openldap-bugs

openldap-bugs@openldap.org

1 participants
21003 discussions

Re: (ITS#7875) lmdb MDB_GET_BOTH unexpected success
by hyc＠symas.com 11 Jun '14

11 Jun '14

mail(a)arnedebruijn.nl wrote: > Full_Name: Arne de Bruijn > Version: mdb.master a93810c > OS: Linux > URL: ftp://ftp.openldap.org/incoming/ > Submission from: (NULL) (94.210.130.16) > > > MDB_GET_BOTH seems to unexpectedly return success if a data-value before the > first data-value for a key is requested. Thanks for the report. This appears to have been broken due to commit 18a07eb7c2dc33372455a6040984cd6b699b41a5 for ITS#7681. Fixed now in mdb.master > > The example program sets 1:b and 1:c and requests 1:a. It prints: > get_both: key=1 val=b > > While I would expect: > get_both: err=MDB_NOTFOUND: No matching key/data pair found > > (curiously it works when it only sets 1:b) > > #include <stdlib.h> > #include <stdio.h> > #include <string.h> > #include <unistd.h> > #include "lmdb.h" > > void set_val(MDB_val *v, char *s) { > v->mv_size = strlen(s); > v->mv_data = s; > } > > int main() { > int rc; > MDB_env *menv = NULL; > MDB_dbi mdbi; > MDB_txn *mtxn = NULL; > MDB_val key, val; > MDB_cursor *mc; > > if ((rc = mdb_env_create(&menv))) > goto err; > if ((rc = unlink("testdb"))) > goto err; > if ((rc = mdb_env_open(menv, "testdb", MDB_NOSUBDIR, 0666))) > goto err; > if ((rc = mdb_txn_begin(menv, NULL, 0, &mtxn))) > goto err; > if ((rc = mdb_dbi_open(mtxn, NULL, MDB_DUPSORT | MDB_CREATE, &mdbi))) > goto err; > set_val(&key, "1"); > set_val(&val, "b"); > if ((rc = mdb_put(mtxn, mdbi, &key, &val, 0))) > goto err; > set_val(&key, "1"); > set_val(&val, "c"); > if ((rc = mdb_put(mtxn, mdbi, &key, &val, 0))) > goto err; > > if ((rc = mdb_cursor_open(mtxn, mdbi, &mc))) > goto err; > > set_val(&key, "1"); > set_val(&val, "a"); > rc = mdb_cursor_get(mc, &key, &val, MDB_GET_BOTH); > if (rc) > printf("get_both: err=%s\n", mdb_strerror(rc)); > else > printf("get_both: key=%.*s val=%.*s\n", key.mv_size, (char *)key.mv_data, > val.mv_size, (char *)val.mv_data); > > mdb_cursor_close(mc); > > if ((rc = mdb_txn_commit(mtxn))) { > mdb_dbi_close(menv, mdbi); > goto err; > } > mdb_env_close(menv); > return 0; > err: > if (mtxn) > mdb_txn_abort(mtxn); > if (menv) > mdb_env_close(menv); > fprintf(stderr, "Error: %s\n", mdb_strerror(rc)); > return 1; > } > > -- -- 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

(ITS#7875) lmdb MDB_GET_BOTH unexpected success
by mail＠arnedebruijn.nl 11 Jun '14

11 Jun '14

Full_Name: Arne de Bruijn Version: mdb.master a93810c OS: Linux URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (94.210.130.16) MDB_GET_BOTH seems to unexpectedly return success if a data-value before the first data-value for a key is requested. The example program sets 1:b and 1:c and requests 1:a. It prints: get_both: key=1 val=b While I would expect: get_both: err=MDB_NOTFOUND: No matching key/data pair found (curiously it works when it only sets 1:b) #include <stdlib.h> #include <stdio.h> #include <string.h> #include <unistd.h> #include "lmdb.h" void set_val(MDB_val *v, char *s) { v->mv_size = strlen(s); v->mv_data = s; } int main() { int rc; MDB_env *menv = NULL; MDB_dbi mdbi; MDB_txn *mtxn = NULL; MDB_val key, val; MDB_cursor *mc; if ((rc = mdb_env_create(&menv))) goto err; if ((rc = unlink("testdb"))) goto err; if ((rc = mdb_env_open(menv, "testdb", MDB_NOSUBDIR, 0666))) goto err; if ((rc = mdb_txn_begin(menv, NULL, 0, &mtxn))) goto err; if ((rc = mdb_dbi_open(mtxn, NULL, MDB_DUPSORT | MDB_CREATE, &mdbi))) goto err; set_val(&key, "1"); set_val(&val, "b"); if ((rc = mdb_put(mtxn, mdbi, &key, &val, 0))) goto err; set_val(&key, "1"); set_val(&val, "c"); if ((rc = mdb_put(mtxn, mdbi, &key, &val, 0))) goto err; if ((rc = mdb_cursor_open(mtxn, mdbi, &mc))) goto err; set_val(&key, "1"); set_val(&val, "a"); rc = mdb_cursor_get(mc, &key, &val, MDB_GET_BOTH); if (rc) printf("get_both: err=%s\n", mdb_strerror(rc)); else printf("get_both: key=%.*s val=%.*s\n", key.mv_size, (char *)key.mv_data, val.mv_size, (char *)val.mv_data); mdb_cursor_close(mc); if ((rc = mdb_txn_commit(mtxn))) { mdb_dbi_close(menv, mdbi); goto err; } mdb_env_close(menv); return 0; err: if (mtxn) mdb_txn_abort(mtxn); if (menv) mdb_env_close(menv); fprintf(stderr, "Error: %s\n", mdb_strerror(rc)); return 1; }

1 0

Re: (ITS#7874) Replication Debian & Ubuntu ( PROBLEM )
by quanah＠zimbra.com 05 Jun '14

05 Jun '14

--On Thursday, June 05, 2014 11:40 AM +0000 lisivoandre(a)gmail.com wrote: > ____________________________________________x____________________________ > ____________ > > Changed: > olcServerID: 001 ldaps://ldap01.example.tld > For: > olcServerID: 001 ldap://<IP> > ________________________________________________________ It was all the > changes i made! a) The URI in olcServerID *must* match the URI used to start slapd (the -h option to slapd) b) ldaps/startTLS do not work with a server IP unless you've added the IP to subjectAltNames c) The Debian/Ubuntu builds are utterly unusable, you need to get a sane build. In any case, this is not a bug report. This ITS will be closed. --Quanah -- Quanah Gibson-Mount Server Architect Zimbra, Inc. -------------------- Zimbra :: the leader in open source messaging and collaboration

1 0

(ITS#7874) Replication Debian & Ubuntu ( PROBLEM )
by lisivoandre＠gmail.com 05 Jun '14

05 Jun '14

Full_Name: André Campos Version: 2.4.31 OS: Debian URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (193.136.194.93) Hi i am having problems with the replication i am using Ubuntu 12.04 LTS to connect to a Debian, the master. With the help from this tutorial: http://grungelabs.com/guides/2012/06/08/ha-openldap-debian/, i did some changes to the overall. I didn't configure the ssl but i did Import the schemata and add memberof. The commands like: ldapmodify -Y EXTERNAL -H ldapi:/// -D olcDatabase={1}hdb,cn=config -f ./repl_db_slave_auth.ldif For: ldapmodify -f repl_master.ldif -D cn=admin,cn=config -x -W ____________________________________________x________________________________________ Changed: olcServerID: 001 ldaps://ldap01.example.tld For: olcServerID: 001 ldap://<IP> ________________________________________________________ It was all the changes i made! But here is the problem i made all the tutorial in the Debian (master) and the last part mean to Ubuntu (slave or ldap02) and when i restart it's ok in the slave but in the master when i restart slapd it gets an error and when i go to the logs, here's the following: Jun 5 10:56:14 controller slapd[11221]: @(#) $OpenLDAP: slapd (Apr 23 2013 12:16:04) $#012#011root@lupin:/tmp/buildd/openldap-2.4.31/debian/build/servers/slapd Jun 5 10:56:14 controller slapd[11221]: read_config: no serverID / URL match found. Check slapd -h arguments. Jun 5 10:56:14 controller slapd[11221]: slapd stopped. Jun 5 10:56:14 controller slapd[11221]: connections_destroy: nothing to destroy. _______________________________________x____________________________________________ I don't know what to do so what do i do?

