Plesk 10.4.4: Bind not working due to malformed master zone

[ThomasR]

After upgrade from Plesk 10.3.1 to Plesk 10.4.4 Bind can not be restarted any more:

# /etc/init.d/named restart
Stopping named: [ OK ]
Starting named:
Error in named configuration:
[...]
zone domainforum.tld/IN: loaded serial 2011111605
dns_master_load: forum.domain.tld:17: www.forum.domain.tld: CNAME and other data
zone forum.domain.tld/IN: loading master file forum.domain.tld: CNAME and other data
_default/forum.domain.tld/IN: CNAME and other data
[FAILED]

The problem occurs because on the same server both domain.tld and domainforum.tld (with forum.domain.tld as a domain alias) are set up as two independent domains (each with their own customer). Apparently the upgrade to Plesk 10.4.4 did put A records for forum.domain both in /var/named/run-root/var/forum.domain.tld and in /var/named/run-root/var/domain.tld.

The only solution is to delete and recreate the domain alias forum.domain.tld. This seems to be a bug in Plesk 10.4 as there was no problem with it in Plesk 10.3 or previous versions.

The same bug makes it impossible to do an migration form Plesk 8.6 to Plesk 10.4.4:

Error: domain "forum.domain.tld"

Execution of /usr/local/psa/admin/plib/api-cli/dns.php --set forum.domain.tld -list /usr/local/psa/PMM/tmp/dns19BZJR -ignore-nonexistent-options failed with return code 1.
Stderr is
Unable to restore the DNS zone: an error occurred while adding the ftp.forum.domain.tld. IN CNAME forum.domain.tld. record: Such DNS record already exists.


Error: domain "forum.domain.tld"

Execution of /usr/local/psa/admin/plib/api-cli/dns.php --set forum.domain.tld -list /usr/local/psa/PMM/tmp/dnsdNDkYb -ignore-nonexistent-options failed with return code 1.
Stderr is
Unable to restore the DNS zone: an error occurred while adding the ftp.forum.domain.tld. IN CNAME forum.domain.tld. record: Such DNS record already exists.


Error: hosting "forum.domain.tld"

Execution of /usr/local/psa/admin/plib/api-cli/domain.php --update forum.domain.tld -hosting true -hst_type phys -do-not-apply-skeleton -ip AA.BB.CC.DD -www-root httpdocs -ssl false -same_ssl false -login forumdomainlogin -passwd '' -hard_quota 20971520 -ignore-nonexistent-options failed with return code 0.


Error: hosting "forum.domain.tld"

Execution of /usr/local/psa/admin/plib/api-cli/domain.php --set-php-settings forum.domain.tld -settings /usr/local/psa/PMM/tmp/phpSettingszaZEwa -ignore-nonexistent-options failed with return code 536305584.

Server and product details:
° Product: Plesk 10.4.4 Update #3, last updated at Nov 22, 2011 12:27 PM
° Operating system: CentOS release 5.7 (Final)
° Architecture: x86_64
° Kernel version 2.6.18-028stab089.1
° Bind version: bind-9.3.6-16.P1.el5_7.1

 

[Christopher McBride]

Thomas, did you get a resolution to this?

I am now facing the same issue.

 

[paulieG]

Hi,

Couldn't you try disabling the DNS synchronisation on the domain alias?

Paul.

 

[Christopher McBride]

My issue is similar, but not exactly the same (but looks like the same cause providing slightly different symptoms).

I just upgraded to 10.4.4 and all subdomains and aliases failed to reload in bind.

My only solution to get at least the main domains running was to run the named reconfigure script given by Parallels.

#!/bin/sh

ADMIN_PASS=`cat /etc/psa/.psa.shadow`
MYSQL_BIN_D=`grep MYSQL_BIN_D /etc/psa/psa.conf | awk '{print $2}'`
mysql="${MYSQL_BIN_D}/mysql -N -uadmin -p${ADMIN_PASS} psa"

query="select name from domains;"
domains=`echo $query | $mysql `

for i in ${domains}; do
    echo "echo $i"
    /usr/local/psa/admin/sbin/dnsmng --update $i
done

This has worked, but no subdomains are included in the new named.conf file.


If i restore a backup of the named.conf file that I took before using the reconfigure script, the same error occurs for each domain with an alias or subdomain.

zone maindomain.org.uk/IN: loaded serial 2011112501
dns_master_load: aliasdomain.com:20: [url]www.aliasdomain.com:[/url] CNAME and other data
zone aliasdomain.com/IN: loading master file aliasdomain.com: CNAME and other data
_default/aliasdomain.com/IN: CNAME and other data

 

[Christopher McBride]

Don't know what the cause was, but a "clutch at straws" moment had me tick all domains in the domains list and click "activate" on them.

