Upgrade gone bad

[kechols]

I tried to upgrade from zimbra 3.0.1 to 4.5.4 over the weekend and got the following errors during the upgrade:

Updating from 3.2.0_M1
ERROR: service.FAILURE (system failure: unable to get config) (cause: javax.naming.NamingException [LDAP: error code 80 - internal error])
ERROR: service.FAILURE (system failure: unable to get config) (cause: javax.naming.NamingException [LDAP: error code 80 - internal error])

Checking 3.2.0_M2
Updating from 3.2.0_M2
ERROR: service.FAILURE (system failure: unable to get config) (cause: javax.naming.NamingException [LDAP: error code 80 - internal error])
ERROR: service.FAILURE (system failure: unable to get config) (cause: javax.naming.NamingException [LDAP: error code 80 - internal error])

Updating from 4.5.0_RC1
Failed to delete cn=timezones,cn=config,cn=zimbra: internal errorChecking 4.5.0_RC2

ERROR: service.FAILURE (system failure: unable to lookup COS by name: default message: [LDAP: error code 80 - internal error]) (cause: javax.naming.NamingException [LDAP: error code 80 - internal error])

When I got in this morning, I have a number of people receiving errors in Outlook saying there is an internal server error: RR System failure: missing blob for id:12824 <changes with user>, change: 4038 <probably also changes with user>.

[kechols]

does anyone know what would cause the errors above?

[jholder]

Do you have a backup?
If so, restore to that. Then do an upgrade from 3.x to 4.x then to 4.5

Upgrades directly from the oldest to the newest arn't supported.

[kechols]

I have restored the backup and got it back up and running. Just wondering what the error means since everything else seemed to go right and if i do see that error again i'd like to know what can be done about it (at least what caused it)

Thanks,
Kyle

[jholder]

It looks like LDAP didn't get upgraded.
If you didn't have a backup, you likely would have had to dump your LDAP DB, and re-import it.

Trust me, it's a nightmare

-john

[kechols]

I had a backup of a snapshot from about 2hrs before luckily. Thanks for letting me know what it looks like happened, I'll probably attempt an upgrade to 4.0.x in the next week or two and make sure it's running right before going to 4.5.x.

@ 1800 users I don't want to import anything I don't have to

Thanks,
Kyle

[kechols]

I do have one other question,

When restoring my backup it looks like some attachments weren't linked back correctly and has caused a fair amount of work in finding the broken emails. Is there a way to reindex all mailboxes and find out which messages are broken so we can remove them more easily?

Thanks,
kyle

[jholder]

I wonder why that happened. . .Did you follow the Wiki backup instructions?
If so, you currently can't reindex from the command line, you have to do it from the admin ui, account by account.

[kechols]

I'm backing up from a snapshot and forgot to cd to the shapshot before running tar so the hardlinks created to attachments for multiple recipients broke.
when restoring from backup, instead of restoring to /opt/zimbra it was restored to /opt/zimbra/media/zimbra-snap (i extracted from the /opt/zimbra directory)

I think that's what caused it at least, or am I wrong in that being the most likely cause?