Problems with a Failover Pair

[David Liedke]

Hi,

i have an Failover Pair with DHCPD Version 4.2.6 on openSUSE 13.2.

Primary:

#OMAPI
omapi-port 7911;
omapi-key primary;
key primary {
     algorithm hmac-md5;
     secret "XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX";
}

failover peer "noc" {
  primary; # declare this to be the primary server
  address XX.XX.XX.XX;
  port 519;
  peer address XX.XX.XX.XX;
  peer port 520;
  max-response-delay 40;
  max-unacked-updates 10;
  load balance max seconds 3;
  mclt 1800;
  split 128;
}


Secondary:

#OMAPI
omapi-port 7911;
omapi-key secondary;
key secondary {
     algorithm hmac-md5;
     secret "XXX";
}

failover peer "noc" {
  secondary; # declare this to be the primary server
  address XX.XX.XX.XX;
  port 520;
  peer address XX.XX.XX.XX;
  peer port 519;
  max-response-delay 40;
  max-unacked-updates 10;
  load balance max seconds 3;

}

On both Server we use a Cronjob to renew the DHCPD Configuration.
Normally there are only changes for the static configuration (for
example: new mac-adress).

Now we had the problem after we execute the script and reboot the dhcpd
server that the server go to the "recover" state. And he doesn´t came
back online (normal state).

Normally we execute the Cronjob on the primary server two times per
hour.  Two minutes after the primary server the cronjob on the secondary
will be executed.
In the last 5 years we use the method to renew the configuration without
any problems.

I want to test the option to send a shutdown signal via OMAPI to the
server i want to renew. But i get always "not connected oder peer
rejected connection".

Can anyone help me ?

Kind regards,
David



_______________________________________________
dhcp-users mailing list
[hidden email]
https://lists.isc.org/mailman/listinfo/dhcp-users

[Steven Carr]

> On both Server we use a Cronjob to renew the DHCPD Configuration.
> Normally there are only changes for the static configuration (for
> example: new mac-adress).
>
> Now we had the problem after we execute the script and reboot the dhcpd
> server that the server go to the "recover" state. And he doesn´t came
> back online (normal state).

What does said script do exactly? (provide the commands if you can).

It sounds like your script is putting the systems into PARTNER-DOWN
mode. Just do a plain restart of the DHCP service, don't tell it that
something is shutting down. When you do a restart the other peer will
go into COMMUNICATIONS-INTERRUPTED then once they see each other again
will work themselves back into NORMAL state.

Steve
_______________________________________________
dhcp-users mailing list
[hidden email]
https://lists.isc.org/mailman/listinfo/dhcp-users

[Shawn Routhier]

You don’t say if this is an issue that started occurring when you switched to 4.2.6.

In 4.2.6 we added a gentle shutdown feature to tidy up the shutdown process.
Unfortunately this created several problems with the servers taking some time
to fully shutdown and so in 4.2.7 and above this feature is no longer on by default.

You might want to upgrade to a newer version of 4.2 or 4.3 and see if that resolves
the issue.

regards,
Shawn

> On Sep 29, 2015, at 1:38 AM, David Liedke <[hidden email]> wrote:
>
> Hi,
>
> i have an Failover Pair with DHCPD Version 4.2.6 on openSUSE 13.2.
>
> Primary:
>
> #OMAPI
> omapi-port 7911;
> omapi-key primary;
> key primary {
>     algorithm hmac-md5;
>     secret "XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX";
> }
>
> failover peer "noc" {
>  primary; # declare this to be the primary server
>  address XX.XX.XX.XX;
>  port 519;
>  peer address XX.XX.XX.XX;
>  peer port 520;
>  max-response-delay 40;
>  max-unacked-updates 10;
>  load balance max seconds 3;
>  mclt 1800;
>  split 128;
> }
>
>
> Secondary:
>
> #OMAPI
> omapi-port 7911;
> omapi-key secondary;
> key secondary {
>     algorithm hmac-md5;
>     secret "XXX";
> }
>
> failover peer "noc" {
>  secondary; # declare this to be the primary server
>  address XX.XX.XX.XX;
>  port 520;
>  peer address XX.XX.XX.XX;
>  peer port 519;
>  max-response-delay 40;
>  max-unacked-updates 10;
>  load balance max seconds 3;
>
> }
>
> On both Server we use a Cronjob to renew the DHCPD Configuration.
> Normally there are only changes for the static configuration (for
> example: new mac-adress).
>
> Now we had the problem after we execute the script and reboot the dhcpd
> server that the server go to the "recover" state. And he doesn´t came
> back online (normal state).
>
> Normally we execute the Cronjob on the primary server two times per
> hour.  Two minutes after the primary server the cronjob on the secondary
> will be executed.
> In the last 5 years we use the method to renew the configuration without
> any problems.
>
> I want to test the option to send a shutdown signal via OMAPI to the
> server i want to renew. But i get always "not connected oder peer
> rejected connection".
>
> Can anyone help me ?
>
> Kind regards,
> David
>
>
> _______________________________________________
> dhcp-users mailing list
> [hidden email]
> https://lists.isc.org/mailman/listinfo/dhcp-users

_______________________________________________
dhcp-users mailing list
[hidden email]
https://lists.isc.org/mailman/listinfo/dhcp-users

[David Liedke]

Am 29.09.2015 um 21:29 schrieb Shawn Routhier:

> You don’t say if this is an issue that started occurring when you switched to 4.2.6.
>
> In 4.2.6 we added a gentle shutdown feature to tidy up the shutdown process.
> Unfortunately this created several problems with the servers taking some time
> to fully shutdown and so in 4.2.7 and above this feature is no longer on by default.
>
> You might want to upgrade to a newer version of 4.2 or 4.3 and see if that resolves
> the issue.
>
> regards,
> Shawn
>

In April i updated one Server to openSUSE 13.2 (DHCPD 4.2.6).  The other
server run with openSUSE 13.1 (DHCPD 4.2.5). With this combination i
have no problems. Last week i update the second server to openSUSE 13.2
(DHCPD 4.2.6) and the problems begin.

Okay thanks for the information. I will test it.


Am 29.09.2015 um 15:10 schrieb Steven Carr:

>> On both Server we use a Cronjob to renew the DHCPD Configuration.
>> Normally there are only changes for the static configuration (for
>> example: new mac-adress).
>>
>> Now we had the problem after we execute the script and reboot the dhcpd
>> server that the server go to the "recover" state. And he doesn´t came
>> back online (normal state).
> What does said script do exactly? (provide the commands if you can).
>
> It sounds like your script is putting the systems into PARTNER-DOWN
> mode. Just do a plain restart of the DHCP service, don't tell it that
> something is shutting down. When you do a restart the other peer will
> go into COMMUNICATIONS-INTERRUPTED then once they see each other again
> will work themselves back into NORMAL state.
>
> Steve

We have a database with the bindings between mac-adress and ip adress.
The script generate the new configuration files.  After generating the
new files the script restart the dhcp daemon without any command to
shutdown or something like that.


Regards,
David


_______________________________________________
dhcp-users mailing list
[hidden email]
https://lists.isc.org/mailman/listinfo/dhcp-users