Uploaded image for project: 'WiredTiger'
  1. WiredTiger
  2. WT-9063

Create a general approach to handle WT_ROLLBACK errors in Python test suite

    XMLWordPrintableJSON

Details

    Description

      In WT-8991, we are fixing test cases that fail due to WT_ROLLBACK. But the issue is potentially more broad than just those few test cases. As of WT-8869, any readonly transactions can now be rolled back, and of course read/write transactions have always had that possibility. WT handles retries of single operations that are not within a transaction, but explicit transactions are used pretty widely in the test suite, so this can come up.  The recommended approach for a production program is to rollback/retry any failing transactions.  Whether that's the right thing for short test cases is part of what we are discussing here.

      Since the problem might require fiddling a bit with a large number of tests, it would be good to discuss the proper approach. There are 183 python test files that use begin_transaction. Of those, 36 files set the cache size to 50M, and these are especially vulnerable to rollback errors.

      Attachments

        Issue Links

          causes

          Bug - A problem which impairs or prevents the functions of the product. WT-9300 Remove retries from test_txn24.py to fix build failure

          • Minor - P4 - Minor loss of function, or other problem where easy workaround is present.
          • Closed
          is depended on by

          Build Failure - WT-9272 failed: test_prepare_hs01 conflict between concurrent operations on macos-1014

          • Major - P3 - Major loss of function.
          • Backlog

          Build Failure - WT-9166 test hs20 WT_ROLLBACK conflict between concurrent operations

          • Major - P3 - Major loss of function.
          • Closed

          Build Failure - WT-9193 test_gc02 WT_ROLLBACK: conflict between concurrent operations

          • Major - P3 - Major loss of function.
          • Closed

          Build Failure - WT-9266 failed: test_rollback_to_stable12 conflict between concurrent operations on macos-1014 [wiredtiger @ f1050387]

          • Major - P3 - Major loss of function.
          • Closed

          Build Failure - WT-9289 failed: test_timestamp18 WT_ROLLBACK on macos-1014 [wiredtiger @ c5e3a8a0]

          • Major - P3 - Major loss of function.
          • Closed
          is duplicated by

          Build Failure - WT-9116 test_hs14: oldest pinned transaction ID rolled back for eviction

          • Major - P3 - Major loss of function.
          • Closed

          Build Failure - WT-9166 test hs20 WT_ROLLBACK conflict between concurrent operations

          • Major - P3 - Major loss of function.
          • Closed

          Build Failure - WT-9266 failed: test_rollback_to_stable12 conflict between concurrent operations on macos-1014 [wiredtiger @ f1050387]

          • Major - P3 - Major loss of function.
          • Closed

          Build Failure - WT-9289 failed: test_timestamp18 WT_ROLLBACK on macos-1014 [wiredtiger @ c5e3a8a0]

          • Major - P3 - Major loss of function.
          • Closed
          is related to

          Build Failure - WT-8991 Implement a retry mechanism for transactions that may rollback

          • Major - P3 - Major loss of function.
          • Closed

          Build Failure - WT-9106 test_checkpoint_snapshot03 oldest pinned transaction ID rolled back for eviction

          • Major - P3 - Major loss of function.
          • Closed
          related to

          Build Failure - WT-9151 Use retry mechanism against rollbacks in test_txn24

          • Major - P3 - Major loss of function.
          • Closed

          Activity

            People

              donald.anderson@mongodb.com Donald Anderson
              donald.anderson@mongodb.com Donald Anderson
              Votes:
              0 Vote for this issue
              Watchers:
              9 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: