File dp.dbxml has LSN 17/1055014, past end of log at 17/678058

Question

There are many many posts in forum with similar questions on why this happens. I know how to recover from the condition but looking for potential reasons why this happens.
We open env using following flags:
DB_CREATE | DB_RECOVER | DB_INIT_LOCK | DB_INIT_LOG |
DB_INIT_TXN | DB_INIT_MPOOL | DB_THREAD | DB_REGISTER;
we run checkpoint every 5 mins in a separate thread:
txn_checkpoint(p_env, 0, 0, 0);
and remove log files:
log_archive(env, &List, DB_ARCH_REMOVE);
When some update happens we also checkpoint if data written is more than X MB (x=2).
txn_checkpoint(p_env, 2048, 0, 0);
log_archive(env, &List, DB_ARCH_REMOVE);
During shutdown we wait for all operations to finish, close the env and exit(). It runs on controlled env and 100% sure that log files do not get removed.
Please let me know if there are potential reasons why we should see it.
It also causes SIGSEGV :
file <something>.dbxml has LSN 17/1055014, past end of log at 17/678058Commonly
caused by moving a database from one database environment to another
without clearing the database LSNs, or by removing all of the logfiles
from a database environment Database handles still open at environment
close Open database handle:
dp.dbxml/secondary_document_statistics_string
Open database handle: dp.dbxml/secondary_document_index_string
Open database handle: dp.dbxml/node_nodestorage Open database handle:
dp.dbxml/secondary_document Open database handle:
dp.dbxml/secondary_dictionary Open database handle:
dp.dbxml/primary_dictionary Open database handle:
dp.dbxml/secondary_sequence Open database handle:
dp.dbxml/secondary_configuration
sigHandler(signo=11):
signo=11[SIGSEGV] code=1[SEGV_MAPERR]
eip=0xcd7f304 addr=0xdbdbdbf3
pid=2712 pid(t)=2747 ppid=2704 pgid=2700 ptid=1446356144 Executable
Path: /usr/bin
Stack Trace:
#00 [0x0cd7f304] __db_close_pp() +0x64 in
/usr/MotoAgent/thirdparty/lib/libdb_cxx-4.8.so (+104304)
#01 [0x0d50d94c] DbXml::DbWrapper::close(unsigned int) +0x30 in
/usr/MotoAgent/thirdparty/lib/libdbxml-2.5.so (+1e294c)
#02 [0x0d510acc] DbXml::DbWrapper::cleanup() +0xa0 in
/usr/MotoAgent/thirdparty/lib/libdbxml-2.5.so (+1e5acc)
#03 [0x0d5114ac] DbXml::DbWrapper::~DbWrapper() +0x38 in
/usr/MotoAgent/thirdparty/lib/libdbxml-2.5.so (+1e64ac)
#04 [0x0d5120d4] DbXml::IndexDatabase::~IndexDatabase() +0x34 in
/usr/MotoAgent/thirdparty/lib/libdbxml-2.5.so (+1e70d4)
#05 [0x0d553c3c] DbXml::SyntaxDatabase::~SyntaxDatabase() +0x60 in
/usr/MotoAgent/thirdparty/lib/libdbxml-2.5.so (+228c3c)
#06 [0x0d4f6a48] DbXml::Container::~Container() +0x114 in
/usr/MotoAgent/thirdparty/lib/libdbxml-2.5.so (+1cba48)
#07 [0x0d4efca4] DbXml::Container::release() +0xac in
/usr/MotoAgent/thirdparty/lib/libdbxml-2.5.so (+1c4ca4)
#08 [0x0d55c050] DbXml::XmlContainer::~XmlContainer() +0x28 in
/usr/MotoAgent/thirdparty/lib/libdbxml-2.5.so (+231050)
#09 [0x0e5a38a4] ma::BDBXml::initialize(std::string const&) +0x1748 in
Edited by: 871139 on Jul 7, 2011 10:05 PM

Answer 1

Hi,
As you mentioned, these errors have been discussed heavily on the Berkeley DB XML. Do a search there on "Database handles still open at environment close".
Actually, I suggest you post the issue on the Berkeley DB XML, as the solution is more related to closing/deleting specific handles (e.g. XmlDocument, XmlResults, XmlContainer, XmlManager etc). It would also be useful if you could provide a simple stand-alone test case program that reproduces the issue.
Regards,
Andrei