[SOLVED] '"zimbra" user fails authenication

[jrefl5]

When running zmprov, as the zimbra user "I su - zimbra" on CENT-OS 5 I get the following;

ERROR: account.AUTH_FAILED (authenication failed for user zimbra)

.

We had a server crash which lead to a reinstall of everything OS & zimbra,
I then used the last good backup set of tapes (outside product) to restore the /opt/zimbra/ structure.

I'm thinking that the file restore over wrote something that is needed.

What was clobbered and how do I fix it?

Thanks

[jrefl5]

Additional Info from the web administration console looking at Manage Mail Queues I get;

Message: system failure: exception during auth {RemoteManager: email.sb.state.az.us->zimbra@email.sb.state.az.us:22}
Error code: service.FAILURE
Method: ZmCsfeCommand.prototype.invoke
Details:soap:Receiver

[mmorse]

Have you reset the ldap password? Directions: Resetting LDAP & MySQL Passwords - Zimbra :: Wiki
Do you have an entry in /etc/passwd? UID's match?
After restores you often need to regenerate the certs: SSL Certificate Problems - Zimbra :: Wiki
Queue problem: Mail Queue Monitoring - Zimbra :: Wiki

[jrefl5]

Thanks,
That did not appear in my searches. zmprov is working. now.

Testing

Partial fix, "Manage Mail Queues" still returning errors.

[mmorse]

Alright getting there, so run me through what you've done - like:
1. check /etc/sshd_config and be sure it's set to 22

port 22

2. stop/start/restart sshd

/etc/init.d/ssh restart

3. su - zimbra
4. Just to double check that it's set:

zmprov ms server.domain.com zimbraRemoteManagementPort 22

5. generate new ssh keys:

cd /opt/zimbra/bin/
./zmsshkeygen

6. deploy the keys

./zmupdateauthkeys

all done, to test this, you can run the following command

ssh -i .ssh/zimbra_identity -o strictHostKeyChecking=no zimbra@server.domain.com -p 22

or -vi
and it should not prompt you for a password

[jrefl5]

checking with

ssh -i .ssh/zimbra_identity -o strictHostKeyChecking=no -p 22

I get the following;

$ ssh -i .ssh/zimbra_identity -o strictHostKeyChecking=no zimbra@FQDN -p 22
Warning: Identity file .ssh/zimbra_identity not accessible: No such file or directory.
Warning: Permanently added 'FQDN,10.168.30.60' (RSA) to the list of known hosts.
zimbra@FQDN's password: 
Permission denied, please try again.

Is there an option to the install.sh script that will reset ALL the access controls without killing the various mailboxes?

[jrefl5]

The final fix was to get a full all accounts backup, and copy /opt/zimbra/backup to another filesystem. Then install.sh -u followed by a reinstall.

1.

zmbackup -a all-f

2.

cp -pr /opt/zimbra/backup /tmp/zimbra/

as root.

3.

install.sh -u

4.

install.sh -s

5.

cp -pr /tmp/zimbra/backup /opt/zimbra/backup

as zimbra
6.

zmrestore -a all -lb last-Full_backup (step 1.)

from the Zimbra Administration console
7. confirm everything was back.

note that I had to restore some accounts from an older backup (these were mostly accounts that had not been used at that time).

8. recheck again.
9. another Full backup

zmbackup --fullbackup --target /opt/zimbra/backup --account all

[mmorse]

Guess you got a good test of your backups out of this!
-marking thread solved

[jrefl5]

Much better than planed