[389-users] db import failure, when setting replication up
Noriko Hosoi
nhosoi at redhat.com
Wed May 25 17:50:17 UTC 2011
On 05/24/2011 06:27 AM, Reinhard Nappert wrote:
> I do that.
> Now, I have two questions:
> So, what db version do you recommend?
Hi Reinhard,
Which OS you are running?
If it's RHEL5 (BDB4.3.29) or RHEL6 (BDB4.7.25), they are patched. But
RHEL4 (BDB4.2.52) was rejected.
> More importantly, is there a migration path or do I have to reload the
> existing data? I could see issues migrating replicated environments.
There's no data change needed. The bug was just in the data
verification code.
This bug has more detailed info.
*Bug 472131* <https://bugzilla.redhat.com/show_bug.cgi?id=472131>
-dbverify: when a duplicate is large enough to have internal page(s),
dbverify issues bogus out-of-order key errors
Thanks,
--noriko
> Thanks,
> -Reinhard
>
> ------------------------------------------------------------------------
> *From:* 389-users-bounces at lists.fedoraproject.org
> [mailto:389-users-bounces at lists.fedoraproject.org] *On Behalf Of
> *Noriko Hosoi
> *Sent:* Monday, May 23, 2011 1:42 PM
> *To:* 389-users at lists.fedoraproject.org
> *Subject:* Re: [389-users] db import failure, when setting replication up
>
> Hi Reinhard,
>
> That was an unfortunate... I was hoping you were using a newer
> version. :) You hit this bug.
>
> *Bug 472131* <https://bugzilla.redhat.com/show_bug.cgi?id=472131>
> -dbverify: when a duplicate is large enough to have internal page(s),
> dbverify issues bogus out-of-order key errors
>
> The bug was fixed by Sleepycat on db4.8. And we ported the fix back
> to 4.3, but no chance to do so to 4.2. So, we cannot use dbverify to
> check if the index file is healthy or not... Could it be possible to
> reindex the ancestorid index and see if the error goes away? (Or you
> could reinitialize the consumer? That would be the cleanest)
>
> Thanks,
> --noriko
>
> Reinhard Nappert wrote:
>> Hi Noriko,
>> I run it on a CentOS 4.4 box (Linux 2.6.24). I use the db 4.2 libs
>> with all the patches.
>> Oh, yes dbverify does complain a lot. I see for all of the db files
>> messages like:
>> [20/May/2011:11:03:05 -0400] DB verify - verify failed(-30976):
>> /var/lib/dirsrv/slapd-ID/db/userRoot/cn.db4
>> [20/May/2011:11:03:06 -0400] - libdb: Page 5: out-of-order key at entry 2
>> [20/May/2011:11:03:06 -0400] - libdb: Page 5: out-of-order key at entry 5
>> [20/May/2011:11:03:06 -0400] - libdb: Page 5: out-of-order key at entry 8
>> [20/May/2011:11:03:06 -0400] - libdb: Page 5: out-of-order key at
>> entry 10
>> [20/May/2011:11:03:06 -0400] - libdb: Page 5: out-of-order key at
>> entry 13
>> [20/May/2011:11:03:06 -0400] - libdb: Page 5: out-of-order key at
>> entry 16
>> [20/May/2011:11:03:06 -0400] - libdb: Page 5: out-of-order key at
>> entry 19
>> [20/May/2011:11:03:06 -0400] - libdb: Page 5: out-of-order key at
>> entry 21
>> [20/May/2011:11:03:07 -0400] DB verify - verify failed(-30976):
>> /var/lib/dirsrv/slapd-ID/db/userRoot/parentid.db4
>> DB verify: Passed
>> This said, I guess I should re-index the entire db. Any idea, why this is happening?
>> Right now, I have a 2 MMR setup, where both masters also have a
>> replication agreement to a third box, which is a dedicated consumer.
>> I do run tests, where I perform simultaneously adds and deletes (not
>> on the same object) on all three boxes. I just want to verify how
>> replication behaves in 1.2.8.
>> -Reinhard
>>
>> ------------------------------------------------------------------------
>> *From:* 389-users-bounces at lists.fedoraproject.org
>> [mailto:389-users-bounces at lists.fedoraproject.org] *On Behalf Of
>> *Noriko Hosoi
>> *Sent:* Thursday, May 19, 2011 5:33 PM
>> *To:* 389-users at lists.fedoraproject.org
>> *Subject:* Re: [389-users] db import failure, when setting replication up
>>
>> Hi Reinhard,
>>
>> Could you tell me the OS version and Berkeley DB version (rpm -q db4)?
>>
>> Could you run "/usr/lib[64]/dirsrv/slapd-ID/dbverify"? Does it
>> complain anything? Especially, the ancestorid index? If it does,
>> you may want to re-create the corrupted index...
>> --noriko
>>
>> Reinhard Nappert wrote:
>>> Noriko,
>>> I observed one more item, which does not bother me right now, but
>>> you may want to see:
>>> I am not sure why and how it happened, but I see the following
>>> message on the supplier:
>>> [18/May/2011:13:59:50 -0400] NSMMReplicationPlugin -
>>> agmt="cn=supplier2consumer" (consumer:389): Consumer failed to
>>> replay change (uniqueid aea3731d-808711e0-83d5fdc8-f32b8f3c, CSN
>>> 4dd4085b004800040000): Operations error. Will retry later.
