how to upgrade the failover pair?

[Aleksey Perov]

Hi.

I have two servers running Debian 6.0.5 with ISC DHCP 4.1.1 in a
failover pair. Now I need to upgrade them to the latest Debian 9 which
is shipped with ISC DHCP 4.3.5. Since Debian 6 is too old, the only way
of upgrading is a clean reinstall (which in most cases is better than
sequential upgrade 6 to 7 to 8 to 9).

I have read
https://kb.isc.org/article/AA-01043/183/Recommendations-for-restarting-a-DHCP-failover-pair.html 
and some other articles, but those documents say nothing about the
upgrade scenario.

So the question is: What is the recommended process of _upgrading_ the
failover pair?

The draft plan is as follows:

- stop the secondary;
- turn the primary into partner-down;
- reinstall the secondary;
- wait until the primary reclaims all addresses in all pools;
- stop the primary;
- copy leases file from primary to secondary;
- start the secondary in partner-down mode;
- reinstall the primary;
- turn the secondary into normal mode;
- start the primary.

Is anything wrong here?

Any advices and recommendations are welcome. Thanks.




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

[perl-list]

This is a layman's opinion as i've not done the described procedure below, but I would think you would want to retain the dhcpd.leases file from the secondary and primary and put those back after re-install.  

In other words, do not attempt to copy leases file from primary to secondary as you listed in steps below, but rather retain the dhcpd.leases file from the original install.  Copy that file back into place before you start the DHCP server on the secondary.

I think there are some differences between the primary and secondary leases files such that you would not want one copied to the other.

---

From: "Aleksey Perov" <[hidden email]>
To: "Users of ISC DHCP" <[hidden email]>
Sent: Thursday, May 31, 2018 3:14:47 AM
Subject: how to upgrade the failover pair?

Hi.

I have two servers running Debian 6.0.5 with ISC DHCP 4.1.1 in a
failover pair. Now I need to upgrade them to the latest Debian 9 which
is shipped with ISC DHCP 4.3.5. Since Debian 6 is too old, the only way
of upgrading is a clean reinstall (which in most cases is better than
sequential upgrade 6 to 7 to 8 to 9).

I have read
https://kb.isc.org/article/AA-01043/183/Recommendations-for-restarting-a-DHCP-failover-pair.html
and some other articles, but those documents say nothing about the
upgrade scenario.

So the question is: What is the recommended process of _upgrading_ the
failover pair?

The draft plan is as follows:

- stop the secondary;
- turn the primary into partner-down;
- reinstall the secondary;
- wait until the primary reclaims all addresses in all pools;
- stop the primary;
- copy leases file from primary to secondary;
- start the secondary in partner-down mode;
- reinstall the primary;
- turn the secondary into normal mode;
- start the primary.

Is anything wrong here?

Any advices and recommendations are welcome. Thanks.




--
Aleksey
_______________________________________________
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

[Simon Hobson]

Aleksey Perov <[hidden email]> wrote:

> The draft plan is as follows:
>
> - stop the secondary;
> - turn the primary into partner-down;
> - reinstall the secondary;
> - wait until the primary reclaims all addresses in all pools;
> - stop the primary;
> - copy leases file from primary to secondary;
> - start the secondary in partner-down mode;
> - reinstall the primary;
> - turn the secondary into normal mode;
> - start the primary.
>
> Is anything wrong here?

Isn't the simplest - provided they are using the same level of failover protocol - to simply start up a clean upgraded secondary and allow it to transfer leases from the primary ? So :

Stop one partner, put the remaining one into partner down state.
Create new upgraded clean partner.
Start up new partner and allow the pair to synchronise.

Repeat for the other partner.


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

[Aleksey Perov]

You mean 4.1.1 and 4.3.5 use the same failover protocol and can talk to
each other without any issue?


