The document I had suggested using ldapsearch and ldapadd to migrate data. If lidf2db commands are faster/better I will use them.

>> Try creating all of the required indexes first, then doing the import of
>> your original LDIF.

I am willing to try this, but It is scary to me. I would have rather you said I must be doing something wrong...because....

Our LDAP database has been in production for 6 years. We add indexes to our i-planet on average twice a year due to new software or new features. Your advice is almost suggesting that adding new indexes can corrupt the database.

I will try again from scratch using everyones advice of course.

Thank you,
Edward





On 12/16/06, Noriko Hosoi < nhosoi@redhat.com> wrote:
Richard Megginson wrote:

> Eddie C wrote:
>
>> I recently did an ldif backup of our iplanet 52 database. Its about
>> an 88 MB ldif file.
>> I took this to a new FDS server Dell 850 3 ghz duel core 2 sata hard
>> disks.
>> I ran an ldapadd  the data imported perfectly.
>
Are there any reason to use ldapadd instead of ldif2db?  ldif2db should
be much faster...

>> Then I tried to cutover some systems and give the database some load.
>>
>> System went 200% processor
>>
>> Eventually I realized I was missing indexes so I added them through
>> the graphical tool.
>>
>> The log seemed to do something like this
>> generating index 1%
>> generating index 2%
>> ....
>> generating index 49%
>> Done
>> Seemed weird that they would jump from 49% to Done
>> At this point the new system was running at 100% processor
>> But the queries are running faster on our old 440 MHZ sparc t1
>> server52 database
>>
>> I ran
>> DB ERROR: db_verify: Page 30: out-of-order key at entry 498
>> DB ERROR: db_verify: DB->verify: db/o_com/channelcontentowner.db4:
>> DB_VERIFY_BAD: Database verification failed
>>
>> then I tried db2_index. The program seemed to be in a tight loop
>> complaining about 1 missing entry.
>>
>> I do not realize how the data can be so corrupted right after an import.
>>
>> These are someone generic symptoms. Any ideas? Thanks
>
> Try creating all of the required indexes first, then doing the import
> of your original LDIF.  Not only will the import+index creation be
> much faster (than doing the import then creating the indexes one at a
> time), but I think your database corruption problems will vanish.
>
>> ------------------------------------------------------------------------
>>
>> --
>> Fedora-directory-users mailing list
>> Fedora-directory-users@redhat.com
>> https://www.redhat.com/mailman/listinfo/fedora-directory-users
>>
>
>------------------------------------------------------------------------
>
>--
>Fedora-directory-users mailing list
>Fedora-directory-users@redhat.com
>https://www.redhat.com/mailman/listinfo/fedora-directory-users
>
>



--
Fedora-directory-users mailing list
Fedora-directory-users@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-directory-users