You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We need to keep track of past retries somewhere so that when an execution fails, we need to know the previous attempts for that same execution to decide if we should reschedule or fail.
The right place to keep track of this information is in the Execution instead of history entries or any other type of entries. Reasons include:
There won't be APIs or reasons to access the retries independently of the execution itself
We want to share the previous attempts to users when they describe an execution
We want to keep track of all retries related to the same execution and not just the retry count as we want to know the number of retries within a sliding window
We want to have efficient access to these retries in the scheduler side. No need to query the execution, and then again query these retry entries, or chain executions and keep querying previous executions.
Chaining of executions might cause problems or limit what we can do with state garbage collection
With that in mind, and while it might seem duplication of state or inefficient writes, every execution model will hold rescheduling events of previous executions
The text was updated successfully, but these errors were encountered:
We need to keep track of past retries somewhere so that when an execution fails, we need to know the previous attempts for that same execution to decide if we should reschedule or fail.
The right place to keep track of this information is in the Execution instead of history entries or any other type of entries. Reasons include:
With that in mind, and while it might seem duplication of state or inefficient writes, every execution model will hold rescheduling events of previous executions
The text was updated successfully, but these errors were encountered: