Trailing space (encode base64) dn:: dWlkPWphdjg3MDIsb3U9dmVuZG9ycyxvdT1wZW

Question

If there is a space at the end of DN, the dn will become encode base64. The DN is looked like this
dn:: dWlkPWphdjg3MDIsb3U9dmVuZG9ycyxvdT1wZW
Put an "-e" option on the ldapsearch can make the "dn:: dWlkPWphdjg3MDIsb3U9dmVuZG9ycyxvdT1wZW" becomes visible.
My question is how I can fix the "dn:: dWlkPWphdjg3MDIsb3U9dmVuZG9ycyxvdT1wZW" back to "dn: cn=abc, dc=xxx, dc=com".
Do I have to delete the whole record and remove the trailing space at the ldif file and insert the record with the fixed ldif file ? Since I have 90% records like that or around 7000 records need to be fixed, and this in production environment. I want to minimize the impact to the production users, please advise a better solution.
Thanks

Answer 1

It's only 7000 entries. Deleting and readding shouldn't pose too much of a problem. Might just be a little scripting involved, depending on how you want to handle the maintenance. If the entries are effectively non-functional due to the base 64 encoding on the DN already, you might not have much to lose in just extracting them to LDIF, using something like sed on the LDIF, and then doing a batch delete / add. If you want to be more sophisticated about it, you can write a script that takes each entry at a time, does some checking on it, deletes and then readds it, and logs a result. That's probably what I'd do. If the data doesn't change much and/or you can drop changes from the past day or two, you can even pull the entries out of an LDIF backup.