Details
-
Task
-
Status: Closed
-
Resolution: Done
-
None
-
None
-
None
Description
agorrod @michaelcahill
Today I pulled the latest develop and ran riak. I restarted it without removing the previous database. It immediately failed taking a divide by zero fault. Here is the (very short) stack:
163 aggressive = chunk_wait / lsm_tree->merge_min;
|
(gdb) bt
|
#0 0x00007f543cd84602 in __wt_lsm_merge_worker (vargs=0x7f53d0005620)
|
at /home/sue/riak/deps/wterl/c_src/wiredtiger-develop/src/lsm/lsm_worker.c:163
|
WT-1 0x00007f5447943c6b in start_thread () from /lib64/libpthread.so.0
|
WT-2 0x00007f544748a5ed in clone () from /lib64/libc.so.6
|
(gdb) p lsm_tree->merge_min
|
$1 = 0
|
(gdb) p chunk_wait
|
$2 = 0
|
If I restart riak with a clean database and it recreates them from scratch then I do not hit this error.