OpenLDAP 2.4.43 Engineering Fixed liblber remove obsolete assert (ITS#8240, ITS#8301) Fixed libldap file URLs on windows (ITS#8273) Fixed libldap microsecond timer for windows (ITS#8295) Fixed slap tools minor one time memory leak (ITS#8082) Fixed slapd to avoid redundant processing of abandon ops (ITS#8232) Fixed slapd syncrepl segv when present list is NULL (ITS#8231, ITS#8042) Fixed slapd segfault with invalid SASL URI (ITS#8218) Fixed slapd configuration parser with unbalanced quotes (ITS#8233) Fixed slapd syncrepl check with config db on windows (ITS#8277) Fixed slapd with mod Increment and inherited attribute type (ITS#8289) Fixed slapd-ldap SEGV after failed retry (ITS#8173) Fixed slapd-ldap to skip client controls in ldap_back_entry_get (ITS#8244) Fixed slapd-null to have an option to return a search entry (ITS#8249) Fixed slapd-relay to correctly handle quoted options (ITS#8284) Fixed slapo-accesslog delta-sync MMR with interrupted refresh phase (ITS#8281) Fixed slapo-dds segfault when using slapo-memberof (ITS#8133) Fixed slapo-ppolicy to allow purging of stale pwdFailureTime attributes (ITS#8185) Fixed slapo-ppolicy to release entry on failure (ITS#7537) Fixed slapo-ppolicy to fall back to default policy if there is a parsing error (ITS#8234) Fixed slapo-syncprov with interrupted refresh phase (ITS#8281) Fixed slapo-refint with subtree renames (ITS#8220) Fixed slapo-rwm missing olcDropUnrequested attribute (ITS#7889) Fixed slapo-rwm parsing to avoid double-escaping rewrite rules (ITS#7964) Build Environment Fixed slapd-tester EOL handling in test output for windows (ITS#8280) Fixed slapd-tester executable suffix for windows (ITS#8216) Fixed ldif-filter option parsing (ITS#8292) Contrib Added libnettle support to pw-pbkdf2 (ITS#8198) Fixed smbk5pwd compiler warnings with libnettle (ITS#8235) Fixed passwd symbol collisions with other crypto libraries (ITS#8294) Documentation Updated guide to reflect changes to how TLS is handled with syncrepl (ITS#7897)
LMDB 0.9.17 Release Engineering Fix ITS#7377 catch calloc failure Fix ITS#8237 regression from ITS#7589 Fix ITS#8238 page_split for DUPFIXED pages Fix ITS#8221 MDB_PAGE_FULL on delete/rebalance Fix ITS#8258 rebalance/split assert Fix ITS#8263 cursor_put cursor tracking Fix ITS#8264 cursor_del cursor tracking Fix ITS#8299 mdb_del cursor tracking Fix ITS#8300 mdb_del cursor tracking Fix ITS#8304 mdb_del cursor tracking Fix ITS#7771 fakepage cursor tracking Fix ITS#7789 ensure mapsize >= pages in use Fix ITS#7971 mdb_txn_renew0() new reader slots Fix ITS#7969 use __sync_synchronize on non-x86 Added mdb_txn_id() (ITS#7994) Added robust mutex support Miscellaneous cleanup/simplification Build Create install dirs if needed (ITS#8256) Fix ThreadProc decl on Win32/MSVC (ITS#8270) Added ssize_t typedef for MSVC (ITS#8067) Use ANSI apis on Windows (ITS#8069)
--
Quanah Gibson-Mount Platform Architect Zimbra, Inc. -------------------- Zimbra :: the leader in open source messaging and collaboration
All tests run through without failures on OS X El Capitan / 10.11.1 with the following configuration:
export CC="gcc -isystem /usr/local/opt/berkeley-db4/include -L/usr/local/opt/berkeley-db4/lib -L/usr/local/opt/berkeley-db4/lib”
./configure --enable-overlays --enable-ldap --enable-meta
BerkeleyDB is version 4.8.30 from Homebrew.
jens
Jens Vagelpohl wrote:
All tests run through without failures on OS X El Capitan / 10.11.1 with the following configuration:
export CC="gcc -isystem /usr/local/opt/berkeley-db4/include -L/usr/local/opt/berkeley-db4/lib -L/usr/local/opt/berkeley-db4/lib”
./configure --enable-overlays --enable-ldap --enable-meta
BerkeleyDB is version 4.8.30 from Homebrew.
Thanks for the report. Why are you still using BDB 4.8? I've run into locking issues with 4.8.30 recently, had to upgrade to 5.3.28.
On 10 Nov 2015, at 10:03 , Howard Chu hyc@symas.com wrote:
Jens Vagelpohl wrote:
export CC="gcc -isystem /usr/local/opt/berkeley-db4/include -L/usr/local/opt/berkeley-db4/lib -L/usr/local/opt/berkeley-db4/lib”
./configure --enable-overlays --enable-ldap --enable-meta
BerkeleyDB is version 4.8.30 from Homebrew.
Thanks for the report. Why are you still using BDB 4.8? I've run into locking issues with 4.8.30 recently, had to upgrade to 5.3.28.
Hi Howard,
It’s simple convenience. Homebrew offers either BDB 4.8.30 as package berkeley-db4 or BDB 6.1.26 as package berkeley-db. I don’t recall any discussion here about BDB 6.x being OK so it seemed OK to stick with 4.8.30 for the moment since it’s just for testing.
For production I don’t use BDB myself, just MDB.
jens
--On Tuesday, November 10, 2015 10:43 AM +0100 Jens Vagelpohl jens@dataflake.org wrote:
Thanks for the report. Why are you still using BDB 4.8? I've run into locking issues with 4.8.30 recently, had to upgrade to 5.3.28.
