-
Notifications
You must be signed in to change notification settings - Fork 12
Description
By recent decisions PREvant processes each request one after another (see worker producer queue, introduced with #40 and related PRs). That also means if there are many requests for the same application, each request will be processed one after another. For example, if there is for each service a merge/pull request and all will be merged at the same time, which then trigger a main line CI run including an API call to PREvant to update the default application, then PREvant processes them one after another. This might then lead to CI timeouts because the processing of all main line CI runs takes longer.
However, PREvant should be able to merge all the requests with the same application that are in line of the queue, into one interaction with the infrastructure (K8s or docker).
- merge queued tasks for PostgreSQL queue (see Merging Queued Tasks #280)
- merge queued tasks for in-memory queue