 'unavailable', 'none', ''] %}
true
{% else %}
{{ (as_timestamp(now()) - as_timestamp(as_datetime(last_reload))) > (reload_cooldown_minutes * 60) }}
{% endif %}
last_alert: "{{ states('input_datetime.watchdog_global_derniere_alerte') }}"
can_alert: >
{% if last_alert in ['unknown', 'unavailable', 'none', ''] %}
true
{% else %}
{{ (as_timestamp(now()) - as_timestamp(as_datetime(last_alert))) > (alert_cooldown_minutes * 60) }}
{% endif %}
condition:
- condition: template
value_template: "{{ bad_entities | count > 0 }}"
- condition: template
value_template: "{{ entry_ids | count > 0 }}"
- condition: template
value_template: "{{ can_reload }}"
action:
- repeat:
for_each: "{{ entry_ids }}"
sequence:
- service: homeassistant.reload_config_entry
data:
entry_id: "{{ repeat.item }}"
- delay: "00:00:05"
- service: input_datetime.set_datetime
target:
entity_id: input_datetime.watchdog_global_dernier_reload
data:
datetime: "{{ now().isoformat() }}"
- delay: "00:00:30"
- variables:
still_bad: >
{% set ns = namespace(items=[]) %}
{% for e in bad_entities %}
{% if states(e) in ['unknown', 'unavailable'] %}
{% set ns.items = ns.items + [e] %}
{% endif %}
{% endfor %}
{{ ns.items }}
- choose:
- conditions:
- condition: template
value_template: "{{ still_bad | count > 0 and can_alert }}"
sequence:
- service: input_text.set_value
target:
entity_id: input_text.watchdog_global_exclusions_temp
data:
value: "{{ still_bad | join('|') }}"
- service: input_datetime.set_datetime
target:
entity_id: input_datetime.watchdog_global_derniere_mise_en_pause
data:
datetime: "{{ now().isoformat() }}"
- repeat:
for_each: "{{ still_bad }}"
sequence:
- service: todo.add_item
target:
entity_id: todo.watchdog_issues
data:
item: "Watchdog: {{ repeat.item }}"
description: >
Entité toujours en {{ states(repeat.item) }} après tentative de reload automatique
le {{ now().strftime('%d.%m.%Y à %H:%M') }}.
Cette entité a été placée en exclusion temporaire jusqu'à acquittement/réarmement.
due_date: "{{ now().date().isoformat() }}"
- service: persistent_notification.create
data:
notification_id: "watchdog_global_ko"
title: "⚠️ Watchdog global — Entités encore KO"
message: >
Après tentative automatique de reload, certaines entités restent
en unknown/unavailable :
{{ still_bad | join(', ') }}.
Elles sont maintenant exclues des prochains cycles jusqu'à acquittement
ou réarmement manuel.
- service: notify.yannick
data:
title: "⚠️ Watchdog global — Entités encore KO"
message: >
Entités encore KO après reload : {{ still_bad | join(', ') }}.
Elles sont temporairement exclues jusqu'à ton acquittement.
data:
tag: watchdog_global_ko
persistent: true
sticky: true
actions:
- action: "WATCHDOG_ACK"
title: "Acquitter"
- action: "WATCHDOG_REARM"
title: "Réarmer"
- service: input_datetime.set_datetime
target:
entity_id: input_datetime.watchdog_global_derniere_alerte
data:
datetime: "{{ now().isoformat() }}"
- id: watchdog_global_ack_mobile
alias: "Watchdog global — Acquittement depuis notification"
mode: single
trigger:
- platform: event
event_type: mobile_app_notification_action
event_data:
action: WATCHDOG_ACK
action:
- service: script.alerte_systeme
data:
niveau: info
titre: "Watchdog acquitté"
message: >
Les entités actuellement exclues restent en pause technique.
Le backlog reste visible dans la liste to-do pour traitement ultérieur.
- id: watchdog_global_rearm_mobile
alias: "Watchdog global — Réarmement depuis notification"
mode: single
trigger:
- platform: event
event_type: mobile_app_notification_action
event_data:
action: WATCHDOG_REARM
action:
- service: input_text.set_value
target:
entity_id: input_text.watchdog_global_exclusions_temp
data:
value: ""
- service: persistent_notification.dismiss
data:
notification_id: "watchdog_global_ko"
- service: script.alerte_systeme
data:
niveau: info
titre: "Watchdog réarmé"
message: "Les exclusions temporaires du watchdog global ont été vidées."
- id: watchdog_global_ack_button
alias: "Watchdog global — Acquittement depuis bouton"
mode: single
trigger:
- platform: state
entity_id: input_button.watchdog_global_acquitter
action:
- service: script.alerte_systeme
data:
niveau: info
titre: "Watchdog acquitté"
message: >
Les exclusions en cours sont conservées.
Les éléments restent visibles dans la to-do list pour traitement plus tard.
- id: watchdog_global_rearm_button
alias: "Watchdog global — Réarmement depuis bouton"
mode: single
trigger:
- platform: state
entity_id: input_button.watchdog_global_rearmer
action:
- service: input_text.set_value
target:
entity_id: input_text.watchdog_global_exclusions_temp
data:
value: ""
- service: persistent_notification.dismiss
data:
notification_id: "watchdog_global_ko"
- service: script.alerte_systeme
data:
niveau: info
titre: "Watchdog réarmé"
message: "Les exclusions temporaires du watchdog global ont été vidées."
And here is the code for the dashboard:
views:
- title: Watchdog
path: watchdog
icon: mdi:shield-alert
cards:
- type: vertical-stack
cards:
- type: entities
title: Watchdog — État
entities:
- entity: sensor.watchdog_global_entites_ko
name: Entités KO actives
- entity: sensor.watchdog_global_integrations_ko
name: Intégrations touchées
- entity: input_text.watchdog_global_exclusions_temp
name: Exclusions temporaires
- entity: input_datetime.watchdog_global_dernier_reload
name: Dernier reload
- entity: input_datetime.watchdog_global_derniere_alerte
name: Dernière alerte
- entity: input_datetime.watchdog_global_derniere_mise_en_pause
name: Dernière mise en pause
- type: horizontal-stack
cards:
- type: button
entity: input_button.watchdog_global_acquitter
name: Acquitter
icon: mdi:check-circle-outline
tap_action:
action: perform-action
perform_action: input_button.press
target:
entity_id: input_button.watchdog_global_acquitter
- type: button
entity: input_button.watchdog_global_rearmer
name: Réarmer
icon: mdi:reload
tap_action:
action: perform-action
perform_action: input_button.press
target:
entity_id: input_button.watchdog_global_rearmer
- type: todo-list
entity: todo.watchdog_issues
title: Watchdog — Backlog
hide_completed: false
display_order: duedate_asc
It’s definitely a work in progress.
Please comment and give me you thoughts on to how to make it better or how you would go about it.
Happy coding fellow smart home owners! 