When initializing a consumer, I get many messages similar to this:
import userRoot: WARNING: Skipping entry "uid=davsmith,ou=win,ou=People,dc=maxim-ic,dc=com" which has no parent, ending at line 0 of file "(bulk import)"
It is believed that all affected entries had been moved within the directory using a moddn change in the past few months (using Apache Directory Studio). I was just able to "break" another account by doing this same action and reinitializing my consumer.
Move entry: #!RESULT OK #!CONNECTION ldap://dalldap001.maxim-ic.com:389 #!DATE 2012-08-28T12:01:56.870 dn: uid=test4phy,ou=People,dc=maxim-ic,dc=com changetype: moddn newrdn: uid=test4phy deleteoldrdn: 1 newsuperior: ou=bri,ou=People,dc=maxim-ic,dc=com
Re-init: [28/Aug/2012:18:05:36 +0100] - import userRoot: WARNING: Skipping entry "uid=test4phy,ou=bri,ou=People,dc=maxim-ic,dc=com" which has no parent, ending at line 0 of file "(bulk import)"
Moving the account back to it's original location and reinitializing the consumer again "fixes" it. The entry does not appear on the consumer until I reinitialize.
If I remember correctly, in the past, I was not able to do a moddn (A few months back I moved from 1.2.5 to 1.2.10). Is moddn considered safe? If I copy and paste the entry to the new location and delete the old entry manually, this works just fine.
I also get one instance of this message on each initialize:
import userRoot: WARNING: Skipping entry "nsuniqueid=2dca7101-594d11df-8622cf4b-918cdd63,uid=fxdtest,ou=chm,ou=People,dc=maxim-ic,dc=com" which has no parent, ending at line 0 of file "(bulk import)"
This one really confuses me since nsuniqueid isn't an entry, it's an attribute. Even stranger is that the value of nsuniqueid in the error message does not match the current value:
dn: uid=fxdtest,ou=chm,ou=People,dc=maxim-ic,dc=com createTimestamp: 20120827200045Z creatorsName: uid=whardin,ou=dal,ou=people,dc=maxim-ic,dc=com entrydn: uid=fxdtest,ou=chm,ou=people,dc=maxim-ic,dc=com entryid: 13315 hasSubordinates: FALSE modifiersName: uid=whardin,ou=dal,ou=people,dc=maxim-ic,dc=com modifyTimestamp: 20120827200214Z nsUniqueId: d2729e81-f08111e1-8cbef100-4e39c57a numSubordinates: 0 parentid: 5877 passwordGraceUserTime: 0 subschemaSubentry: cn=schema
I have deleted and recreated this account in the past few days. The value of nsuniqueid in the error message appears to be the value of the now deleted entry.
These are the packages on the server, which is CentOS 5.7: 389-admin-1.1.29-1.el5.x86_64 389-admin-console-1.1.8-1.el5.noarch 389-adminutil-1.1.15-1.el5.x86_64 389-console-1.1.7-3.el5.noarch 389-ds-base-1.2.10.4-5.el5.x86_64 389-ds-base-libs-1.2.10.4-5.el5.x86_64 389-ds-console-1.2.6-1.el5.noarch
The consumer is also CentOS 5.7, but slightly newer 389-ds: 389-admin-1.1.29-1.el5.x86_64 389-admin-console-1.1.8-1.el5.noarch 389-adminutil-1.1.15-1.el5.x86_64 389-console-1.1.7-3.el5.noarch 389-ds-base-1.2.10.13-1.el5.x86_64 389-ds-base-libs-1.2.10.13-1.el5.x86_64 389-ds-console-1.2.6-1.el5.noarch
On 08/28/2012 11:31 AM, Wes Hardin wrote:
When initializing a consumer, I get many messages similar to this:
import userRoot: WARNING: Skipping entry "uid=davsmith,ou=win,ou=People,dc=maxim-ic,dc=com" which has no parent, ending at line 0 of file "(bulk import)"
It is believed that all affected entries had been moved within the directory using a moddn change in the past few months (using Apache Directory Studio). I was just able to "break" another account by doing this same action and reinitializing my consumer.
