[Fedora-directory-users] ldapmodify to add OU failed, and led to "ldap_search: Operations error"

[ken]

ldapmodify to add OU failed, and led to "ldap_search: Operations 
error"

I set up a directory and am feeling my way towards making it 
live by doing one thing at a time. I successfully added quite a 
large number of users using ldapmodify, and could retrieve them 
with ldapsearch and db2ldif.

Then I tried to add some new OUs in order to copy a subset of 
the OU hierarchy we have on Windows. But when I ran the 
ldapmodify it failed.

 From that moment on, every ldapsearch I tried resulted in:
   ldap_search: Operations error

I stopped and restarted the ldap daemon and  now every search I 
try produces:
   ldap_search: No such object

The database looks as if it is empty

What did I do wrong?  Can a botched modify so easily wipe out 
what is already there? Or are they secretly hiding somewhere? Is 
there a utility that can show me what is actually in the 
database even if I don't know what its root name is? (I thought 
db2ldif might do that and it shows nothing now)

I know I can wipe this database entirely (it looks like I 
already have!) and re-install. BUT I want to know what I did 
wrong so I don't do it again. The LDIFs to be imported into the 
directory will come from a program I wrote (it gathers 
information from various sources such as an SQL database, WAD, 
/etc/passwd...) and I really don't want to risk repeating my 
mistake in batch runs at 2am after we've gone live and coming in 
to find no-one can use the directory.


Any clues?

Examples of what I did:

Command used to import LDIF to define an OU

========================
ldapmodify -a -B "dc=bbk,dc=ac,dc=uk" -D "cn=directory manager" 
   -w [PWD]
========================

the LDIF that was used:

========================
dn: ou=students,ou=people,dc=bbk,dc=ac,dc=uk
objectClass: top
objectClass: organizationalunit
ou: students
========================

Error log from ldapmodify:

==========================================
[09/Jun/2008:19:54:47 +0100] - WARNING: Import is running with 
    nsslapd-db-private-import-mem on; No other process is 
allowed to access the database
[09/Jun/2008:19:54:47 +0100] - Bulk import: begin import on 
'dc=bbk,dc=ac,dc=uk'.
[09/Jun/2008:19:54:56 +0100] - import userRoot: WARNING: 
Skipping entry "ou=students,ou=people,dc=bbk,dc=ac,dc=uk" which 
has no parent, ending at line 0 of file "(bulk import)"
[09/Jun/2008:19:55:05 +0100] - import userRoot: Workers 
finished; cleaning up...
[09/Jun/2008:19:55:05 +0100] - import userRoot: Workers cleaned up.
[09/Jun/2008:19:55:05 +0100] - import userRoot: Indexing 
complete.  Post-processing...
[09/Jun/2008:19:55:05 +0100] - Nothing to do to build ancestorid 
index
[09/Jun/2008:19:55:05 +0100] - import userRoot: Flushing caches...
[09/Jun/2008:19:55:05 +0100] - import userRoot: Closing files...
[09/Jun/2008:19:55:05 +0100] - import userRoot: Import complete. 
  Processed 1 entries (1 were skipped) in 18 seconds. (0.06 
entries/sec)
[09/Jun/2008:19:55:05 +0100] - Bulk import completed successfully.
==========================================

There was no error message on the screen, but the log says the 
object "has no parent". Though as far as I can tell it has the 
same parent as the user entries I added successfully, such as:

========================
# ldapsearch -b 'dc=bbk,dc=ac,dc=uk' -D "cn=directory manager" 
-w [PWD] '(objectclass=person)'
version: 1
dn: cn=xlean99,ou=people,dc=bbk,dc=ac,dc=uk
cn: xlean99
description: A mythical person to test LDAP with
objectClass: person
objectClass: top
sn: Lean
========================

Which was in there but is no longer.

