openldap-bugs

openldap-bugs@openldap.org

1 participants
21001 discussions

(ITS#8931) deref'ing non-existent attributes should not result in protocolError
by michael＠stroeder.com 03 Nov '18

03 Nov '18

Full_Name: Version: RE24 OS: URL: Submission from: (NULL) (46.183.103.8) In aehostd I try to limit the number of required search requests. Therefore I'm using the deref control to read group and sudoers entries referenced in service groups. If there are no such references (yet) slapd currently returns protocolError. IMO this is wrong. It should simply omit the deref response control and the client has to deal with that. (No, to save round-trips, log-space and data traffic I don't want to read the service groups first to determine whether which deref attributes are present.)

1 0

Re: (ITS#8752) MMR delta-sync deadlock using slapd.conf
by hyc＠symas.com 01 Nov '18

01 Nov '18

ondra(a)mistotebe.net wrote: > On Thu, Oct 05, 2017 at 03:32:38AM +0000, quanah(a)openldap.org wrote: >> Full_Name: Quanah Gibson-Mount >> Version: 2.4.45 >> OS: Linux >> URL: ftp://ftp.openldap.org/incoming/ >> Submission from: (NULL) (47.208.148.239) >> >> >> There is a reproducible lockup with delta-sync based multimaster replication. >> The more masters exist the easier it is to trigger, with 100% reproducibility at >> 4 masters. >> >> Generally: >> >> Load a database onto a server with no syncprov/accesslog configuration. >> >> Stop slapd >> >> Reconfigure slapd.conf to include syncprov/accesslog overlays, and cn=accesslog >> DB for changes, and replication agreements with other servers >> >> Configure other servers for syncprov/accesslog, but no database >> >> start slapd on all servers >> >> execute a MOD operation sequentially across the servers (1, 2, 3, 4) >> >> slapd will deadlock on server 3 or 4 100% of the time > > This is still an issue at least in master - running the its8752 > regression test often results in the following deadlock: This says that thread 4 is still holding the LMDB writer mutex while running inside the accesslog overlay. Which it should not be, since back-mdb acquires and commits its LMDB transactions before letting any other module have control. > > (gdb) deadlock > Cycles: > Thread #5 (LWP 8024) in mdb_txn_renew0() at ./../../../libraries/liblmdb/mdb.c:2749 > Thread #4 (LWP 8023) in accesslog_op_mod() at accesslog.c:1988 > deadlock from root > Thread #3 (LWP 8022) in mdb_txn_renew0() at ./../../../libraries/liblmdb/mdb.c:2749 > > (gdb) thread apply all bt > > Thread 9 (Thread 0x7f9199ffd700 (LWP 15035)): > #3 0x00007f91b815f20d in ldap_pvt_thread_cond_wait (cond=0xe11cb8, mutex=0xe11c90) at thr_posix.c:288 > > Thread 8 (Thread 0x7f91a09fa700 (LWP 15034)): > #3 0x00007f91b815f20d in ldap_pvt_thread_cond_wait (cond=0xe11cb8, mutex=0xe11c90) at thr_posix.c:288 > > Thread 7 (Thread 0x7f91b12b7700 (LWP 8026)): > #3 0x00007f91b815f20d in ldap_pvt_thread_cond_wait (cond=0xe11cb8, mutex=0xe11c90) at thr_posix.c:288 > > Thread 6 (Thread 0x7f91b1ab8700 (LWP 8025)): > #3 0x00007f91b815f20d in ldap_pvt_thread_cond_wait (cond=0xe11cb8, mutex=0xe11c90) at thr_posix.c:288 > > Thread 5 (Thread 0x7f91b22b9700 (LWP 8024)): > #0 0x00007f91b5dda352 in __pthread_mutex_lock_full [waiting on thread 4 (LWP 8023)] (mutex=0x7f91b8114040) at ../nptl/pthread_mutex_lock.c:313 > #1 0x000000000051a881 in mdb_txn_renew0 (txn=0x1027ac0) at ./../../../libraries/liblmdb/mdb.c:2749 > #2 0x000000000051afab in mdb_txn_begin (env=0xf24e10, parent=0x0, flags=0, ret=0x7f91b22b7190) at ./../../../libraries/liblmdb/mdb.c:2907 > #3 0x00000000005792f4 in mdb_opinfo_get (op=Modify request = {...