Releases: freepik-company/notifik
v0.7.0
Changes:
globals.DeleteWatcherFromWatcherPool()
is replaced byglobals.DisableWatcherFromWatcherPool()
to avoid some concurrency problems. From now, empty watchers will be killed not to consume resources, but their registration will be kept in memory. This way we avoid collisions betweencontroller.NotificationsReconciler
andxyz.WorkloadController
during their operation and checks in common places
Contributors:
@achetronic
v0.6.0
Changes:
-
--informer-seconds-to-resync
is now--informer-duration-to-resync
to simplify some pieces of code and give more freedom to the users at the same time -
📜 Update README to include changes related to main templating scope
. := dict "eventType" (string) "object" (dict) "previousObject" (dict)
Contributors:
@achetronic
v0.5.0
Features:
-
🤯 Add more data available to the conditions/message template. Previously the main scope
.
on templating included only the object triggering the event. From this release this was replaced by a more complete object. = {eventType, object, previousObject}
. This will help to handle some use cases where it's needed to distinguish the events to send notifications. For example, like using Notifik to set markers on Grafana when a new ReplicaSet is created -
❤️ Replace Watchers with RetryWatchers and Informers. Previously, resources in Kubernetes were watched using pure client-go watchers. On network issues or Kubernetes API server timeouts, watcher was simply recreated, triggering all the events again, spamming Kubernetes API server from time to time. Now, this was replaced by RetryWatchers and Informers, that are two newer patterns to do the same with some retries and cache in the middle. This is intended to hide this problems from the final user, being much more resilient and causing less spam to the API. Both are included (Informers are used by default) and you can chose your favorite strategy with a flag
-
🏳️ Add several flags
--use-watchers
to enable using RetryWatchers instead of Informers.--informer-seconds-to-resync
to set the amount of seconds to wait before resyncing all the objects when using Informers--watcher-events-per-second
to replace previous--events-per-second
-
📜 Update README to include changes
Contributors:
@achetronic
notifik-helm-chart-0.7.0
A Helm chart for Notifik, a Kubernetes operator to watch groups of resources and send notifications if conditions are met (realtime)
notifik-helm-chart-0.6.0
A Helm chart for Notifik, a Kubernetes operator to watch groups of resources and send notifications if conditions are met (realtime)
notifik-helm-chart-0.5.0
A Helm chart for Notifik, a Kubernetes operator to watch groups of resources and send notifications if conditions are met (realtime)
v0.4.0
Feature:
- ♻️ Add WatcherPool cleaner process: this is a process that is triggered on
Notification
related events to kill watchers that are not claimed anymore by anyNotification
resource. - 🏳️ Add flag
--enable-watcher-cleaner
to enable WatcherPool cleaner process. This is disabled by default while it's experimental feature - 📜 Update README to include changes
Contributors:
@achetronic
notifik-helm-chart-0.4.0
A Helm chart for Notifik, a Kubernetes operator to watch groups of resources and send notifications if conditions are met (realtime)
v0.3.0
Fixes
- Improve logs for Alertmanager integration (sorry for the previous ones, past me)
Feature:
- Add
Blocked
flag intoWatcher
type to allow preventing a watcher from being started in future releases
Contributors:
@achetronic @jmjimenezlp
notifik-helm-chart-0.3.0
A Helm chart for Notifik, a Kubernetes operator to watch groups of resources and send notifications if conditions are met (realtime)