Hi Howard,
It's simple convenience. Homebrew offers either BDB 4.8.30 as package berkeley-db4 or BDB 6.1.26 as package berkeley-db. I don't recall any discussion here about BDB 6.x being OK so it seemed OK to stick with 4.8.30 for the moment since it's just for testing.
BDB 6.x had a licensing change that makes it non compatible with OSS software. I'm surprised Homebrew ships it at all. Essentially, the 5.x series is the last series of BDB that's usable with OpenLDAP, etc. If it's just for testing, 4.8's fine. ;)
--Quanah
--
Quanah Gibson-Mount Platform Architect Zimbra, Inc. -------------------- Zimbra :: the leader in open source messaging and collaboration
Tried to build it on CentOS 5 32bits (LTB packaging spec file) and got this error:
gcc -DOPENLDAP_FD_SETSIZE=4096 -O2 -g -I../../../include -I../../../include -I.. -I./.. -I./../../../libraries/liblmdb -I/usr/local/berkeleydb/include -I/usr/kerberos/include -c ./../../../libraries/liblmdb/mdb.c -o mdb.o ./../../../libraries/liblmdb/mdb.c: In function 'mdb_env_setup_locks': ./../../../libraries/liblmdb/mdb.c:4579: error: 'PTHREAD_MUTEX_ROBUST' undeclared (first use in this function) ./../../../libraries/liblmdb/mdb.c:4579: error: (Each undeclared identifier is reported only once ./../../../libraries/liblmdb/mdb.c:4579: error: for each function it appears in.) make[3]: *** [mdb.lo] Error 1 make[3]: Leaving directory `/home/clement/build/BUILD/openldap-44982bf/servers/slapd/back-mdb' make[2]: *** [.backend] Error 1 make[2]: Leaving directory `/home/clement/build/BUILD/openldap-44982bf/servers/slapd' make[1]: *** [all-common] Error 1 make[1]: Leaving directory `/home/clement/build/BUILD/openldap-44982bf/servers' make: *** [all-common] Error 1 erreur: Mauvais status de sortie pour /var/tmp/rpm-tmp.20759 (%build)
Could it be a problem on my build machine?
On Tue, 10 Nov 2015, Cl?ment OUDOT wrote:
Could it be a problem on my build machine?
Not directly IMO. Some (older?) pthreads implementations are PTHREAD_MUTEX_ROBUST_NP, etc.
one thing to try would be:
#ifdef PTHREAD_MUTEX_ROBUST_NP #define PTHREAD_MUTEX_ROBUST PTHREAD_MUTEX_ROBUST_NP #define pthread_mutexattr_setrobust pthread_mutexattr_setrobust_np #define pthread_mutex_consistent pthread_mutex_consistent_np #endif
which might also need:
|| (rc = pthread_mutexattr_setrobust(&mattr, PTHREAD_MUTEX_ROBUST)) + || (rc = pthread_mutexattr_setprotocol(&mattr, PTHREAD_PRIO_INHERIT)) || (rc = pthread_mutex_init(env->me_txns->mti_rmutex, &mattr))
or you can just -DMDB_USE_ROBUST=0, but I don't think the platform is *that* hopeless...
Aaron Richton wrote:
On Tue, 10 Nov 2015, Cl?ment OUDOT wrote:
Could it be a problem on my build machine?
Not directly IMO. Some (older?) pthreads implementations are PTHREAD_MUTEX_ROBUST_NP, etc.
one thing to try would be:
#ifdef PTHREAD_MUTEX_ROBUST_NP #define PTHREAD_MUTEX_ROBUST PTHREAD_MUTEX_ROBUST_NP #define pthread_mutexattr_setrobust pthread_mutexattr_setrobust_np #define pthread_mutex_consistent pthread_mutex_consistent_np #endif
Unfortunately, PTHREAD_MUTEX_ROBUST_NP is an enum, not a macro, so #ifdef won't work to detect it.
The attached patch should work. Please report back; I won't merge it unless we know it actually helps.
which might also need:
|| (rc = pthread_mutexattr_setrobust(&mattr,PTHREAD_MUTEX_ROBUST))
|| (rc = pthread_mutexattr_setprotocol(&mattr,PTHREAD_PRIO_INHERIT)) || (rc = pthread_mutex_init(env->me_txns->mti_rmutex, &mattr))
or you can just -DMDB_USE_ROBUST=0, but I don't think the platform is *that* hopeless...
Le 11/11/2015 01:09, Howard Chu a écrit :
Aaron Richton wrote:
On Tue, 10 Nov 2015, Cl?ment OUDOT wrote:
Could it be a problem on my build machine?
Not directly IMO. Some (older?) pthreads implementations are PTHREAD_MUTEX_ROBUST_NP, etc.
one thing to try would be:
#ifdef PTHREAD_MUTEX_ROBUST_NP #define PTHREAD_MUTEX_ROBUST PTHREAD_MUTEX_ROBUST_NP #define pthread_mutexattr_setrobust pthread_mutexattr_setrobust_np #define pthread_mutex_consistent pthread_mutex_consistent_np #endif
Unfortunately, PTHREAD_MUTEX_ROBUST_NP is an enum, not a macro, so #ifdef won't work to detect it.
The attached patch should work. Please report back; I won't merge it unless we know it actually helps.
Hello Howard,
I confirm the patch works for CentOS5/RHEL5 32 bits and 64 bits.
I can build without the patch on CentOS 6.
I can now take my plane for LDAPCon ;) See you soon.
-
Aaron Richton -
Clément OUDOT -
Howard Chu -
Jens Vagelpohl -
Quanah Gibson-Mount