}, mdb=0x7f91b543b010, rdonly=0, moip=0x7f91b22b7178) at id2entry.c:740 > #4 0x000000000052f989 in mdb_modify (op=Modify request = {...}, rs=0x7f91b22b89b8) at modify.c:584 > #5 0x00000000004f6d65 in overlay_op_walk (op=Modify request = {...}, rs=0x7f91b22b89b8, which=op_modify, oi=0xe29b10, on=0x0) at backover.c:706 > #6 0x00000000004f9d61 in over_op_func (op=Modify request = {...}, rs=0x7f91b22b89b8, which=op_modify) at backover.c:766 > #7 0x00000000004f8fb2 in over_op_modify (op=Modify request = {...}, rs=0x7f91b22b89b8) at backover.c:808 > #8 0x0000000000472e4e in fe_op_modify (op=Modify request = {...}, rs=0x7f91b22b89b8) at modify.c:307 > #9 0x0000000000471bf5 in do_modify (op=Modify request = {...}, rs=0x7f91b22b89b8) at modify.c:177 > #10 0x000000000044e1ce in connection_operation (ctx=0x7f91b22b8b78, arg_v=0x7f919c100fa0) at connection.c:1169 > #11 0x000000000044c3a1 in connection_read_thread (ctx=0x7f91b22b8b78, argv=0xf) at connection.c:1326 > > Thread 4 (Thread 0x7f91b2aba700 (LWP 8023)): > #2 0x00007f91b815f275 in ldap_pvt_thread_mutex_lock [waiting on thread 5 (LWP 8024)] (mutex=0xe7c328) at thr_posix.c:307 > #3 0x00007f91b54fffbc in accesslog_op_mod (op=Modify request = {...}, rs=0x7f91b2ab8bd8) at accesslog.c:1988 > #4 0x00000000004f6cc1 in overlay_op_walk (op=Modify request = {...}, rs=0x7f91b2ab8bd8, which=op_modify, oi=0xe29b10, on=0xe7c0c0) at backover.c:691 > #5 0x00000000004f9d61 in over_op_func (op=Modify request = {...}, rs=0x7f91b2ab8bd8, which=op_modify) at backover.c:766 > #6 0x00000000004f8fb2 in over_op_modify (op=Modify request = {...}, rs=0x7f91b2ab8bd8) at backover.c:808 > #7 0x00000000004ec259 in syncrepl_updateCookie (si=0xe7aa90, op=Modify request = {...}, syncCookie=0x7f91b2ab91f0, save=1) at syncrepl.c:4051 > #8 0x00000000004e978e in do_syncrep2 (op=Modify request = {...}, si=0xe7aa90) at syncrepl.c:1206 > #9 0x00000000004e2556 in do_syncrepl (ctx=0x7f91b2ab9b78, arg=0xe7b030) at syncrepl.c:1604 > > Thread 3 (Thread 0x7f91b32bb700 (LWP 8022)): > #0 0x00007f91b5dda352 in __pthread_mutex_lock_full [waiting on thread 4 (LWP 8023)] (mutex=0x7f91b8114040) at ../nptl/pthread_mutex_lock.c:313 > #1 0x000000000051a881 in mdb_txn_renew0 (txn=0x1027ac0) at ./../../../libraries/liblmdb/mdb.c:2749 > #2 0x000000000051afab in mdb_txn_begin (env=0xf24e10, parent=0x0, flags=0, ret=0x7f91a4001170) at ./../../../libraries/liblmdb/mdb.c:2907 > #3 0x00000000005792f4 in mdb_opinfo_get (op=Search request = {...}, mdb=0x7f91b543b010, rdonly=0, moip=0xe7b348) at id2entry.c:740 > #4 0x000000000057989d in mdb_txn (op=Search request = {...}, txnop=1, ptr=0xe7b348) at id2entry.c:812 > #5 0x00000000004ee059 in syncrepl_entry (si=0xe7b0f0, op=Search request = {...}, entry="dc=example,dc=com", modlist=0x7f91b32ba1a8, syncstate=1, syncUUID=0x7f91b32ba140, syncCSN=0x0) at syncrepl.c:3087 > #6 0x00000000004e8cac in do_syncrep2 (op=Search request = {...}, si=0xe7b0f0) at syncrepl.c:1056 > #7 0x00000000004e2556 in do_syncrepl (ctx=0x7f91b32bab78, arg=0xe7b6d0) at syncrepl.c:1604 > > Thread 2 (Thread 0x7f91b3abc700 (LWP 8021)): > #0 0x00007f91b5d0b207 in epoll_wait (epfd=6, events=0xdebc50, maxevents=1024, timeout=7000) at ../sysdeps/unix/sysv/linux/epoll_wait.c:30 > #1 0x000000000044591a in slapd_daemon_task (ptr=0x1232700) at daemon.c:2745 > > Thread 1 (Thread 0x7f91b5761780 (LWP 8006)): > #1 0x00007f91b815f11d in ldap_pvt_thread_join (thread=140263761364736, thread_return=0x0) at thr_posix.c:208 > #2 0x0000000000443a1b in slapd_daemon () at daemon.c:3157 > #3 0x000000000041ec5b in main (argc=8, argv=0x7ffcee967dc8) at main.c:1018 > -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

