Refs 4739: metric check for failing snap tasks on RH repos #833
Conversation
dominikvagner
changed the title
|
|
||
| subQuery := d.db.WithContext(ctx). | ||
| Model(&models.RepositoryConfiguration{}). | ||
| Select("repository_configurations.uuid, bool_or(tasks.status ILIKE ? AND tasks.finished_at > ?) AS has_successful_tasks", fmt.Sprintf("%%%s%%", config.TaskStatusCompleted), date). |
There was a problem hiding this comment.
| Select("repository_configurations.uuid, bool_or(tasks.status ILIKE ? AND tasks.finished_at > ?) AS has_successful_tasks", fmt.Sprintf("%%%s%%", config.TaskStatusCompleted), date). | |
| Select("repository_configurations.uuid, bool_or(tasks.status = ? AND tasks.finished_at > ?) AS has_successful_tasks", config.TaskStatusCompleted), date). |
this can be simplified slightly
There was a problem hiding this comment.
This query looks good. I believe it would miss the (very unlikely) case where there's no snapshot tasks for a repository. I don't think this would happen naturally, but maybe could happen if there's a bug with task cleanup.
There was a problem hiding this comment.
What do you think about handling that case? I could see just adding a second query and adding the two counts together as a simple solution.
| @@ -134,6 +134,7 @@ type MetricsDao interface { | |||
| PendingTasksAverageLatency(ctx context.Context) float64 | |||
| PendingTasksCount(ctx context.Context) int64 | |||
| PendingTasksOldestTask(ctx context.Context) float64 | |||
| RHReposNoSuccessfulSnapshotTaskIn36Hours(ctx context.Context) int64 | |||
There was a problem hiding this comment.
| RHReposNoSuccessfulSnapshotTaskIn36Hours(ctx context.Context) int64 | |
| RHReposSnapshotNotCompletedInLast36HoursCount(ctx context.Context) int64 |
This is a nitpick, but what do you think about renaming this metric?
"No" can also be an abbreviation for "number", so "RHReposNoSuccessfulSnapshotTaskIn36Hours" could be interpreted as "Number of sucessful snapshot tasks"
Summary
This PR adds a new metric for checking if the snapshot tasks are failing on RH repos, specifically if there isn't a successful one in the last 36 hours and returns number of repos on which is this happening.
Testing steps