[Samba] PVE HA for DCs

[Luis Peromarta]

Thanks so much Matthias.

This exactly how I am doing it, knowing the “built in” redundancy in Samba is there by adding DCs. I have 4.

I was just wondering if it was a good idea to activate HA on these DCs - like I have in other VMs and CTs- , witch I agree is not needed.

What I am using is CEPH storage, and I think you are right and these DCs can and should be run on local storage.

There is enough redundancy built into the system.

Thank you all.

LP
On 5 Jun 2024 at 07:01 +0100, Matthias Kühne | Ellerhold Aktiengesellschaft via samba <samba at lists.samba.org>, wrote:
> Hello Luis,
>
> I think youre going at this problem the wrong way.
>
> Samba DC is built with multiple servers in mind. There is no need for
> external HA if you have enough DCs already.
>
> e. g. you have 3 nodes in your cluster: put a DC on each node and
> disable Proxmox HA alltogether. If one (or even two) nodes go down, your
> domain is still active because of the last DC.
>
> Its best to use different storages as well, so that a storage failure
> does not bring down your whole domain and locks you out of admin
> interfaces etc...
>
> As soon as your nodes recover, reboot the DCs on them and they'll sync
> again.
>
> If you have to decommision a node, then you can just move the DC from
> the old node to the new one and boot them there.
>
> Hope this helps.
>
> Regards, Matthias.
>
> Am 04.06.24 um 21:53 schrieb Luis Peromarta via samba:
> > Not that there’s a software problem but maybe like a power failure. Say a non-gracefully shut down server.
> >
> > Replication wise, is it ok to just reboot it ?
> >
> > LP
> > On 4 Jun 2024 at 20:11 +0100, samba at lists.samba.org <samba at lists.samba.org>, wrote:
> > > Well, a Samba DC shouldn't crash, but if it does, you need to find out
> > > why it crashed, there really isn't much point in restarting a DC if it
> > > is just going to crash again.
> > >
> > > If the 'crash' was caused by an easily fixed error and can be fixed
> > > quickly, you should be able to restart it and it should replicate from
> > > any other DCs. You can, theoretically, restart a DC quite some time
> > > after it is stopped, but it may just be easier to replace it.
> > >
> > > So, to put it into a nutshell, if a DC is stopped for a few hours, then
> > > it should be safe to restart it, but if comes to days or weeks, then I
> > > would replace it, better safe than sorry :-)
>
> --
> Senior Webentwickler
> Datenschutzbeauftragter
>
> Ellerhold Aktiengesellschaft
> Friedrich-List-Str. 4
> 01445 Radebeul
>
> Telefon: +49 (0) 351 83933-61
> Web: www.ellerhold.de
> Facebook: www.facebook.com/ellerhold.gruppe
> Instagram: www.instagram.com/ellerhold.gruppe
> LinkedIn: www.linkedin.com/company/ellerhold-gruppe
>
> Amtsgericht Dresden / HRB 23769
> Vorstand: Stephan Ellerhold, Maximilian Ellerhold
> Vorsitzender des Aufsichtsrates: Frank Ellerhold
>
>
>
> ---Diese E-Mail und Ihre Anlagen enthalten vertrauliche Mitteilungen. Sollten Sie nicht der beabsichtigte Adressat sein, so bitten wir Sie um Mitteilung und um sofortiges löschen dieser E-Mail und der Anlagen.
>
> Unsere Hinweise zum Datenschutz finden Sie hier: http://www.ellerhold.de/datenschutz/
>
> This e-mail and its attachments are privileged and confidential. If you are not the intended recipient, please notify us and immediately delete this e-mail and its attachments.
>
> You can find our privacy policy here: http://www.ellerhold.de/datenschutz/
>
>
> --
> To unsubscribe from this list go to the following URL and read the
> instructions: https://lists.samba.org/mailman/options/samba