1 0

Re: (ITS#8752) MMR delta-sync deadlock using slapd.conf
by ondra＠mistotebe.net 01 Nov '18

01 Nov '18

On Thu, Oct 05, 2017 at 03:32:38AM +0000, quanah(a)openldap.org wrote: > Full_Name: Quanah Gibson-Mount > Version: 2.4.45 > OS: Linux > URL: ftp://ftp.openldap.org/incoming/ > Submission from: (NULL) (47.208.148.239) > > > There is a reproducible lockup with delta-sync based multimaster replication. > The more masters exist the easier it is to trigger, with 100% reproducibility at > 4 masters. > > Generally: > > Load a database onto a server with no syncprov/accesslog configuration. > > Stop slapd > > Reconfigure slapd.conf to include syncprov/accesslog overlays, and cn=accesslog > DB for changes, and replication agreements with other servers > > Configure other servers for syncprov/accesslog, but no database > > start slapd on all servers > > execute a MOD operation sequentially across the servers (1, 2, 3, 4) > > slapd will deadlock on server 3 or 4 100% of the time This is still an issue at least in master - running the its8752 regression test often results in the following deadlock: (gdb) deadlock Cycles: Thread #5 (LWP 8024) in mdb_txn_renew0() at ./../../../libraries/liblmdb/mdb.c:2749 Thread #4 (LWP 8023) in accesslog_op_mod() at accesslog.c:1988 deadlock from root Thread #3 (LWP 8022) in mdb_txn_renew0() at ./../../../libraries/liblmdb/mdb.c:2749 (gdb) thread apply all bt Thread 9 (Thread 0x7f9199ffd700 (LWP 15035)): #3 0x00007f91b815f20d in ldap_pvt_thread_cond_wait (cond=0xe11cb8, mutex=0xe11c90) at thr_posix.c:288 Thread 8 (Thread 0x7f91a09fa700 (LWP 15034)): #3 0x00007f91b815f20d in ldap_pvt_thread_cond_wait (cond=0xe11cb8, mutex=0xe11c90) at thr_posix.c:288 Thread 7 (Thread 0x7f91b12b7700 (LWP 8026)): #3 0x00007f91b815f20d in ldap_pvt_thread_cond_wait (cond=0xe11cb8, mutex=0xe11c90) at thr_posix.c:288 Thread 6 (Thread 0x7f91b1ab8700 (LWP 8025)): #3 0x00007f91b815f20d in ldap_pvt_thread_cond_wait (cond=0xe11cb8, mutex=0xe11c90) at thr_posix.c:288 Thread 5 (Thread 0x7f91b22b9700 (LWP 8024)): #0 0x00007f91b5dda352 in __pthread_mutex_lock_full [waiting on thread 4 (LWP 8023)] (mutex=0x7f91b8114040) at ../nptl/pthread_mutex_lock.c:313 #1 0x000000000051a881 in mdb_txn_renew0 (txn=0x1027ac0) at ./../../../libraries/liblmdb/mdb.c:2749 #2 0x000000000051afab in mdb_txn_begin (env=0xf24e10, parent=0x0, flags=0, ret=0x7f91b22b7190) at ./../../../libraries/liblmdb/mdb.c:2907 #3 0x00000000005792f4 in mdb_opinfo_get (op=Modify request = {...}, mdb=0x7f91b543b010, rdonly=0, moip=0x7f91b22b7178) at id2entry.c:740 #4 0x000000000052f989 in mdb_modify (op=Modify request = {...}, rs=0x7f91b22b89b8) at modify.c:584 #5 0x00000000004f6d65 in overlay_op_walk (op=Modify request = {...}, rs=0x7f91b22b89b8, which=op_modify, oi=0xe29b10, on=0x0) at backover.c:706 #6 0x00000000004f9d61 in over_op_func (op=Modify request = {...}, rs=0x7f91b22b89b8, which=op_modify) at backover.c:766 #7 0x00000000004f8fb2 in over_op_modify (op=Modify request = {...}, rs=0x7f91b22b89b8) at backover.c:808 #8 0x0000000000472e4e in fe_op_modify (op=Modify request = {...