Move entry: #!RESULT OK #!CONNECTION ldap://dalldap001.maxim-ic.com:389 #!DATE 2012-08-28T12:01:56.870 dn: uid=test4phy,ou=People,dc=maxim-ic,dc=com changetype: moddn newrdn: uid=test4phy deleteoldrdn: 1 newsuperior: ou=bri,ou=People,dc=maxim-ic,dc=com
Re-init: [28/Aug/2012:18:05:36 +0100] - import userRoot: WARNING: Skipping entry "uid=test4phy,ou=bri,ou=People,dc=maxim-ic,dc=com" which has no parent, ending at line 0 of file "(bulk import)"
Moving the account back to it's original location and reinitializing the consumer again "fixes" it. The entry does not appear on the consumer until I reinitialize.
If I remember correctly, in the past, I was not able to do a moddn (A few months back I moved from 1.2.5 to 1.2.10). Is moddn considered safe?
Yes.
If I copy and paste the entry to the new location and delete the old entry manually, this works just fine.
I also get one instance of this message on each initialize:
import userRoot: WARNING: Skipping entry "nsuniqueid=2dca7101-594d11df-8622cf4b-918cdd63,uid=fxdtest,ou=chm,ou=People,dc=maxim-ic,dc=com" which has no parent, ending at line 0 of file "(bulk import)"
This one really confuses me since nsuniqueid isn't an entry, it's an attribute. Even stranger is that the value of nsuniqueid in the error message does not match the current value:
dn: uid=fxdtest,ou=chm,ou=People,dc=maxim-ic,dc=com createTimestamp: 20120827200045Z creatorsName: uid=whardin,ou=dal,ou=people,dc=maxim-ic,dc=com entrydn: uid=fxdtest,ou=chm,ou=people,dc=maxim-ic,dc=com entryid: 13315 hasSubordinates: FALSE modifiersName: uid=whardin,ou=dal,ou=people,dc=maxim-ic,dc=com modifyTimestamp: 20120827200214Z nsUniqueId: d2729e81-f08111e1-8cbef100-4e39c57a numSubordinates: 0 parentid: 5877 passwordGraceUserTime: 0 subschemaSubentry: cn=schema
I have deleted and recreated this account in the past few days.The value of nsuniqueid in the error message appears to be the value of the now deleted entry.
Right. It is a tombstone entry. We did have some problems early on in 1.2.10 with replication and tombstone entries, but those should have been fixed by 1.2.10.4. Was this 1.2.10.4 an upgrade from an earlier version of 1.2.10?
These are the packages on the server, which is CentOS 5.7: 389-admin-1.1.29-1.el5.x86_64 389-admin-console-1.1.8-1.el5.noarch 389-adminutil-1.1.15-1.el5.x86_64 389-console-1.1.7-3.el5.noarch 389-ds-base-1.2.10.4-5.el5.x86_64 389-ds-base-libs-1.2.10.4-5.el5.x86_64 389-ds-console-1.2.6-1.el5.noarch
The consumer is also CentOS 5.7, but slightly newer 389-ds: 389-admin-1.1.29-1.el5.x86_64 389-admin-console-1.1.8-1.el5.noarch 389-adminutil-1.1.15-1.el5.x86_64 389-console-1.1.7-3.el5.noarch 389-ds-base-1.2.10.13-1.el5.x86_64 389-ds-base-libs-1.2.10.13-1.el5.x86_64 389-ds-console-1.2.6-1.el5.noarch
Try this on your 1.2.10.4-5 system: 1) stop dirsrv 2) export your database to LDIF using db2ldif 3) upgrade to the latest 1.2.10.14-1.el5 from epel-testing (yum update --enablerepo=epel-testing 389-ds-base) on your system that is currently using 1.2.10.4-5 NOTE: I've pushed this package to Stable but it won't show up in the mirrors for a couple of days 4) import the database from your saved LDIF using ldif2db 5) start dirsrv 6) reinitialize the consumer
On 08/28/2012 12:50 PM, Rich Megginson wrote:
On 08/28/2012 11:31 AM, Wes Hardin wrote:
When initializing a consumer, I get many messages similar to this:
import userRoot: WARNING: Skipping entry "uid=davsmith,ou=win,ou=People,dc=maxim-ic,dc=com" which has no parent, ending at line 0 of file "(bulk import)"
It is believed that all affected entries had been moved within the directory using a moddn change in the past few months (using Apache Directory Studio). I was just able to "break" another account by doing this same action and reinitializing my consumer.