But now I see things like:

========================
[ken@~]$  /usr/lib/mozldap/ldapsearch -b 'dc=bbk,dc=ac,dc=uk' -D 
"cn=uxxxxxx" '(cn=uxxxxxx)'
ldap_search: No such object

[ken@~]$ ldapsearch -b 'dc=bbk,dc=ac,dc=uk' -D "cn=directory 
manager" -w [PWD] '(objectclass=person)'
ldap_search: No such object

[ken@~]$ ldapsearch  -b 'dc=bbk,dc=ac,dc=uk' '(objectclass=*)'
ldap_search: No such object
========================

And most worryingly of all:

========================
[ken@~]$ ldapsearch -b "" -s base -D "cn=directory manager" -w 
[PWD] '(objectclass=*)' namingContext
version: 1
dn:
[ken@~]$
========================

:-(

It's not a permissions problem because this works:

========================
ldapsearch -b o=netscaperoot "objectclass=*" cn
========================

This also still works:

========================
[]# /usr/lib/dirsrv/slapd-ldap1/db2ldif -n NetscapeRoot -a 
/tmp/stuff2.ldif
Exported ldif file: /tmp/stuff2.ldif
ldiffile: /tmp/stuff2.ldif
[11/Jun/2008:14:03:47 +0100] - export NetscapeRoot: Processed 95 
entries (100%).
[11/Jun/2008:14:03:47 +0100] - All database threads now stopped
========================

But this produces nothing:

========================
[]# /usr/lib/dirsrv/slapd-ldap1/db2ldif -n UserRoot -a 
/tmp/stuffU.ldif
Exported ldif file: /tmp/stuffU.ldif
ldiffile: /tmp/stuffU.ldif
[11/Jun/2008:14:38:23 +0100] - All database threads now stopped
[]# more /tmp/stuffU.ldif
version: 1
========================

[Rich Megginson]

ken wrote:
>  ldapmodify to add OU failed, and led to "ldap_search: Operations
> error"
>
>  I set up a directory and am feeling my way towards making it live by 
>  doing one thing at a time. I successfully added quite a large number 
>  of users using ldapmodify, and could retrieve them with ldapsearch and 
>  db2ldif.
>
>  Then I tried to add some new OUs in order to copy a subset of the OU 
>  hierarchy we have on Windows. But when I ran the ldapmodify it failed.
>
>  From that moment on, every ldapsearch I tried resulted in:
>    ldap_search: Operations error
>
>  I stopped and restarted the ldap daemon and  now every search I try 
>  produces:
>    ldap_search: No such object
>
>  The database looks as if it is empty
>
>  What did I do wrong?  Can a botched modify so easily wipe out what is 
>  already there? Or are they secretly hiding somewhere? Is there a 
>  utility that can show me what is actually in the database even if I 
>  don't know what its root name is? (I thought db2ldif might do that and 
>  it shows nothing now)
>
>  I know I can wipe this database entirely (it looks like I already 
>  have!) and re-install. BUT I want to know what I did wrong so I don't 
>  do it again. The LDIFs to be imported into the directory will come 
>  from a program I wrote (it gathers information from various sources 
>  such as an SQL database, WAD, /etc/passwd...) and I really don't want 
>  to risk repeating my mistake in batch runs at 2am after we've gone 
>  live and coming in to find no-one can use the directory.
>
>
>  Any clues?
>
>  Examples of what I did:
>
>  Command used to import LDIF to define an OU
>
>  ========================
>  ldapmodify -a -B "dc=bbk,dc=ac,dc=uk" -D "cn=directory manager"   -w
>
>  [PWD] 
Why are you using the -B option here?  In mozldap ldapmodify, -B means 
"bulk online import" which means
1) wipe out the database
2) use the database import code to import the given LDIF