}, rs=0x7f91b22b89b8) at modify.c:307 #9 0x0000000000471bf5 in do_modify (op=Modify request = {...}, rs=0x7f91b22b89b8) at modify.c:177 #10 0x000000000044e1ce in connection_operation (ctx=0x7f91b22b8b78, arg_v=0x7f919c100fa0) at connection.c:1169 #11 0x000000000044c3a1 in connection_read_thread (ctx=0x7f91b22b8b78, argv=0xf) at connection.c:1326 Thread 4 (Thread 0x7f91b2aba700 (LWP 8023)): #2 0x00007f91b815f275 in ldap_pvt_thread_mutex_lock [waiting on thread 5 (LWP 8024)] (mutex=0xe7c328) at thr_posix.c:307 #3 0x00007f91b54fffbc in accesslog_op_mod (op=Modify request = {...}, rs=0x7f91b2ab8bd8) at accesslog.c:1988 #4 0x00000000004f6cc1 in overlay_op_walk (op=Modify request = {...}, rs=0x7f91b2ab8bd8, which=op_modify, oi=0xe29b10, on=0xe7c0c0) at backover.c:691 #5 0x00000000004f9d61 in over_op_func (op=Modify request = {...}, rs=0x7f91b2ab8bd8, which=op_modify) at backover.c:766 #6 0x00000000004f8fb2 in over_op_modify (op=Modify request = {...}, rs=0x7f91b2ab8bd8) at backover.c:808 #7 0x00000000004ec259 in syncrepl_updateCookie (si=0xe7aa90, op=Modify request = {...}, syncCookie=0x7f91b2ab91f0, save=1) at syncrepl.c:4051 #8 0x00000000004e978e in do_syncrep2 (op=Modify request = {...}, si=0xe7aa90) at syncrepl.c:1206 #9 0x00000000004e2556 in do_syncrepl (ctx=0x7f91b2ab9b78, arg=0xe7b030) at syncrepl.c:1604 Thread 3 (Thread 0x7f91b32bb700 (LWP 8022)): #0 0x00007f91b5dda352 in __pthread_mutex_lock_full [waiting on thread 4 (LWP 8023)] (mutex=0x7f91b8114040) at ../nptl/pthread_mutex_lock.c:313 #1 0x000000000051a881 in mdb_txn_renew0 (txn=0x1027ac0) at ./../../../libraries/liblmdb/mdb.c:2749 #2 0x000000000051afab in mdb_txn_begin (env=0xf24e10, parent=0x0, flags=0, ret=0x7f91a4001170) at ./../../../libraries/liblmdb/mdb.c:2907 #3 0x00000000005792f4 in mdb_opinfo_get (op=Search request = {...}, mdb=0x7f91b543b010, rdonly=0, moip=0xe7b348) at id2entry.c:740 #4 0x000000000057989d in mdb_txn (op=Search request = {...}, txnop=1, ptr=0xe7b348) at id2entry.c:812 #5 0x00000000004ee059 in syncrepl_entry (si=0xe7b0f0, op=Search request = {...}, entry="dc=example,dc=com", modlist=0x7f91b32ba1a8, syncstate=1, syncUUID=0x7f91b32ba140, syncCSN=0x0) at syncrepl.c:3087 #6 0x00000000004e8cac in do_syncrep2 (op=Search request = {...}, si=0xe7b0f0) at syncrepl.c:1056 #7 0x00000000004e2556 in do_syncrepl (ctx=0x7f91b32bab78, arg=0xe7b6d0) at syncrepl.c:1604 Thread 2 (Thread 0x7f91b3abc700 (LWP 8021)): #0 0x00007f91b5d0b207 in epoll_wait (epfd=6, events=0xdebc50, maxevents=1024, timeout=7000) at ../sysdeps/unix/sysv/linux/epoll_wait.c:30 #1 0x000000000044591a in slapd_daemon_task (ptr=0x1232700) at daemon.c:2745 Thread 1 (Thread 0x7f91b5761780 (LWP 8006)): #1 0x00007f91b815f11d in ldap_pvt_thread_join (thread=140263761364736, thread_return=0x0) at thr_posix.c:208 #2 0x0000000000443a1b in slapd_daemon () at daemon.c:3157 #3 0x000000000041ec5b in main (argc=8, argv=0x7ffcee967dc8) at main.c:1018 -- Ondřej Kuzník Senior Software Engineer Symas Corporation http://www.symas.com Packaged, certified, and supported LDAP solutions powered by OpenLDAP

