hi,

I run a thread which write records into a database continually,
another thread which do txn_checkpoint every 100M log writing. The db
type is reco and it's in a transaction env.

After writing about 30000 records, on average each record is about
0.6K, the testing program coredumped. a gdb backtrace shows:
(gdb) bt
#0 0x400b1537 in __log_flush_int () from /home/joy/lib/libdb-4.4.so
#1 0x400b0d2d in __log_flush_commit () from /home/joy/lib/libdb-4.4.so
#2 0x400b095e in __log_put () from /home/joy/lib/libdb-4.4.so
#3 0x400c78e7 in __txn_ckp_log () from /home/joy/lib/libdb-4.4.so
#4 0x400c8901 in __txn_checkpoint () from /home/joy/lib/libdb-4.4.so
#5 0x400c8678 in __txn_checkpoint_pp () from
/home/joy/lib/libdb-4.4.so
#6 0x0804a641 in check_thread__FPv (args=0x809c140) at urlrec.c:729
#7 0x400f0941 in pthread_start_thread () from
/lib/i686/libpthread.so.0

I run the same test case again and everything is OK.

any help is appreciated.

I found the bug in my own testing program. the root cause is that while
txn_checkpoint is running, the main thread is trying close the db.

sorry to bother here.

thanks