Move entry: #!RESULT OK #!CONNECTION ldap://dalldap001.maxim-ic.com:389 #!DATE 2012-08-28T12:01:56.870 dn: uid=test4phy,ou=People,dc=maxim-ic,dc=com changetype: moddn newrdn: uid=test4phy deleteoldrdn: 1 newsuperior: ou=bri,ou=People,dc=maxim-ic,dc=com
Re-init: [28/Aug/2012:18:05:36 +0100] - import userRoot: WARNING: Skipping entry "uid=test4phy,ou=bri,ou=People,dc=maxim-ic,dc=com" which has no parent, ending at line 0 of file "(bulk import)"
Moving the account back to it's original location and reinitializing the consumer again "fixes" it. The entry does not appear on the consumer until I reinitialize.
If I remember correctly, in the past, I was not able to do a moddn (A few months back I moved from 1.2.5 to 1.2.10). Is moddn considered safe?
Yes.
If I copy and paste the entry to the new location and delete the old entry manually, this works just fine.
I also get one instance of this message on each initialize:
import userRoot: WARNING: Skipping entry "nsuniqueid=2dca7101-594d11df-8622cf4b-918cdd63,uid=fxdtest,ou=chm,ou=People,dc=maxim-ic,dc=com" which has no parent, ending at line 0 of file "(bulk import)"
This one really confuses me since nsuniqueid isn't an entry, it's an attribute. Even stranger is that the value of nsuniqueid in the error message does not match the current value:
dn: uid=fxdtest,ou=chm,ou=People,dc=maxim-ic,dc=com createTimestamp: 20120827200045Z creatorsName: uid=whardin,ou=dal,ou=people,dc=maxim-ic,dc=com entrydn: uid=fxdtest,ou=chm,ou=people,dc=maxim-ic,dc=com entryid: 13315 hasSubordinates: FALSE modifiersName: uid=whardin,ou=dal,ou=people,dc=maxim-ic,dc=com modifyTimestamp: 20120827200214Z nsUniqueId: d2729e81-f08111e1-8cbef100-4e39c57a numSubordinates: 0 parentid: 5877 passwordGraceUserTime: 0 subschemaSubentry: cn=schema
I have deleted and recreated this account in the past few days.The value of nsuniqueid in the error message appears to be the value of the now deleted entry.
Right. It is a tombstone entry. We did have some problems early on in 1.2.10 with replication and tombstone entries, but those should have been fixed by 1.2.10.4. Was this 1.2.10.4 an upgrade from an earlier version of 1.2.10?
These are the packages on the server, which is CentOS 5.7: 389-admin-1.1.29-1.el5.x86_64 389-admin-console-1.1.8-1.el5.noarch 389-adminutil-1.1.15-1.el5.x86_64 389-console-1.1.7-3.el5.noarch 389-ds-base-1.2.10.4-5.el5.x86_64 389-ds-base-libs-1.2.10.4-5.el5.x86_64 389-ds-console-1.2.6-1.el5.noarch
The consumer is also CentOS 5.7, but slightly newer 389-ds: 389-admin-1.1.29-1.el5.x86_64 389-admin-console-1.1.8-1.el5.noarch 389-adminutil-1.1.15-1.el5.x86_64 389-console-1.1.7-3.el5.noarch 389-ds-base-1.2.10.13-1.el5.x86_64 389-ds-base-libs-1.2.10.13-1.el5.x86_64 389-ds-console-1.2.6-1.el5.noarch
Try this on your 1.2.10.4-5 system:
- stop dirsrv
- export your database to LDIF using db2ldif
- upgrade to the latest 1.2.10.14-1.el5 from epel-testing (yum update
--enablerepo=epel-testing 389-ds-base) on your system that is currently using 1.2.10.4-5 NOTE: I've pushed this package to Stable but it won't show up in the mirrors for a couple of days 4) import the database from your saved LDIF using ldif2db 5) start dirsrv 6) reinitialize the consumer
I found the latest packages in Koji and followed your instructions. This appears to have fixed my replication issues. I plan to force all my consumers to reinitialize today and will definitely report back if I find any further errors.
Thanks for your help, Rich.
389-users@lists.fedoraproject.org
-
Rich Megginson -
Wes Hardin