1 0

Re: (ITS#8930) Setting source interface of LDAP session the ldap client uses to connect to server
by ryan＠nardis.ca 29 Oct '18

29 Oct '18

Hello Adrian, I think someone is already working on this feature, please see: ITS#8847 https://openldap.org/its?findid=8847 ITS#8654 https://openldap.org/its?findid=8654 http://www.openldap.org/lists/openldap-devel/201810/msg00008.html

1 0

(ITS#8930) Setting source interface of LDAP session the ldap client uses to connect to server
by ac1471＠att.com 29 Oct '18

29 Oct '18

Full_Name: adrian cepleanu Version: 0.5.3 OS: Ubuntu 16.04LTS URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (144.160.130.91) In a system with multiple interfaces software initiating network connections may choose the source interface (or IP address) that the request is initiated from. I am looking for such a setting in the ldap-auth-client for Linux but cannot find it. Perhaps it doesn't exist. If the latter please advise if there is a way to submit a feature request. thanks

1 0

Re: (ITS#8866) RFE: slapo-unique to return filter used in diagnostic message
by michael＠stroeder.com 26 Oct '18

26 Oct '18

On 10/26/18 4:21 PM, Ondřej Kuzník wrote: > Yes, but `key` had already been freed a few lines earlier and using > o_tmpalloc reliably exposes the issue where ch_malloc just masks it. Ouch! > This is now fixed in master. Thanks. Everything now works like a charm also with RE24. Ciao, Michael.

