Uploaded image for project: 'WiredTiger'
  1. WiredTiger
  2. WT-2846

Several bugs related to reconfiguring eviction server at runtime

    Details

    • Type: Bug
    • Status: Closed
    • Priority: Major - P3
    • Resolution: Fixed
    • Affects Version/s: None
    • Fix Version/s: WT2.9.0, 3.3.12, 3.2.10
    • Labels:
      None

      Description

      There was a Jenkins failure when running the new reconfigure test called wt2719_reconfig. The call stacks from the failure are:

      FAIL: test_wt2719_reconfig
      ==========================
       
      ASAN:SIGSEGV
      =================================================================
      ==17893==ERROR: AddressSanitizer: SEGV on unknown address 0x000000000000 (pc 0x0000005064cf bp 0x7f312d7ebed0 sp 0x7f312d7ebb00 T102)
          #0 0x5064ce in __evict_thread_run /home/jenkins/jenkins/workspace/wiredtiger-clang-sanitizer/build_posix/../src/evict/evict_lru.c:184:2
          #1 0x3e6ce07554 in start_thread (/lib64/libpthread.so.0+0x3e6ce07554)
          #2 0x3e6cb02dec in __clone (/lib64/libc.so.6+0x3e6cb02dec)
       
      AddressSanitizer can not provide additional info.
      SUMMARY: AddressSanitizer: SEGV /home/jenkins/jenkins/workspace/wiredtiger-clang-sanitizer/build_posix/../src/evict/evict_lru.c:184 __evict_thread_run
      Thread T102 created by T92 here:
          #0 0x48d95f in __interceptor_pthread_create (/home/jenkins/jenkins/workspace/wiredtiger-clang-sanitizer/build_posix/test/csuite/test_wt2719_reconfig+0x48d95f)
          #1 0x55a15f in __wt_thread_create /home/jenkins/jenkins/workspace/wiredtiger-clang-sanitizer/build_posix/../src/os_posix/os_thread.c:22:84
          #2 0x3e6ce07554 in start_thread (/lib64/libpthread.so.0+0x3e6ce07554)
       
      Thread T92 created by T1 here:
          #0 0x48d95f in __interceptor_pthread_create (/home/jenkins/jenkins/workspace/wiredtiger-clang-sanitizer/build_posix/test/csuite/test_wt2719_reconfig+0x48d95f)
          #1 0x55a15f in __wt_thread_create /home/jenkins/jenkins/workspace/wiredtiger-clang-sanitizer/build_posix/../src/os_posix/os_thread.c:22:84
          #2 0x3e6ce07554 in start_thread (/lib64/libpthread.so.0+0x3e6ce07554)
       
      Thread T1 created by T0 here:
          #0 0x48d95f in __interceptor_pthread_create (/home/jenkins/jenkins/workspace/wiredtiger-clang-sanitizer/build_posix/test/csuite/test_wt2719_reconfig+0x48d95f)
          #1 0x55a15f in __wt_thread_create /home/jenkins/jenkins/workspace/wiredtiger-clang-sanitizer/build_posix/../src/os_posix/os_thread.c:22:84
          #2 0x4c35d5 in main /home/jenkins/jenkins/workspace/wiredtiger-clang-sanitizer/build_posix/test/csuite/../../../test/csuite/wt2719_reconfig/main.c:212:28
          #3 0x3e6ca206ff in __libc_start_main (/lib64/libc.so.6+0x3e6ca206ff)
       
      ==17893==ABORTING
      

      The failure didn't reproduce immediately for me locally, but did reproduce within 100 runs of the test. I suspect there is a race when switching between shared and non-shared cache via reconfigure.

      It's likely that this bug has existed for a long time, but is only now being uncovered because of additional testing.

        Attachments

          Activity

            People

            • Assignee:
              alexander.gorrod Alexander Gorrod
              Reporter:
              alexander.gorrod Alexander Gorrod
            • Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: