openldap-bugs

openldap-bugs@openldap.org

1 participants
20963 discussions

(ITS#7513) added TCP keepalive support to back_meta
by tedcheng＠symas.com 30 Jan '13

30 Jan '13

Full_Name: Ted C. Cheng Version: Head OS: Linux/Solaris URL: https://dl.dropbox.com/u/94235048/back_meta_tcp_keepalive.patch Submission from: (NULL) (76.174.253.197) Please find in the link a patch against HEAD for TCP keepalive support for back_meta.

1 0

(ITS#7512) MDB page leaks
by h.b.furuseth＠usit.uio.no 30 Jan '13

30 Jan '13

Full_Name: Hallvard B Furuseth Version: mdb.master 27aaecc744955d08d2bfe7a3ca786d742267c5bd OS: Linux x86_64 URL: Submission from: (NULL) (193.69.163.163) Submitted by: hallvard If the malloc at mdb_page_malloc() line 1348 fails, me_pgfirst has already been incremented and those pages can be lost. I'll fix that one. Not this, however: If a child transaction aborts, it loses pages it used up: It does not give them back to me_pghead, nor to the parent's mt_free_pgs. It does not rewind me_pgfirst/me_last to their state at txn_begin. This program illustrates. The DB grows to 107 pages. mdb_stat can see 10 of them, same as if everything was done in the main txn. MainDB will have max 1 entry while running. #include <lmdb.h> #include <stdio.h> #include <stdlib.h> int main(void) { int rc, i; MDB_val key = { sizeof(i), &i }, data = { 0, "" }; MDB_env *env; MDB_txn *txn, *ctxn; MDB_dbi dbi; # define dbname "test.mdb" # define E(e) { rc = (e); if (rc) { fprintf(stderr, "%d: %s: %s\n", \ __LINE__, #e, mdb_strerror(rc)); return rc; } } E(mdb_env_create(&env)); E(mdb_env_open(env, dbname, MDB_NOSYNC|MDB_NOSUBDIR, 0666)); for (i = 0; i < 100; ) { E(mdb_txn_begin(env, NULL, 0, &txn)); if (!i) { E(mdb_dbi_open(txn, NULL, MDB_CREATE, &dbi)); } else { E(mdb_txn_begin(env, txn, 0, &ctxn)); E(mdb_put(ctxn, dbi, &key, &data, 0)); mdb_txn_abort(ctxn); E(mdb_del(txn, dbi, &key, NULL)); } i++; E(mdb_put(txn, dbi, &key, &data, 0)); E(mdb_txn_commit(txn)); if (i % 20 == 0) system("du " dbname); } mdb_env_close(env); return 0; }

1 0

Re: (ITS#7505) liblmdb can fail with MDB_PAGE_FULL on a delete
by hyc＠symas.com 29 Jan '13

29 Jan '13

Sung Ruo wrote: > Hi Howard: > Our test program has been running for about a day now and the fix is holding up good still. > We used to reproduce this in about 1~2 days. So I'll let it runs for another day. > > Will update again sometime tomorrow. That's good news, thanks for the followup. -- -- 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#7505) liblmdb can fail with MDB_PAGE_FULL on a delete
by quanah＠zimbra.com 29 Jan '13

29 Jan '13

--On Tuesday, January 29, 2013 5:36 PM +0000 sruo(a)vmware.com wrote: > Hi Howard: > Our test program has been running for about a day now and the fix is > holding up good still. We used to reproduce this in about 1~2 days. So > I'll let it runs for another day. > > Will update again sometime tomorrow. Hi Sung, I don't know if you saw, but Howard made a tweak to the 7505 commit yesterday. If you could try current mdb.master from git, that would be helpful. It also has a commit addressing ITS#7511. Thanks, Quanah -- Quanah Gibson-Mount Sr. Member of Technical Staff Zimbra, Inc A Division of VMware, Inc. -------------------- Zimbra :: the leader in open source messaging and collaboration

1 0

Re: (ITS#7505) liblmdb can fail with MDB_PAGE_FULL on a delete
by sruo＠vmware.com 29 Jan '13