>  ========================
>
>  the LDIF that was used:
>
>  ========================
>  dn: ou=students,ou=people,dc=bbk,dc=ac,dc=uk
>  objectClass: top
>  objectClass: organizationalunit
>  ou: students
>  ========================
>
>  Error log from ldapmodify:
>
>  ==========================================
>  [09/Jun/2008:19:54:47 +0100] - WARNING: Import is running with    
>  nsslapd-db-private-import-mem on; No other process is allowed to 
>  access the database
>  [09/Jun/2008:19:54:47 +0100] - Bulk import: begin import on 
>  'dc=bbk,dc=ac,dc=uk'.
>  [09/Jun/2008:19:54:56 +0100] - import userRoot: WARNING: Skipping 
>  entry "ou=students,ou=people,dc=bbk,dc=ac,dc=uk" which has no parent, 
>  ending at line 0 of file "(bulk import)"
>  [09/Jun/2008:19:55:05 +0100] - import userRoot: Workers finished; 
>  cleaning up...
>  [09/Jun/2008:19:55:05 +0100] - import userRoot: Workers cleaned up.
>  [09/Jun/2008:19:55:05 +0100] - import userRoot: Indexing complete.  
>  Post-processing...
>  [09/Jun/2008:19:55:05 +0100] - Nothing to do to build ancestorid index
>  [09/Jun/2008:19:55:05 +0100] - import userRoot: Flushing caches...
>  [09/Jun/2008:19:55:05 +0100] - import userRoot: Closing files...
>  [09/Jun/2008:19:55:05 +0100] - import userRoot: Import complete. 
>   Processed 1 entries (1 were skipped) in 18 seconds. (0.06 entries/sec)
>  [09/Jun/2008:19:55:05 +0100] - Bulk import completed successfully.
>  ==========================================
>
>  There was no error message on the screen, but the log says the object 
>  "has no parent". Though as far as I can tell it has the same parent as 
>  the user entries I added successfully, such as:
>
>  ========================
>  # ldapsearch -b 'dc=bbk,dc=ac,dc=uk' -D "cn=directory manager" -w 
>  [PWD] '(objectclass=person)'
>  version: 1
>  dn: cn=xlean99,ou=people,dc=bbk,dc=ac,dc=uk
>  cn: xlean99
>  description: A mythical person to test LDAP with
>  objectClass: person
>  objectClass: top
>  sn: Lean
>  ========================
>
>  Which was in there but is no longer.
>
>  But now I see things like:
>
>  ========================
>  [ken@~]$  /usr/lib/mozldap/ldapsearch -b 'dc=bbk,dc=ac,dc=uk' -D 
>  "cn=uxxxxxx" '(cn=uxxxxxx)'
>  ldap_search: No such object
>
>  [ken@~]$ ldapsearch -b 'dc=bbk,dc=ac,dc=uk' -D "cn=directory manager" 
>  -w [PWD] '(objectclass=person)'
>  ldap_search: No such object
>
>  [ken@~]$ ldapsearch  -b 'dc=bbk,dc=ac,dc=uk' '(objectclass=*)'
>  ldap_search: No such object
>  ========================
>
>  And most worryingly of all:
>
>  ========================
>  [ken@~]$ ldapsearch -b "" -s base -D "cn=directory manager" -w [PWD]
>
>  '(objectclass=*)' namingContext
>  version: 1
>  dn:
>  [ken@~]$
>  ========================
>
>  :-(
>
>  It's not a permissions problem because this works:
>
>  ========================
>  ldapsearch -b o=netscaperoot "objectclass=*" cn
>  ========================
>
>  This also still works:
>
>  ========================
>  []# /usr/lib/dirsrv/slapd-ldap1/db2ldif -n NetscapeRoot -a 
>  /tmp/stuff2.ldif
>  Exported ldif file: /tmp/stuff2.ldif
>  ldiffile: /tmp/stuff2.ldif
>  [11/Jun/2008:14:03:47 +0100] - export NetscapeRoot: Processed 95 
>  entries (100%).
>  [11/Jun/2008:14:03:47 +0100] - All database threads now stopped
>  ========================
>
>  But this produces nothing:
>
>  ========================
>  []# /usr/lib/dirsrv/slapd-ldap1/db2ldif -n UserRoot -a /tmp/stuffU.ldif
>  Exported ldif file: /tmp/stuffU.ldif
>  ldiffile: /tmp/stuffU.ldif
>  [11/Jun/2008:14:38:23 +0100] - All database threads now stopped
>  []# more /tmp/stuffU.ldif
>  version: 1
>  ========================
>
>
>
>
>
>
>
>
>
>  -- 
>  Fedora-directory-users mailing list
>  Fedora-directory-users(a)redhat.com
>  https://www.redhat.com/mailman/listinfo/fedora-directory-users

[ken]

Rich Megginson wrote:

> > ========================
> > ldapmodify -a -B "dc=bbk,dc=ac,dc=uk" 
>  Why are you using the -B option here?  In mozldap ldapmodify, -B
> means 
>  "bulk online import" which means
>  1) wipe out the database
>  2) use the database import code to import the given LDIF 
That's my error!    A simple typo. Probably the result of my 
habit of cutting and pasting commands from one place to another.

One of those things that I didn't spot in two days of worrying 
about the command because when I looked at it I must have 
assumed it said what I intended it to say rather than what it 
actually said.

Thanks!