Single user can't send or receive email after upgrade to 8.5.0

[langs]

Friday night I upgraded to 8.5.0 all seemed to go well except when I sent the notice to all users to say the upgrade was complete one account bounced back to me immediately with;

554 5.0.0 Permanent message delivery failure (in reply to end of DATA command)

In mailbox.log I spied;

14-10-25 16:33:58,319 INFO [LmtpServer-8] [name=user@ourdomain;mid=506;ip=10.12.1.3;] lmtp - rejecting message from=user@ourdomain,to=user@ourdomain 

com.zimbra.cs.mailbox.MailServiceException: object with that id already exists: 9066 
ExceptionId:LmtpServer-8:1414218838319:9c21a1690db12394 
Code:mail.ALREADY_EXISTS Arg:(itemId, IID, "9066") 
at com.zimbra.cs.mailbox.MailServiceException.ALREADY_EXISTS(MailServiceException.java:455) 
at com.zimbra.cs.db.DbMailItem.create(DbMailItem.java:204) 

Caused by: com.mysql.jdbc.exceptions.jdbc4.MySQLIntegrityConstraintViolationException: Duplicate entry '506-9066' for key 'PRIMARY' 
Query being executed when exception was thrown: 
INSERT INTO mboxgroup6.mail_item(mailbox_id, id, type, parent_id, folder_id, index_id, imap_id, date, size, locator, blob_digest, unread, flags, tag_names, sender, recipients, subject, name, metadata, mod_metadata, change_date, mod_content, uuid) VALUES (506, 9066, 5, null, 2, 0, 9066, 1414218834, 2768, '1', 'rS5iIGzMXNGH52+Osxe04O69kHSoAurB2LHmZEzHBnw=', 1, 0, null, 'Me', 'everyone', 'Maintenance Completed', null, 'd1:f151:Everyone, All forward facing services have now been restored, there will still be some tasks being performed in the background this evening however ...1:s38:ME <user@ourdomain>1:t32:everyone <everyone@ourdomain>1:vi10ee', 21351, 1414218838, 21351, null) 

The key point being the error "object with that id already exists: 9066"

I also logged in as this user and attempted to send an email and was met with the same error;

 An object with that name already exists.

method: [unknown]

msg: object with that id already exists: 9066

code: mail.ALREADY_EXISTS

I have reindexed this users account a number of times and the error persists, I have logged this with support but 2 day's later they still have not responded. 

Anyone have suggestions?

[langs]

In the end I just exported the users mail, delete the account, recreated it and imported the old mail. This is not an acceptable solution, however I could no longer wait for support to *maybe* get back to me.

[jorgedlcruz]

Hi idiomatically,

How did you create the account the first time? Some script? Or manually?



Best regards

[langs]

The account was created manually both times, out of the 100's of users on the system this was the only account effected and they are all created the same way.

[jorgedlcruz]

Hi idiomatically,

I very strange the error, if you have it next time please force a GAL sync,

zmgsautil forceSync -a gal@yourdomain.com -n InternalGAL

 

Best regards

[ccelis5215]

Jorge, clearly a bug and you as a zimbra employee or whatever, please take note and report accordingly.



ccelis

[jorgedlcruz]

Hi ccelis5215,

This error could be a database corruption, we don't know if in the upgrade process the option "check the database Integrity" was selected or not. But of course I've taken note this issue, and if I'll see other similar issue, even in our Customers in Network Edition, or here, we will open a bug for work on it.



Best regards

[danielfarrelly]

Also had/having this issue after upgrading. Don't think it's related to v8.5, rather how you go about upgrading. We did an OS update first, then upgraded Zimbra from v8.0 to v8.5. The former prevents Zimbra from running during the update, so it cannot perform a db check during install. As idiomatically said, simply exporting/reimporting account data into a new account works fine, but it's a rather cludgy solution.

Since there will be a lot of users moving to an updated OS over the next releases due to EOL of older OS releases, maybe Zimbra can come up with a way of performing an offline db check during installation.

[langs]

Jorge,



I can confirm for you I always use the "check the database integrity" option, it was not a db corruption. Also I am noting that the DOSS filter stuff is a pain in the arse with this release.



Users get the "Network Service Error" when they switch between accounts too fast which is due to the over zealous DOSS filter.

[jorgedlcruz]

Hi danielfarrelly,

I will take a look and try to reproduce the issue, then I will open a Bugm with this steps I will be able to reproduce it. How many users do you have? Maybe in my labs will work with 10 o 20 users, I will talk also with Support Team, maybe we had some case too.



Thank you again for your help and input, danielfarrelly, ccelis5215, idiomatically.



Best regards