For confirmed bugs, please report:
- Version: 8.12/main
- Operating System: all
- Discuss Forum URL: N/A
- Steps to Reproduce:
- schedule and upgrade on fleet for 2 min form now
- check the agent report the upgrade is scheduled
- restart the agent
- the status will not show upgrade details
UPG_SCHEDULED anymore
- then after the 2 min has passed, the upgrade will start and fail due to this bug
The logic to set the upgrade details for scheduled upgrades is on the action dispatcher:
|
// Dispatch dispatches an action using pre-registered set of handlers. |
|
// Dispatch will handle action queue operations, and retries. |
|
// Any action that implements the ScheduledAction interface may be added/removed from the queue based on StartTime. |
|
// Any action that implements the RetryableAction interface will be rescheduled if the handler returns an error. |
|
func (ad *ActionDispatcher) Dispatch(ctx context.Context, detailsSetter details.Observer, acker acker.Acker, actions ...fleetapi.Action) { |
|
var err error |
|
span, ctx := apm.StartSpan(ctx, "dispatch", "app.internal") |
|
defer func() { |
|
apm.CaptureError(ctx, err).Send() |
|
span.End() |
|
}() |
|
|
|
ad.removeQueuedUpgrades(actions) |
|
reportNextScheduledUpgrade(actions, detailsSetter, ad.log) |
|
actions = ad.queueScheduledActions(actions) |
|
actions = ad.dispatchCancelActions(ctx, actions, acker) |
|
queued, expired := ad.gatherQueuedActions(time.Now().UTC()) |
|
ad.log.Debugf("Gathered %d actions from queue, %d actions expired", len(queued), len(expired)) |
|
ad.log.Debugf("Expired actions: %v", expired) |
|
actions = append(actions, queued...) |
However it does not take into account the actions loaded from disk when the agent starts
