Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: DRBD: Users

General question about DRBD and HA

 

 

DRBD users RSS feed   Index | Next | Previous | View Threaded


prolinux at web

Apr 23, 2012, 11:17 PM

Post #1 of 3 (361 views)
Permalink
General question about DRBD and HA

Hello,

on the DRBD homepage I found the following sentence:

"After an outage of a node DRBD automatically resynchronizes the temporarily unavailable node to the latest version of the data, in the background, without interfering with the service running."

In a HA scenario where the master server has gone down, DRBD takes care of syncing the data from slave to master.
But how does the HA software know, when syncing has finished, so that the master server can take his role again?
I know this is not a DRBD question, but I guess that somehow DRBD has to tell the HA software, that everything is okay again. Silly question?

Thanks for any hints/advice,

Stefan



Ihr WEB.DE Postfach immer dabei: die kostenlose WEB.DE Mail App für iPhone und Android.
https://produkte.web.de/freemail_mobile_startseite/"]https:// produkte.web.de/freemail_mobile_startseite/


pascal.berton3 at free

Apr 24, 2012, 4:39 AM

Post #2 of 3 (331 views)
Permalink
Re: General question about DRBD and HA [In reply to]

Stefan,

If the master fails, then the slave is promoted master and takes the
service over. At this time, it's the only node alive. When the former
master returns to life, its data will be outdated compared to the
current master. When it joins again the cluster, DRBD will detect that
inconsistency and sync its data (former master datas) against the
official master one (current master). DRBD embeds a bitmap scheme to
identify the outdated datas, it knows precisely what is up to date and
what is not. There are a couple of pages discussing that by the end of
the DRBD Users Guide.

Best regards,

Pascal.

Le 24/04/2012 08:17, "Stefan Günther" a écrit :
> Hello,
>
> on the DRBD homepage I found the following sentence:
>
> "After an outage of a node DRBD automatically resynchronizes the
> temporarily unavailable node to the latest version of the data, in the
> background, without interfering with the service running."
>
> In a HA scenario where the master server has gone down, DRBD takes
> care of syncing the data from slave to master.
> But how does the HA software know, when syncing has finished, so that
> the master server can take his role again?
> I know this is not a DRBD question, but I guess that somehow DRBD has
> to tell the HA software, that everything is okay again. Silly question?
>
> Thanks for any hints/advice,
>
> Stefan
>
>
>
> Ihr WEB.DE Postfach immer dabei: die kostenlose WEB.DE Mail App für
> iPhone und Android.
> *https:// produkte.web.de/freemail_mobile_startseite/*
> <https://produkte.web.de/freemail_mobile_startseite/>
>
>
>
> _______________________________________________
> drbd-user mailing list
> drbd-user [at] lists
> http://lists.linbit.com/mailman/listinfo/drbd-user


ff at mpexnet

May 2, 2012, 1:44 AM

Post #3 of 3 (281 views)
Permalink
Re: General question about DRBD and HA [In reply to]

On 04/24/2012 01:39 PM, Pascal BERTON wrote:
> Stefan,
>
> If the master fails, then the slave is promoted master and takes the
> service over. At this time, it's the only node alive. When the former
> master returns to life, its data will be outdated compared to the
> current master. When it joins again the cluster, DRBD will detect that
> inconsistency and sync its data (former master datas) against the
> official master one (current master). DRBD embeds a bitmap scheme to
> identify the outdated datas, it knows precisely what is up to date and
> what is not. There are a couple of pages discussing that by the end of
> the DRBD Users Guide.

Also, DRBD is very flexible as long as a stable connection exists
between the nodes. As soon as the cluster is reconnected, the (possibly
still partially outdated) original master can become primary again and
start hosting services.
Of course, this is not the safest thing to do (in case of another
breakage during the sync).

A minor gripe with pacemaker at the moment (afaik) is that it does in
fact *not* have much knowledge of DRBD consistency states. I'm not sure
if it can make good decisions that take this sort of resync into account.

HTH,
Felix

> Le 24/04/2012 08:17, "Stefan Günther" a écrit :
>> Hello,
>>
>> on the DRBD homepage I found the following sentence:
>>
>> "After an outage of a node DRBD automatically resynchronizes the
>> temporarily unavailable node to the latest version of the data, in the
>> background, without interfering with the service running."
>>
>> In a HA scenario where the master server has gone down, DRBD takes
>> care of syncing the data from slave to master.
>> But how does the HA software know, when syncing has finished, so that
>> the master server can take his role again?
>> I know this is not a DRBD question, but I guess that somehow DRBD has
>> to tell the HA software, that everything is okay again. Silly question?

--
MPeX.net GmbH / Werner-Voß-Damm 62 / D-12101 Berlin / Germany
MPeXnetworks / www.mpexnetworks.de
Tel: ++49-30-78097 180 / Fax: ++49-30-78097 181

Sitz, Registergericht: Berlin, Amtsgericht Charlottenburg, HRB 76688
Geschäftsführer: Lars Bräuer, Gregor Rex-Lawatscheck
_______________________________________________
drbd-user mailing list
drbd-user [at] lists
http://lists.linbit.com/mailman/listinfo/drbd-user

DRBD users RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact Gossamer Threads
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.