1 0

Re: (ITS#8866) RFE: slapo-unique to return filter used in diagnostic message
by ondra＠mistotebe.net 26 Oct '18

26 Oct '18

On Fri, Oct 26, 2018 at 08:12:13AM +0000, michael(a)stroeder.com wrote: > On 10/26/18 4:01 AM, Quanah Gibson-Mount wrote: >> This patch has been applied to OpenLDAP HEAD with a correction to the >> memory allocation functions. I'll discuss with Howard about whether or >> not to add it to 2.4.47. > > With your memory allocation correction, op->o_tmpalloc() and > op->o_tmpfree(), this does not work as expected with 2.4.x. It does > *not* return the filter used for uniqueness check. > (I did not test master yet.) > > [...] > Result with patch backported from git master using op->o_tmpalloc() and > op->o_tmpfree(): > > non-unique attributes found with non-unique attribute > ^^^^^^^^^^^^^^^^^^^^ > Seems to repeat parts of the buffer? > > In my original patch the use of ch_malloc() and ch_free() was simply > copied from other overlay code. There are many occurences of ch_malloc() > and ch_free() throughout the whole code. Yes, but `key` had already been freed a few lines earlier and using o_tmpalloc reliably exposes the issue where ch_malloc just masks it. This is now fixed in master. -- Ondřej Kuzník Senior Software Engineer Symas Corporation http://www.symas.com Packaged, certified, and supported LDAP solutions powered by OpenLDAP

1 0

Re: (ITS#8866) RFE: slapo-unique to return filter used in diagnostic message
by michael＠stroeder.com 26 Oct '18

26 Oct '18

On 10/26/18 4:01 AM, Quanah Gibson-Mount wrote: > --On Monday, July 30, 2018 6:58 PM +0000 michael(a)stroeder.com wrote: > >> Can someone correct the subject line of the ticket? >> >> Should of course mention slapo-unique instead of slapo-constraint. > > This patch has been applied to OpenLDAP HEAD with a correction to the > memory allocation functions. I'll discuss with Howard about whether or > not to add it to 2.4.47. With your memory allocation correction, op->o_tmpalloc() and op->o_tmpfree(), this does not work as expected with 2.4.x. It does *not* return the filter used for uniqueness check. (I did not test master yet.) Example for expected diagnostic message (with my original patch applied to 2.4.46): non-unique attributes found with (|(gidNumber=30041)) Result with patch backported from git master using op->o_tmpalloc() and op->o_tmpfree(): non-unique attributes found with non-unique attribute ^^^^^^^^^^^^^^^^^^^^ Seems to repeat parts of the buffer? In my original patch the use of ch_malloc() and ch_free() was simply copied from other overlay code. There are many occurences of ch_malloc() and ch_free() throughout the whole code. Does op->o_tmpalloc() and op->o_tmpfree() work correctly in RE24 branch? Ciao, Michael.

1 0

Re: (ITS#8925) Please remove the entry for mirror.switch.ch from the Download page
by quanah＠symas.com 25 Oct '18

25 Oct '18

--On Thursday, October 18, 2018 10:23 PM -0400 Thomas Lenggenhager <thomas.lenggenhager(a)switch.ch> wrote: > Hello Quanah > > Many thanks for your action and answer. In the mean time I noticed that > in this file > ftp://ftp.openldap.org/pub/OpenLDAP/MIRRORS > the mirrors are still listed. Could you please update that as well? Thanks, This is now fixed as well. :) Warm regards, Quanah -- Quanah Gibson-Mount Product Architect Symas Corporation Packaged, certified, and supported LDAP solutions powered by OpenLDAP: <http://www.symas.com>

1 0

Re: (ITS#8866) RFE: slapo-unique to return filter used in diagnostic message
by quanah＠symas.com 25 Oct '18

25 Oct '18

--On Monday, July 30, 2018 6:58 PM +0000 michael(a)stroeder.com wrote: > Can someone correct the subject line of the ticket? > > Should of course mention slapo-unique instead of slapo-constraint. This patch has been applied to OpenLDAP HEAD with a correction to the memory allocation functions. I'll discuss with Howard about whether or not to add it to 2.4.47. --Quanah -- Quanah Gibson-Mount Product Architect Symas Corporation Packaged, certified, and supported LDAP solutions powered by OpenLDAP: <http://www.symas.com>

1 0

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

openldap-bugs