I encountered 2 error

Question

the code:
u_int32_t id;
u_int32_t dataSize;
char dataBuf[32];
dataSize = getData(dataBuf,..);
Dbt key(&id, sizeof(id));
key.set_flag(DB_DBT_USERMEM);
key.set_ulen(sizeof(id));
Dbt data(buffer, buf_size);
Dbc* cursorp;
db.cursor(NULL, &cursorp, 0);
int ret = cursorp->get(&key, &data, DB_GET_BOTH);
if (0 == ret) {
// here I saw dataBuf had been changed, it was be filled with the data in db
// but I had not called data.set_flag(DB_DBT_USERMEM), was it a bug? or it was the design?
// my dup_compare only compare part of data
ret = cursor.del(0); assert(0 == ret);
// after I found the error, I call my getData(...) again here:
newID = updateID(....);
dataSize = getData(dataBuf,..);
data.set_data(dataBuf);
data.set_size(dataSize);
key.set_data(&newID);
ret = db.put(NULL, &key, &data, 0); assert(0 == ret);
ret = cursorp->close(); assert(0 == ret);
another error, when i call db.sync(0), I get a message:
memp_sync: correcting dirty count 7 4
I used multi different Db in one DbEnv, and every Db was used in different thread, DbEnv was opened with DB_THREAD, and Db was opened without DB_THREAD. when muti thread was joined, I call every db.sync(0), then I saw this message.

Answer 1

Hi,
Could you post a ready to be run small reproducible testcase that demonstrate these issues? If so, mention the Berkeley DB release you're using and on what platform/OS.
Regards,
Andrei