31.05.2018 12:41, Simon Hobson пишет:

> Aleksey Perov <[hidden email]> wrote:
>
>> The draft plan is as follows:
>>
>> - stop the secondary;
>> - turn the primary into partner-down;
>> - reinstall the secondary;
>> - wait until the primary reclaims all addresses in all pools;
>> - stop the primary;
>> - copy leases file from primary to secondary;
>> - start the secondary in partner-down mode;
>> - reinstall the primary;
>> - turn the secondary into normal mode;
>> - start the primary.
>>
>> Is anything wrong here?
>
> Isn't the simplest - provided they are using the same level of failover protocol - to simply start up a clean upgraded secondary and allow it to transfer leases from the primary ? So :
>
> Stop one partner, put the remaining one into partner down state.
> Create new upgraded clean partner.
> Start up new partner and allow the pair to synchronise.
>
> Repeat for the other partner.
>
>
> _______________________________________________
> dhcp-users mailing list
> [hidden email]
> https://lists.isc.org/mailman/listinfo/dhcp-users
>

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

[Bill Shirley-2]

I am running a fail-over pair:
primary = dhcp-server-4.3.6-9.fc27.x86_64
secondary = dhcp-4.2.5-23.fc19.x86_64

So yes, I think 4.1.1 and 4.3.5 would speak the same language.

Bill


On 5/31/2018 6:40 AM, Алексей Перов wrote:

> You mean 4.1.1 and 4.3.5 use the same failover protocol and can talk to each other without any issue?
>
>
> 31.05.2018 12:41, Simon Hobson пишет:
>> Aleksey Perov <[hidden email]> wrote:
>>
>>> The draft plan is as follows:
>>>
>>> - stop the secondary;
>>> - turn the primary into partner-down;
>>> - reinstall the secondary;
>>> - wait until the primary reclaims all addresses in all pools;
>>> - stop the primary;
>>> - copy leases file from primary to secondary;
>>> - start the secondary in partner-down mode;
>>> - reinstall the primary;
>>> - turn the secondary into normal mode;
>>> - start the primary.
>>>
>>> Is anything wrong here?
>>
>> Isn't the simplest - provided they are using the same level of failover protocol - to simply start up a clean upgraded
>> secondary and allow it to transfer leases from the primary ? So :
>>
>> Stop one partner, put the remaining one into partner down state.
>> Create new upgraded clean partner.
>> Start up new partner and allow the pair to synchronise.
>>
>> Repeat for the other partner.
>>
>>
>> _______________________________________________
>> 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

[Bob Harold]

On Thu, May 31, 2018 at 11:26 AM Bill Shirley <[hidden email]> wrote:

I am running a fail-over pair:
primary = dhcp-server-4.3.6-9.fc27.x86_64
secondary = dhcp-4.2.5-23.fc19.x86_64

So yes, I think 4.1.1 and 4.3.5 would speak the same language.

Bill


On 5/31/2018 6:40 AM, Алексей Перов wrote:
> You mean 4.1.1 and 4.3.5 use the same failover protocol and can talk to each other without any issue?

You could try it, and if it fails to sync it should give some error in the logs.

-- 

Bob Harold

 

>
> 31.05.2018 12:41, Simon Hobson пишет:
>> Aleksey Perov <[hidden email]> wrote:
>>
>>> The draft plan is as follows:
>>>
>>> - stop the secondary;
>>> - turn the primary into partner-down;
>>> - reinstall the secondary;
>>> - wait until the primary reclaims all addresses in all pools;
>>> - stop the primary;
>>> - copy leases file from primary to secondary;
>>> - start the secondary in partner-down mode;
>>> - reinstall the primary;
>>> - turn the secondary into normal mode;
>>> - start the primary.
>>>
>>> Is anything wrong here?
>>
>> Isn't the simplest - provided they are using the same level of failover protocol - to simply start up a clean upgraded
>> secondary and allow it to transfer leases from the primary ? So :
>>
>> Stop one partner, put the remaining one into partner down state.
>> Create new upgraded clean partner.
>> Start up new partner and allow the pair to synchronise.
>>
>> Repeat for the other partner.

 

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