Patrick Kaufmann schrieb am 06.04.2009 14:13 Hi,
We experience some serious problems with our openldap. We got about 50 databases in our openldap. If I now request single databases everything works fine until I request the 30st database.
From there on every database which I haven't already requested will just return an internal error.
After some further investigation I finally figured out that the bdb backend somehow has some problems. If I start the sldap with d -1 I get following error: bdb_locker_id: Not enough space(12)
We use openLDAP 2.3 together with DBD 4.2.52
Whole error message: daemon: activity on 1 descriptor daemon: activity on: 4052r daemon: read activity on 4052 daemon: select: listen=1912 active_threads=0 tvp=NULL connection_get(4052) connection_get(4052): got connid=0 connection_read(4052): checking for input on id=0 ber_get_next ldap_read: want=8, got=8 0000: 30 35 02 01 6e 63 30 04 05..nc0. ldap_read: want=47, got=47 0000: 0f 6f 3d 74 65 73 74 2e 6d 61 69 6c 34 2e 63 68 .o=test.domain 0010: 0a 01 00 0a 01 00 02 02 03 e8 02 01 1e 01 01 00 ................ 0020: 87 0b 6f 62 6a 65 63 74 63 6c 61 73 73 30 00 ..objectclass0. ber_get_next: tag 0x30 len 53 contents: ber_dump: buf=086F29D8 ptr=086F29D8 end=086F2A0D len=53 0000: 02 01 6e 63 30 04 0f 6f 3d 74 65 73 74 2e 6d 61 ..nc0..o=test.ma 0010: 69 6c 34 2e 63 68 0a 01 00 0a 01 00 02 02 03 e8 il4.ch.......... 0020: 02 01 1e 01 01 00 87 0b 6f 62 6a 65 63 74 63 6c ........objectcl 0030: 61 73 73 30 00 ass0. ber_get_next ldap_read: want=8 error=Unknown error daemon: activity on 1 descriptor daemon: waked daemon: select: listen=1912 active_threads=0 tvp=NULL do_search ber_scanf fmt ({miiiib) ber: ber_dump: buf=086F29D8 ptr=086F29DB end=086F2A0D len=50 0000: 63 30 04 0f 6f 3d 74 65 73 74 2e 6d 61 69 6c 34 c0..o=test.mail4 0010: 2e 63 68 0a 01 00 0a 01 00 02 02 03 e8 02 01 1e .ch............. 0020: 01 01 00 87 0b 6f 62 6a 65 63 74 63 6c 61 73 73 .....objectclass 0030: 30 00 0.
dnPrettyNormal: <o=test.domain>
=> ldap_bv2dn(o=test.domain,0) <= ldap_bv2dn(o=test.domain)=0 => ldap_dn2bv(272) <= ldap_dn2bv(o=test.domain)=0 => ldap_dn2bv(272) <= ldap_dn2bv(o=test.domain)=0 <<< dnPrettyNormal: <o=test.domain>, <o=test.domain> SRCH "o=test.domain" 0 0 1000 30 0 begin get_filter PRESENT ber_scanf fmt (m) ber: ber_dump: buf=086F29D8 ptr=086F29FE end=086F2A0D len=15 0000: 87 0b 6f 62 6a 65 63 74 63 6c 61 73 73 30 00 ..objectclass0. end get_filter 0 filter: (objectClass=*) ber_scanf fmt ({M}}) ber: ber_dump: buf=086F29D8 ptr=086F2A0B end=086F2A0D len=2 0000: 00 00 .. attrs: conn=0 op=109 SRCH base="o=test.domain" scope=0 deref=0 filter="(objectClass=*)" ==> limits_get: conn=0 op=109 dn="[anonymous]" => bdb_search bdb_locker_id: err Not enough space(12) send_ldap_result: conn=0 op=109 p=3 send_ldap_result: err=80 matched="" text="internal error" send_ldap_response: msgid=110 tag=101 err=80 ber_flush: 28 bytes to sd 4052 0000: 30 1a 02 01 6e 65 15 0a 01 50 04 00 04 0e 69 6e 0...ne...P....in 0010: 74 65 72 6e 61 6c 20 65 72 72 6f 72 ternal error ldap_write: want=28, written=28 0000: 30 1a 02 01 6e 65 15 0a 01 50 04 00 04 0e 69 6e 0...ne...P....in 0010: 74 65 72 6e 61 6c 20 65 72 72 6f 72 ternal error conn=0 op=109 SEARCH RESULT tag=101 err=80 nentries=0 text=internal error
Any help is appreciated.