Some with manual dns settings failed, but a quick manual check on those and an update saw everything back online.

 

[ShoneA]

Bind issue with legacy domain aliases

We had this issue after upgrading to 10.4.4 bind would not start.
Turns out some older domain alaises had an A record for WWW and a CNAME for WWW

Steps we took to get it resolved
1) We got a list of problem subdomains bind was having issues with

2) In Plesk we turned off dns sync for the subdomain

3) Then for the we clicked the manage DNS for that subdomain.

4) Click the Restore Defaults (DNS) for that subdomain.


Once we got through the list of sites bind was able to start. We left off domain sync for the time being.

Hope this helps someone

Shone
www.xpcwebhosting.com

 

[burnleyvic]

This can be a show stopper if upgrading existing 10.3.1 installations. I'd like to see a resolution for this from Parallels, too.

 

[IgorG]

Problem has been forwarded to developers and under their investigation now. I will keep this thread updated with results.

 

[ThomasR]

Thomas, did you get a resolution to this?

For me, the only solution was to delete and recreate the domain alias (just disabling DNS sync and re-enabling did not work).

@IgorG: do you have any ETA for this?

 

[HarryBleckert]

Plesk garbled interfaces file unusable

On Nov. 3, while still using Plesk 10.3.1, the network interface file /etc/networking/interfaces was completely garbled.
On Nov 26, after upgrading to Plesk 10.4.4, I encountered the same problem with statistics not running anymore as described in this thread.

Following a hunch, I decided to reboot and see what happens...
Should not have done it: Server was nonresponsive after reboot.

Cost me 2 hours in rescue mode anylyzing log files, till I realized that file /etc/networking/interfaces was completely garbled. Almost all descriptions were duplicated, in result no IP address was working!!!

So this problem is NOT 10.4.4 related, but was produced by 10.3.1 and only became effective on reboot.

- Parallels Plesk Panel v10.4.4_build1013111102.18 os_Ubuntu 10.04

 

[HernanR]

Hi, i have the same issue too!
after upgrade to 10.4.4 all my subdomains dont resolve
adding a new subdomain into any domain account with plesk i get:
"Error: Unable to update domain data: Domain::update failed: Failed to switch on the webmail service.: IP pool is empty."

i tried with:
- turn off dns sync for the subdomain
- reboot system
- slave dns for domain

I have

CentOS release 5.5 (Final)
Parallels Plesk 10.4.4

 

[IgorG]

Hi, i have the same issue too!
after upgrade to 10.4.4 all my subdomains dont resolve
adding a new subdomain into any domain account with plesk i get:
"Error: Unable to update domain data: Domain::update failed: Failed to switch on the webmail service.: IP pool is empty."

i tried with:
- turn off dns sync for the subdomain
- reboot system
- slave dns for domain

I have

CentOS release 5.5 (Final)
Parallels Plesk 10.4.4

http://kb.parallels.com/en/9658

 

[IgorG]

ThomasR, we can't reproduce reported problem on our test environment. Please send me in PM login credentials to your server and developers will check problem directly on your server, if it is possible.

 

[sergius]

Hello, Gents. The issue has been reproduced and localized. Thanks for reporting.

The bug fix will be delivered with the next micro-update 10.4.4 MU#6.

 

[agarzon]

I make the upgrade today 10.4.4 #6 and now I have the same problem (so, not was fixed)

 

[kadoudal]

running 10.4.4#6 ..not fixed

It's a big mess now .... and no clear info on how to run domain/subdomains ... whatever existi,g solution it should be explained clearly ....

 

[sergius]

It's a big mess now .... and no clear info on how to run domain/subdomains ... whatever existi,g solution it should be explained clearly ....

Bind is not restarted on MU#6 installing. So those who have Bind failed to restart have to start it manually after MU installation. Hope it helps.

 

[saatchi]

an abnormal method to solve this problem...

Hello,

I've exactly, the same problem today and I search it to resolve it. All of my domains stopped.

in the Tools & Settings > DNS Template Settings , I tried to return all setting to default and after that I re-config it by my own zone settings.
After that it's need to applying changes. So I click Apply and it start to syncing. then I restart PSA service by ssh and Named was start!

I don't know that what's the main reason of this issue, but by this method, my problem has resolved.
I write it here for the others that were exposed of this issue.