1 0

(ITS#7873) slapo-refint interferes with cn=config
by hyc＠OpenLDAP.org 04 Jun '14

04 Jun '14

Full_Name: Howard Chu Version: 2.4.39 OS: URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (78.155.233.73) Submitted by: hyc The refint overlay performs its referential integrity updates in a separately queued thread. On a large database the necessary updates may take a long time, and currently the refint_qtask doesn't check for pending cn=config pause requests. Fix coming shortly.

1 0

(ITS#7872) slapd-mdb memory leak in mdb_entry_get
by hyc＠OpenLDAP.org 04 Jun '14

04 Jun '14

Full_Name: Howard Chu Version: 2.4.39 OS: URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (78.155.233.73) Submitted by: hyc Actually the leak is in dn2entry but it only occurs when called from mdb_entry_get. A fix is coming shortly.

1 0

Re: (ITS#6970) OpenLDAP 2.4.25 MemberOf + AutoGroup user has stale "memberof" attributes for target group after removal from trigger group
by ryan＠nardis.ca 04 Jun '14

04 Jun '14

Hi, The last comments on this ITS mentioned crashes or deadlocks when using autogroup and memberof together. I tried this today with git master on both mdb and hdb backends. Adding a new entry that triggers an autogroup member to be added: I don't get a crash or deadlock, but the new entry doesn't have a memberOf attribute. It looks like the internal Modify adding the member happens before the Add completes, so memberof tries to add memberOf to an entry that doesn't exist yet. Changing autogroup to trigger on an alteration to an existing entry, for example addition of a specific attribute or auxiliary objectClass: with mdb this works and the memberOf attribute is added; with hdb I get the same deadlock Landry Breuil already reported. Looks like a recursive deadlock on the same thread. I think those problems are both different from the one this ITS is about, though. Maybe they should be filed separately. I also confirmed the problem originally reported on current git master, and the analysis sounds right, but I haven't had time to think through the proposed solution fully. I'm not totally sure whether it's autogroup or memberof that actually needs fixing. Maybe both. cheers, Ryan

1 0

Re: (ITS#7871) slapcat ldif-wrap option produced wrong column length
by hyc＠symas.com 04 Jun '14

04 Jun '14

mwarren(a)symas.com wrote: > Full_Name: Mark Warren > Version: 2.4.39 > OS: Debian 6 x86_64 > URL: ftp://ftp.openldap.org/incoming/ > Submission from: (NULL) (72.35.133.119) > > > It appears that the slapcat option '-o ldif-wrap=n' is not properly followed. > > e.g, this: > > slapcat -o ldif-wrap=30 > > will produce a file with 32 columns not including the newline character. Yes, we already detailed this in ITS#7859 followup #1. -- -- 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

(ITS#7871) slapcat ldif-wrap option produced wrong column length
by mwarren＠symas.com 03 Jun '14

03 Jun '14

Full_Name: Mark Warren Version: 2.4.39 OS: Debian 6 x86_64 URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (72.35.133.119) It appears that the slapcat option '-o ldif-wrap=n' is not properly followed. e.g, this: slapcat -o ldif-wrap=30 will produce a file with 32 columns not including the newline character. Regards, Mark

1 0

(ITS#7870) slapd-mdb slapcat failure
by hyc＠OpenLDAP.org 03 Jun '14

03 Jun '14

Full_Name: Howard Chu Version: 2.4.39 OS: URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (78.155.233.73) Submitted by: hyc If an index was configured in slapd.conf but not created yet (e.g. slapindex wasn't run) then slapcat will fail because it tried to open the index DB in read-only mode and it doesn't exist yet. slapcat shouldn't even bother to open the index DBs since it doesn't use them. Fix coming shortly.

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