DirectSlave - solution to add and remove zones on slave without DirectAdmin

What do you mean by "read all TXT value from master"?

Since now (and i'm using it since months), i've never faced a single problem (except for perl softwares on centOS 5)

Regards
 
That's a nice question actually xD I dont remember if Roman gave me this or the 2.1 on first page is actually 2.2.

Regarding DNS, how the older version does work?

At this point (but as far as i remember i always had this way), DA send the new domain info's to the slave, than yes the slave named require to the master all the needed data's and store them locally, and when on master something get changed, named send notification to slaves (and so should do DA aswell).

As i wrote, as far as i can remember i've used this always this way, dont remember different way to go on previous release (i'm using DS since was made public).

Regards
 
In older version you can choose named behavior, like master or slave. In version 2.1 master is deprecated. Read documentation :)
 
Ok, i did read.. sorry it is passed a lot of time since i did setup the nameservers xD

But it shouldn't be a big problem this way to act, is it?

Regards
 
I still have a problem with DNSSEC. After a month the DNSSEC does not work correctly anymore. It seems like the keys are not updated like they should.

If i do create a new a record for the domain on the directadmin server it works again because the records are updated. So it seems like Directslave does not update the DNSSEC records like it should.
 
I have read almost all pages , i try to find a solution for and extra DNS server.

I only have 2 questions.

1) how is the author (Roman) doing ? since i see is living in Ukraine. (we all know the situation there)

2) should i use this option on CentOS 6 ? you guys recommend this ? (my option is still for one DA license only for DNS and use multi server option, since i see that the bugs are fixed - (google apps mx transfer , double A zones .. and others) )

Thanks
 
Honestly i dont know how is Roman, is quite a bit i dont read his posts..

Regarding your second question, Yes, DS does work perfectly on CentOS 6. I do use DS in production from about 2 years (started using it just after was out or a few days/months later) and is working pretty fine.

I do use it on CentOS 5 and the only problem i had lately was perl issue, but on CentOS 6 you would not have those issue.

I'm forced to use CentOS 5 OS cause my hardware are old DELL Poweredge 850 (which are perfect for DNS Services) and CentSO 6 apparently doenst work well with those :)

Best regards
 
If I recall correctly, Roman started the project with some of my code from Master2Slave DNS Replicator.

I'm working on a new project which will be offering SlaveDNS on multiple servers for less than the cost of most low-end Virtual Private Servers.

And I'm basing my code on DirectSlave. So in the future I'll need to maintain it if necessary.

I'd say 'no worries'. :)

Jeff
 
Hi!
I'm trying this tool, but it's not working for me.
I'm getting this error on DA server (master) when I try an action=rewrite:

2015:04:04-01:25:08: Cluster: Error writing remote dns zone at host xxx.com for domain (yyy.com): (null) (retry=0)
2015:04:04-01:25:08: Cluster socket:

I tried configuring "allow-notify" and "notify" as I read in this thread, with no success.

Also, the slave server adds a line with the zone to /etc/named/directslave.conf , but doesn't add the domain to /etc/named/secondary/
well, at least didn't until I patched dns_admin() function with something I think makes sense. When action is "rawsave", apparently it only calls add() and then reload(), but add() seems to only add the line to directslave.conf, not create the zone itself. My patch was to call rewrite() before reload(). Anyway, the problem persists.
And I'm also not sure my patch makes sense, because I'm pretty sure someone would have noticed this before me.

Any ideas?
 
I found out the sync is made by AXFR, but it's still not working. I'm getting this:
Apr 4 06:28:48 xxx named[26760]: client yyy#31657: received notify for zone 'zzz': not authoritative


my master config has:
notify yes;
also-notify { SLAVE-SERVER; };


my slave config has:
allow-query { MASTER-SERVER; };
allow-transfer { localhost; MASTER-SERVER; };
allow-notify { localhost; MASTER-SERVER; };

(AFAIK it's not really needed since the zone is configured with the master, but I was just trying to make sure)
 
Don't you need also the allow-transfer for the slave server in your master config?
I've got both allow-notify and allow-transfer for the slave server in there and it's working fine.
 
Also, the slave server adds a line with the zone to /etc/named/directslave.conf
My slave server does not put anything in /etc/named/directslave.conf which I've got in /usr/local/directslave/etc.
It however does put lines in /etc/secondary/named.conf but this may depend on how you set it up.

You did use the /32 in the main directslave.conf for your master server, correct?
Code:
host            SLAVE.SERVER.IP
allow           master.server.ip/32
correct?
 
huh! you gave me a light!
by default directslave sets named_conf /etc/named/directslave.conf (actually, /namedb/, but I changed it)
the problem is that I didn't think about adding this include to /etc/named.conf :p
now I did and everything seems to be working
it's replicating correctly, although I still get this on DA master log:

Cluster socket:
Cluster: Error writing remote dns zone at host SLAVE-SERVER.com for domain

is that normal?
 
As far as I know, you should not receive writing error messages.
However, you helped me too, because I went to check and discovered my zones are not transferred anymore either.
I get a "Unable to open a socket" when I try a test connection from my master server, so I have to investigate why this is.
 
@Sellerone: Do you know what I can do to fix this?
I have the correct ip's, but I can't even connect via "telnet localhost 2222" it says "connection refused" and from the master server it says "unable to open socket".
I'm confused because I did not change anything.
 
Very odd. So I did another test connection from my master server, with ssl enabled. Connection was fine.
Then I did another test connection without ssl enabled and it also gives an OK. I had restarted Directslave in between, but at first it still gave the socket error.

I just did a test and I don't have that "writing remote dns zone" which you have, but this was done via the datatask.queue.
Code:
Apr  4 18:09:02 server19 named[8515]: client 141.xxx.xxx.xxx#36147: transfer of 'testdomain.com/IN': AXFR-style IXFR started
No error notices.

So I did another test by creating a new user with a new domain which gave this:
Code:
Apr  4 18:11:02 server19 named[8515]: zone test2domain.com/IN: sending notifies (serial 2015040400)
Apr  4 18:11:02 server19 named[8515]: client 141.xxx.xxx.xxx#37314: transfer of 'test2domain.com/IN': AXFR started
Apr  4 18:11:02 server19 named[8515]: client 141.xxx.xxx.xxx#37314: transfer of 'test2domain.com/IN': AXFR ended
Apr  4 18:11:02 server19 proftpd[1700]: 217.xxx.xxx.xxx - received SIGHUP -- master server reparsing configuration file

As you can see, no error notices again. So imho there should not be any writing error notices.

The only issue which I seem to have is that when I delete a user (and so his domain) on the master server, the domain won't get deleted on the slave server.
 
Last edited:
Back
Top