-
Type: Improvement
-
Resolution: Fixed
-
Priority: Major - P3
-
Affects Version/s: None
-
Component/s: Testing Infrastructure
-
None
-
Query Execution
-
Fully Compatible
-
200
This is a suggestion to improve the debugging experience around failures raised by the txnRunCommandOverride functionality used on our testing infrastructure to retry upon transient transaction errors.
These suggestions came up after debugging a couple of build failures caused by a txn error.
For some context, the overridden method for retrying txn can stop for two reasons: either the number of retries has exceeded a limit or a timeout has expired.
I found it not trivial to understand the timeout had expired by just looking at the logs, that's why I'm suggesting to improve the logging on the following points:
- Print the remaining time on every retry as we do with the remaining attempts.
- Print a message when we stop retrying with the stopping reason and the last failure. This is especially interesting when the operation times out since it's confusing to see a number of remaining attempts bigger than 1 as the last retry without succeeding.
- is related to
-
SERVER-94538 Increase txn timeout in multi_stmt_txn_jscore_passthrough_with_migration
- Closed