Thanks Patrick
Patrick Kaufmann Softwareentwicklung
Hauptsitz: Hardstrasse 72 CH - 5430 Wettingen Tel. +41 56 437 80 40 Fax + 41 56 437 80 59 pkaufmann@zubler.chmailto:pkaufmann@zubler.ch www.zubler.chhttp://www.zubler.ch [cid:aussenstelle.jpg]
--On Monday, April 06, 2009 2:13 PM +0200 Patrick Kaufmann pkaufmann@zubler.ch wrote:
Patrick Kaufmann schrieb am 06.04.2009 14:13
Hi,
We experience some serious problems with our openldap.
We got about 50 databases in our openldap. If I now request single databases everything works fine until I request the 30st database.
From there on every database which I haven't already requested will just return an internal error.
After some further investigation I finally figured out that the bdb backend somehow has some problems.
If I start the sldap with d -1 I get following error:
bdb_locker_id: Not enough space(12)
At a guess, you're running on a 32-bit system and have run out of resources.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc -------------------- Zimbra :: the leader in open source messaging and collaboration
Hm, thats true, it's a 32 bit system. What can I do about it? I have tried various things in the DB_CONFIG file and the also the openldap config file, because I already thought of the resource problem.
Patrick
________________________________________ Von: Quanah Gibson-Mount [quanah@zimbra.com] Gesendet: Montag, 6. April 2009 18:33 An: Patrick Kaufmann; openldap-technical@openldap.org Betreff: Re: problem with a lot of databases in openldap 2.3
--On Monday, April 06, 2009 2:13 PM +0200 Patrick Kaufmann pkaufmann@zubler.ch wrote:
Patrick Kaufmann schrieb am 06.04.2009 14:13
Hi,
We experience some serious problems with our openldap.
We got about 50 databases in our openldap. If I now request single databases everything works fine until I request the 30st database.
From there on every database which I haven't already requested will just return an internal error.
After some further investigation I finally figured out that the bdb backend somehow has some problems.
If I start the sldap with d -1 I get following error:
bdb_locker_id: Not enough space(12)
At a guess, you're running on a 32-bit system and have run out of resources.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc -------------------- Zimbra :: the leader in open source messaging and collaboration
--On Wednesday, April 08, 2009 8:20 AM +0200 Patrick Kaufmann pkaufmann@zubler.ch wrote:
Hm, that's true, it's a 32 bit system. What can I do about it?
Use a 64-bit system that doesn't have the paltry resources a 32-bit one has.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc -------------------- Zimbra :: the leader in open source messaging and collaboration
Ok, now that's somehow a problem with our system, because the whole thing runs on a Windows Server. It's an iMail-Server, so we can't just shift the whole system onto a linux based server, although this would certainly be better for the openldap server.
Patrick
-----Ursprüngliche Nachricht----- Von: Quanah Gibson-Mount [mailto:quanah@zimbra.com] Gesendet: Mittwoch, 8. April 2009 18:23 An: Patrick Kaufmann; openldap-technical@openldap.org Betreff: Re: AW: problem with a lot of databases in openldap 2.3
--On Wednesday, April 08, 2009 8:20 AM +0200 Patrick Kaufmann pkaufmann@zubler.ch wrote:
Hm, that's true, it's a 32 bit system. What can I do about it?