29 Jan '13

Hi Howard: Our test program has been running for about a day now and the fix is holding up good still. We used to reproduce this in about 1~2 days. So I'll let it runs for another day. Will update again sometime tomorrow. Thanks. --Sung

1 0

(ITS#7511) EINVAL in internal mdb_cursor_del()
by h.b.furuseth＠usit.uio.no 28 Jan '13

28 Jan '13

Full_Name: Hallvard B Furuseth Version: master, d235778c71a50da91741b8b17c572a44b143831b OS: Linux x86_64 URL: Submission from: (NULL) (193.69.163.163) Submitted by: hallvard mdb_cursor_put() can return MDB_SUCCESS with mc.mc_flags == 0. mdb_cursor_del() requires mc.mc_flags & C_INITIALIZED. mdb_txn_commit() can call _del() which fails after _put() succeeds, at the /* was completely used up */ case. This breaks slapadd -l frontier.ldif (without -q) at ada.openldap.org, at "uid=smithhehehe,ou=people,dc=frontiernet,dc=net" (line=527182). Config ~hallvard/frontiercf.ldif. Or to trigger it a lot quicker, insert assert(rc || (mc->mc_flags & C_INITIALIZED)); before 'done:' in mdb_cursor_put(). I'm guessing that is required, since ldbm.h says _put() positions the cursor at the newly inserted item. Or is that only when _inserting_ a new item, not _modifying_ an existing one? The mdb_cursor_del() check for C_INITIALIZED is incorrect, it is evaluated as if ((!mc->mc_flags) & C_INITIALIZED). Maybe that's why this has not been caught before. Same with the test in mdb_cursor_get case MDB_GET_CURRENT.

1 0

(ITS#7510) Group ACL's not cached
by tioteath＠gmail.com 28 Jan '13

28 Jan '13

Full_Name: Tio Teath Version: 2.4.33 OS: Debian GNU Linux URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (178.120.33.222) For some reason group ACL's not cached by slapd during single connection. This behavior lead to major efficiency decrease, especially when groups provided by ldap-proxy. Each ACL produces separate request to the back-end database. Not sure if ACL cached at all, but I tend to think this is inappropriate behavior.

1 0

(ITS#7509) 2.4 Admin Guide olcDatabase={0}config Clarification w.r.t. olcRootDN
by selling＠vml.com 28 Jan '13

28 Jan '13

Full_Name: Steven Elling Version: 2.4.* OS: Irrelevant URL: http://www.openldap.org/doc/admin24/ Submission from: (NULL) (205.240.253.201) The OpenLDAP 2.4 Admin Guide makes no mention of the fact that the "olcRootDN" defined in "olcDatabase={0}config,cn=config" has full access to the "cn=config" DIT (i.e. That there is an implied "olcSuffix" value of "cn=config"). While I now understand this important detail, I think it is best to add this information to the documentation for clarification (or as a reminder) to new and current users of the OpenLDAP software.

1 0

(ITS#7508) 2.4 Administrator's Guide Missing Date on PDF Version
by selling＠vml.com 28 Jan '13

28 Jan '13

Full_Name: Steven Elling Version: 2.4.* OS: Irrelevant URL: http://www.openldap.org/doc/admin24/OpenLDAP-Admin-Guide.pdf Submission from: (NULL) (205.240.253.201) The modified date of the OpenLDAP Software 2.4 Administrator's Guide really needs to be included in the PDF version like it is for both HTML versions. Without the modified date, it is difficult to tell whether a downloaded copy of the PDF is out of date or not.

1 0

(ITS#7507) OpenLDAP Software 2.4 Administrator's Guide PDF Correction (N-Way Multi-Master)
by selling＠vml.com 28 Jan '13

28 Jan '13

Full_Name: Steven Elling Version: 2.4.* OS: Irrelevant URL: http://www.openldap.org/doc/admin24/OpenLDAP-Admin-Guide.pdf Submission from: (NULL) (205.240.253.201) In the "18.3.3. N-Way Multi-Master" section page 171 of the PDF formatted OpenLDAP Software 2.4 Administrator's Guide, the "olcLimits" option of the "olcDatabase={1}$BACKEND,cn=config" example provided is incomplete. It appears the PDF generator cut off the text of "olcLimits" that extended beyond the bounding box.

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