• If you are still using CentOS 7.9, it's time to convert to Alma 8 with the free centos2alma tool by Plesk or Plesk Migrator. Please let us know your experiences or concerns in this thread:
    CentOS2Alma discussion

Question Partially Problems Syncing with Slave - DNS -

Powie

Regular Pleskian
We are trying to debug a problem in syncing zones form our plesk servers to our 2 slaves. We have no idea what is happening. It's unknown since when the problem comes to plesk. It's possible that it is triggered by a Plesk Update, or maybee by updates from bind or rndc....

Our Setup
We have a couple of Plesk Servers in our Stack, All Plesk Obsidian, latest, Debian 9 and Debian 8.
All set with the same setup for slaves in the DNS Template.
2 Debian 10 Instances, without Plesk, as Slave DNS.
SlaveDNS Extension configured, and showing the correct status.
This setup was without problems since 2 years, comming from Plesk Onyx.....

Problem
Slave DNS servers are loosing Zones. (after expire) . It's impossible to Resync this zones to the the Slaves.
Bind logs: named[30612]: client @0x7fef100c71c0 x.x.x.x#24351: received notify for zone 'domain': not authoritative

And now to the confusing point:

- Problem is never on Debian 8 servers
- Some of the Debian 9 servers without any problems.
- Some of the Debian 9 servers with partially problems, not all domains syncing
- 2 Debian 9 servers with problems for all domains.....

What we have tried
We have read and worked trough all KB articles about such problems.
We have reinstalled 1 Slave DNS . no help . same zones with problems
We have read many topics in forums etc, about such problems and possible solutions, theres no hint for us....
We have tried to transfer zones manually to the slave servers with rndc.

Workarround actually
Syncing works after we create the slave zone manually on the Slave DNS server. Without entries. Will be populated from the master on the next sync..........
... impossible workarround because we have to many domains with this problem now.

Question
What to debug next?
 
I have invested some hours for looking into the slave-dns extension and other code from plesk.

I don't know the reason why the zones are gone from the Slave DNS.... I can manage to add the zones again to the slave dns by rndc whe I do it manually! Zones are added without problems and gets also updates from the master later.

I think that somewhere in plesk is stored if a zone is already added to a slave dns. Looking for it!
"rndc .... addzone" I think is only running in 2 cases:
- when the domain is added to the server
- when a slave dns is added

When something is wrong between master and slave, they get out of sync and readding the domain to the slave is impossible. RESYNC or plesk repair dns only sends a refresh to the slaves....
To remove and readd the domain from the server is not an option :)
 
Guy's. I have reslved this issue.

I have invested many many hours in doing deep analysis of plesk and the Slave DNS extension. It's in fact a bug of Plesk, and not of the extension, I think.

I don't want to make a bad information to public at this time. If someone also has this problem pls contact me. If this solution also helps on other servers, I will release the information on what is going wrong.
 
It's good to hear that you've sorted this out.

If there is anything sensitive in the information, I suggest beginning by sending a private message to @IgorG or contacting Plesk Support directly. If not, publishing your finds and getting a possible bug fixed would certainly be beneficial to all...
 
Last edited:
No one from plesk careas about this issue.....

Example: Zones are not transferred to slave DNS configured via Slave DNS Manager

And theres no chance to get help from Plesk, because we have a licence from a plesk partner.... And I'm not sure if my solution is the right way, and if it is the same problem. But what I have learned in the last days is how this problem comes into plesk, how I can reproduce it, how I can fix it and what's wrong inside....
So for me first, I'd like to double check this problem. So that nobody can crash his plesk installation in case of false informations...
 
Hi @tho .

can you go to bash and post the result of this plesk script:

Code:
plesk db "select * from ServiceNodeConfiguration Where section ='dnsConnector' "
 
  • Like
Reactions: tho
Great to hear that.....

It's a Plesk bug, you can only have ONE DNS extension at this time at he server, but there's no blocks to install more. This breaks this configuration.....
 
  • Like
Reactions: tho
Great to hear that.....

It's a Plesk bug, you can only have ONE DNS extension at this time at he server, but there's no blocks to install more. This breaks this configuration.....
Would it be possible to add multiple DNS extension through the same panel or maybe by adding third party plugin ?? Can you check this out
 
This hast to be answered by the Plesk developers.
I think the must first resolve this bug and build a solution for that issue that more than one DNS extension can be used and configured.
 
Got it,

Code:
plesk db dump psa>psa.dump

plesk db "select * from ServiceNodeConfiguration Where section ='dnsConnector' "

this showed both plesk and DigitalOceans DNS manager were in play even though the DigiOcean manager was removed

plesk db "delete from ServiceNodeConfiguration where value like '%digitalocean%'; "

Uninstall the Plesk DNS manager

Reinstall the Plesk DNS manager

reconfigure slave DNS servers and you should be good to go.

The DNS managers should be able to play with each other or not be installed together.

Thanks Powie!

Dave_W
 
Yes..... the rule is: you can only have one slave dns extension at a time. Install a second one breaks the other.....
 
Hi Powie,
Just to update, Ive flagged this with Plesk and theyre releasing an update on the DNS manager to show a warning if there is already a manager installed. Supposed to be out this month.
Thanks
Dave
 
Back
Top