Use a 64-bit system that doesn't have the paltry resources a 32-bit one has.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc -------------------- Zimbra :: the leader in open source messaging and collaboration
--On Tuesday, April 14, 2009 3:35 PM +0200 Patrick Kaufmann pkaufmann@zubler.ch wrote:
Ok, now that's somehow a problem with our system, because the whole thing runs on a Windows Server. It's an iMail-Server, so we can't just shift the whole system onto a linux based server, although this would certainly be better for the openldap server.
Can't you configure the rest of the services to query an LDAP server that's not located on the same box? LDAP is generally for doing distributed work anyhow...
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc -------------------- Zimbra :: the leader in open source messaging and collaboration
On Monday 06 April 2009 14:13:17 Patrick Kaufmann wrote:
Patrick Kaufmann schrieb am 06.04.2009 14:13 Hi,
We experience some serious problems with our openldap. We got about 50 databases in our openldap. If I now request single databases everything works fine until I request the 30st database.
From there on every database which I haven't already requested will just return an internal error.
What you haven't explained so far is why you need so many databases. We have ~ 1.6 million entries in 3 databases (they are separate because they have different replication requirements).
We use the ldap function oft he IMail server. And no, up to now Ipswitch also hasn't got a solution for the problem. They create a new database for each domain you've got on the imail server. As far as I know you can't change this default behaviour.
On Monday 06 April 2009 14:13:17 Patrick Kaufmann wrote:
Patrick Kaufmann schrieb am 06.04.2009 14:13 Hi,
We experience some serious problems with our openldap. We got about 50 databases in our openldap. If I now request single databases everything works fine until I request the 30st database.
From there on every database which I haven't already requested will just return an internal error.
What you haven't explained so far is why you need so many databases. We have ~ 1.6 million entries in 3 databases (they are separate because they have different replication requirements).
--On Wednesday, April 15, 2009 8:36 AM +0200 Patrick Kaufmann pkaufmann@zubler.ch wrote:
We use the ldap function oft he IMail server. And no, up to now Ipswitch also hasn't got a solution for the problem. They create a new database for each domain you've got on the imail server. As far as I know you can't change this default behaviour.
Root your database at "" and use a single database. It's trivial.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc -------------------- Zimbra :: the leader in open source messaging and collaboration
The problem is, when I sync the single database out of the mailserver it will just overwrite my settings and recreate the databases, because as I see it, there isn't an option to set the root directory of my openldap database(s). Or am I to dumb to see what you mean by that? :)
-----Ursprüngliche Nachricht----- Von: Quanah Gibson-Mount [mailto:quanah@zimbra.com] Gesendet: Mittwoch, 15. April 2009 08:53 An: Patrick Kaufmann; openldap-technical@openldap.org Betreff: Re: AW: problem with a lot of databases in openldap 2.3
--On Wednesday, April 15, 2009 8:36 AM +0200 Patrick Kaufmann pkaufmann@zubler.ch wrote:
We use the ldap function oft he IMail server. And no, up to now Ipswitch also hasn't got a solution for the problem. They create a new database for each domain you've got on the imail server. As far as I know you can't change this default behaviour.
Root your database at "" and use a single database. It's trivial.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc -------------------- Zimbra :: the leader in open source messaging and collaboration
--On Wednesday, April 15, 2009 9:19 AM +0200 Patrick Kaufmann pkaufmann@zubler.ch wrote:
The problem is, when I sync the single database out of the mailserver it will just overwrite my settings and recreate the databases, because as I see it, there isn't an option to set the root directory of my openldap database(s). Or am I to dumb to see what you mean by that? :)
Have a single database definition in slapd.conf, where the suffix used is "", rather than "dc=example,dc=com" etc. By having the suffix rooted at "", all entries exist beneath it.
--Quanah
--
Quanah Gibson-Mount Principal Software Engineer Zimbra, Inc -------------------- Zimbra :: the leader in open source messaging and collaboration
openldap-technical@openldap.org