Clustered DNS - Cannot Recreate Accounts Once Deleted

[vittle]

We are having a major issue here for the past few weeks, with our "clustered" DNS system. Firstly, the way our DNS servers work is that we have three central servers hosting the DNS entries for all our 5 servers. They are called bionic, vs12 and helicase, as you will see in the below output. Now, the problem is that recently (we upgraded to WHM 10.8.0 cPanel 10.8.2-R119 ), after we terminate account, we can't re-create the account with the same domain, as WHM gives an error that says "Sorry, a DNS entry for testacct12.net already exists, please delete it first (from all servers in the dns cluster)"

We checked /var/named as well as /usr/local/apache/sites as well as named.conf and all of them don't show up with the domain on ANY server. So we're at loss to explain how CPanel thinks that the entries are not deleted. Yes, we restarted apache/bind to make sure they refresh after any edits, but no luck.

Below is basically the way we follow to reproduce the DNS error:


How to Reproduce the Clustered DNS Error

PART I : Create a new test domain account via WHM
-----
1) Login to ANY server's WHM, bionic, benzene, primus or helicase
2) Go to "Create New Account"
3) Enter any domain as the domain, such as "testacct13.com", the username must not contain "test" itself, however (can be like "accttest") and enter a temp password and select any feature package for now. Uncheck the "Make owner" checkbox and then press Submit
4) Once the account is created successfully, go to Part II

PART II: Terminating the account

1) Click "Terminate an Account" option in WHM
2) Select the above account/domain you created and click Terminate
3) You will notice an error as follows:
................
Removing PostgreSQL databases and users....

Done
Removing User....Done
Removing Group....Done
Removing Named entry...testacct12.net -> deleted from bionic.

testacct12.net -> deleted from helicase.
Unable remove zone.
Unable to remove zone testacct12.net from the Bind configuration (named.conf).
The zone was possibly removed earlier.
testacct12.net -> deleted from vs12.
Done
......................

4) Now proceed to Part III

PART III: Recreate the account -- it will fail!

Follow the exact same steps as part I, and this time, the account creation will NOT work. You will see instead the following error message:

......................
WWWAcct 10.0 (c) 1997-2005 cPanel, Inc.

Sorry, a DNS entry for testacct12.net already exists, please delete it first (from all servers in the dns cluster)



*Account Creation Complete!!!*

......................

As you can see, the account is never added, and you can never re-add the domain. This functionality is NOT normal, you should be able to recreate the account. For some reason, the DNS says the domain exists, when any DNS entry doesn't show it does, as there are no traces of it in any file on any of the servers.

Any help on this issue is appreciated. Thanks!

[vittle]

anyone? Thanks

[jbw]

I wrote a program that checks each of the nameservers directly for the domains. None of the nameservers answer for any of the domains that cpanel erroniosly says ""Sorry, a DNS entry for testacct12.net already exists, please delete it first (from all servers in the dns cluster)" and none of the servers have an entry in /var/named/ for the domains.

I traced this message to being in the files whostmgr and whostmgr2 , which both seem to be 'compiled' perl so undebuggable, fixable from this point.

Is cpanel a supported product. if so how do we get some support on this bug?

[protocol]

We are also experiencing this problem if the server is not one of the nameservers i.e. not ns1, ns2 etc. then accounts cannot be added after being removed without manually removing the entry from the named.conf on that machine. I don't see why the machine needs to be doing anything with named anyway as it is not nameserver.

[ekhwan]

I am having the same problem. Can anyone point to how to delete the dns entries from the cluster and recreate the account.

[Anishts]

I think you have to do /scripts/killdns domain.com on all cluster servers.. then recreate the account.

[ekhwan]

I think you have to do /scripts/killdns domain.com on all cluster servers.. then recreate the account.

Can you kindly write an how-to for this. It would be most help full. I am a complete dummy for this.

[Anishts]

Assume that you are having 3 clustered servers . server1, server2,server3.

login to server 1 through ssh and issue the command.

#scripts/killdns your_problem_domain.com

then login to server2 and issue the command.

#scripts/killdns your_problem_domain.com

finally login to server3 and issue the command.

#scripts/killdns your_problem_domain.com

Then go to first clustered server and create the dns for the domain again.

[vittle]

This does not seem to work still:

root@primus [~]# /scripts/killdns testdomain12.net
rndc: connection to remote host closed
This may indicate that the remote server is using an older version of the command protocol, this host is not authorized to connect, or the key is invalid.

testdomain12.net -> missing from named.conf on primusUnable to remove zone testdomain12.net from the Bind configuration (named.conf).
The zone was possibly removed earlier.
Unable to unlink /var/named/testdomain12.net.db, file does not exist.testdomain12.net -> deleted from primus.

testdomain12.net -> missing from named.conf on helicaseUnable to remove zone testdomain12.net from the Bind configuration (named.conf).
The zone was possibly removed earlier.
Unable to unlink /var/named/testdomain12.net.db, file does not exist.testdomain12.net -> deleted from helicase.

testdomain12.net -> missing from named.conf on bionicUnable to remove zone testdomain12.net from the Bind configuration (named.conf).
The zone was possibly removed earlier.
Unable to unlink /var/named/testdomain12.net.db, file does not exist.testdomain12.net -> deleted from bionic.
Unable remove zone.

Unable to remove zone testdomain12.net from the Bind configuration (named.conf).
The zone was possibly removed earlier.
Unable to unlink /var/named/testdomain12.net.db, file does not exist.testdomain12.net -> deleted from vs12.

Afterwards, going to server to recreate the account still produces the DNS exists error...