>>> And I see the following on the consumer:
>>> [18/May/2011:13:59:29 -0400] - idl_new.c BAD 22, err=-30988
>>> DB_PAGE_NOTFOUND: Requested page not found
>>> [18/May/2011:13:59:29 -0400] - ancestorid BAD 13120, err=-30988
>>> DB_PAGE_NOTFOUND: Requested page not found
>>> Any idea, what happened there....
>>> Thanks,
>>> -Reinhard
>>>
>>>
>>>
>>> ------------------------------------------------------------------------
>>> *From:* 389-users-bounces at lists.fedoraproject.org
>>> [mailto:389-users-bounces at lists.fedoraproject.org] *On Behalf Of
>>> *Noriko Hosoi
>>> *Sent:* Tuesday, May 17, 2011 4:02 PM
>>> *To:* General discussion list for the 389 Directory server project.
>>> *Subject:* Re: [389-users] db import failure, when setting
>>> replication up
>>>
>>> Hi Reinhard,
>>>
>>> Reinhard Nappert wrote:
>>>> Hi Noriko,
>>>> I have to correct myself. The box which had the import issue was on
>>>> a 1.2.7.5 system. The other box was running 1.2.8.2.
>>>> So, it looks like you have fixed the issue with 1.2.8.2.
>>> *relieved* Thanks for testing it on 1.2.8.2!
>>> --noriko
>>>> Thanks,
>>>> -Reinhard
>>>>
>>>> ------------------------------------------------------------------------
>>>> *From:* 389-users-bounces at lists.fedoraproject.org
>>>> [mailto:389-users-bounces at lists.fedoraproject.org] *On Behalf Of
>>>> *Reinhard Nappert
>>>> *Sent:* Tuesday, May 17, 2011 3:21 PM
>>>> *To:* General discussion list for the 389 Directory server project.
>>>> *Subject:* Re: [389-users] db import failure, when setting
>>>> replication up
>>>>
>>>> 1.2.8.2
>>>> -Reinhard
>>>>
>>>> ------------------------------------------------------------------------
>>>> *From:* 389-users-bounces at lists.fedoraproject.org
>>>> [mailto:389-users-bounces at lists.fedoraproject.org] *On Behalf Of
>>>> *Noriko Hosoi
>>>> *Sent:* Tuesday, May 17, 2011 2:16 PM
>>>> *To:* General discussion list for the 389 Directory server project.
>>>> *Subject:* Re: [389-users] db import failure, when setting
>>>> replication up
>>>>
>>>> It looks to me you have hit this bug... Which version of
>>>> 389-ds-base you are running?
>>>>
>>>> *Bug 684996*
>>>> <https://bugzilla.redhat.com/show_bug.cgi?id=684996> -Exported
>>>> tombstone cannot be imported correctly.
>>>>
>>>> The patch should be in the version 1.2.8.2.
>>>> Thanks,
>>>> --noriko
>>>>
>>>> On 05/17/2011 11:03 AM, Reinhard Nappert wrote:
>>>>> Hi,
>>>>> I have seen the following:
>>>>> I set 2 systems up in MMR. Replication worked. For some reason, I
>>>>> needed to take one of the boxes out of the replication and
>>>>> disabled replication. Later on, I enabled it again and created the
>>>>> shadowing agreement to the other box. Now, I saw the following
>>>>> errors during the import of the db:
>>>>> [17/May/2011:11:46:04 -0400] NSMMReplicationPlugin -
>>>>> multimaster_be_state_change
>>>>> : replica o=base is going offline; disabling replication
>>>>> [17/May/2011:11:46:07 -0400] - WARNING: Import is running with
>>>>> nsslapd-db-privat
>>>>> e-import-mem on; No other process is allowed to access the database
>>>>> [17/May/2011:11:46:08 -0400] - import userRoot: WARNING: Skipping
>>>>> entry "nsuniqu
>>>>> eid=06869502-7fe011e0-8f589300-7e7b2163,ou=sample,o=base" which
>>>>> has no parent,
>>>>> ending at line 0 of file "(bulk import)"
>>>>> [17/May/2011:11:46:08 -0400] - import userRoot: WARNING: bad
>>>>> entry: ID 453
>>>>> .....
>>>>> Any idea, what is going on there?
>>>>> Thanks,
>>>>> -Reinhard
>>>>>
>>>>>
>>>>> --
>>>>> 389 users mailing list
>>>>> 389-users at lists.fedoraproject.org
>>>>> https://admin.fedoraproject.org/mailman/listinfo/389-users
>>>>
>>>>
>>>> --
>>>> 389 users mailing list
>>>> 389-users at lists.fedoraproject.org
>>>> https://admin.fedoraproject.org/mailman/listinfo/389-users
>>>
>>>
>>> --
>>> 389 users mailing list
>>> 389-users at lists.fedoraproject.org
>>> https://admin.fedoraproject.org/mailman/listinfo/389-users
>>
>>
>> --
>> 389 users mailing list
>> 389-users at lists.fedoraproject.org
>> https://admin.fedoraproject.org/mailman/listinfo/389-users
>
>
> --
> 389 users mailing list
> 389-users at lists.fedoraproject.org
> https://admin.fedoraproject.org/mailman/listinfo/389-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.fedoraproject.org/pipermail/389-users/attachments/20110525/5e49eb6f/attachment.html>
More information about the 389-users
mailing list