New hang in internal page split

XMLWordPrintableJSON

    • Type: Bug
    • Resolution: Done
    • Priority: Major - P3
    • WT2.9.2, 3.2.13, 3.4.3, 3.5.4
    • Affects Version/s: None
    • Component/s: None
    • None
    • Storage 2017-02-13
    • None

      Several of our Jenkins jobs have hung, and it appears to be related to internal page splits from the call stacks. That means it's likely due to the changes in either WT-3170 or WT-3088. There have been several changes merged recently, so I opened a new ticket.

      The call stacks I see are:

            7 select,__wt_sleep,__wt_page_in_func,__wt_page_swap_func,__wt_row_search,__cursor_row_search,__wt_btcur_search,__curfile_search,worker,start_thread,clone
            4 select,__wt_sleep,__wt_writelock,__split_internal_lock,__wt_split_multi,__evict_page_dirty_update,__wt_evict,__evict_page,__wt_cache_eviction_worker,__wt_cache_eviction_check,__cursor_enter,__curfile_enter,__cursor_func_init,__wt_btcur_insert,__curfile_insert,worker,start_thread,clone
            2 select,__wt_sleep,__wt_writelock,__split_internal_lock,__wt_split_multi,__evict_page_dirty_update,__wt_evict,__evict_page,__evict_lru_pages,__wt_evict_thread_run,__wt_thread_run,start_thread,clone
            2
            1 select,__wt_sleep,__wt_page_in_func,__wt_page_swap_func,__tree_walk_internal,__wt_tree_walk,__wt_btcur_next,__curfile_next,do_range_reads,worker,start_thread,clone
            1 select,__wt_sleep,__wt_page_in_func,__split_verify_intl,__split_internal,__split_parent_climb,__wt_split_multi,__evict_page_dirty_update,__wt_evict,__evict_page,__evict_lru_pages,__wt_evict_thread_run,__wt_thread_run,start_thread,clone
            1 pthread_join,stop_threads,execute_workload,start_run,start_all_runs,main
            1 pthread_cond_timedwait@@GLIBC_2.3.2,__wt_cond_wait_signal,__wt_cond_wait,__sweep_server,start_thread,clone
            1 pthread_cond_timedwait@@GLIBC_2.3.2,__wt_cond_wait_signal,__wt_cond_wait,__statlog_server,start_thread,clone
            1 pthread_cond_timedwait@@GLIBC_2.3.2,__wt_cond_wait_signal,__wt_cond_wait,__evict_pass,__evict_server,__wt_evict_thread_run,__wt_thread_run,start_thread,clone
      

      From the Jenkins wtperf test job: http://build.wiredtiger.com:8080/job/wiredtiger-test-wtperf/2433/

      The configuration being run is:

      ./bench/wtperf/wtperf -O ../bench/wtperf/stress//btree-split-stress.wtperf -o verbose=2
      

              Assignee:
              Michael Cahill (Inactive)
              Reporter:
              Alexander Gorrod
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

                Created:
                Updated:
                Resolved: