[DRBD-user] configuring 2 services on 2 hosts

Digimer linux at alteeve.com
Wed Jan 5 21:35:35 CET 2011

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.

On 01/05/2011 03:26 PM, J wrote:
> On 1/5/2011 12:49 PM, Digimer wrote:
>> On 01/05/2011 11:48 AM, J wrote:
>>> Hello,
>>> This is my first go at setting up any HA services, so I wanted to see if
>>> anyone on the list had a similar setup or any recommendations on doing
>>> this better using drbd.
>>> I'm setting up 2 servers with 2 application services on them. One
>>> application service is postgresql database (about 5GB of data in
>>> database). The other is custom software that does a lot of file-based
>>> I/O (about 15GB of files) with a custom TCP server in front of it.
>>> My idea is to have one server being the master for the custom software
>>> and the other being the master
>>> for the postgresql database. The other server will be a stand by for the
>>> other service. Both will run run ubuntu 10.04 LTS and be connected by 2
>>> cat6 1Gb cables.
>>> This will hopefully distribute the load, as I am worried about the
>>> postgresql clobbering the custom software (which requires very high
>>> responsiveness).
>>> So is setting up 2 brbd's in opposite configurations on 2 hosts a good
>>> idea? I'm mostly worried about
>>> performance and the headache of admin if the links go down. 2 cases of
>>> split brain on 2 boxes
>>> doesn't sound fun at all.
>>> Thanks!
>>> J
>> If you are using platter-based HDDs, then I'd strongly recommend
>> creating two separate resources on two separate HDD(s). This way you
>> won't suffer I/O penalties with the read/write heads flying back and
>> forth.
>> Give your relatively small storage requirements though, I'd even more
>> highly recommend SSDs providing that the write-cycle limit is within
>> reason. In this setup, then you should have no trouble running it all on
>> one resource as there are no heads to flog.
>> In any case, be sure to have (at least) RAID 1 on each node backing the
>> DRBD devices to help minimize downtime. Drives fail fairly frequently...
>> software RAID 1 is an inexpensive route to much better uptime. :)
> Thanks for the recommendation.
> I agree, I think that SSD's will be a great solution and I'll put in for
> 4 256GB ones *sweet*.  I've still got a lot of things to figure out, so
> I'm sure I'll be back. Thanks for your help!
> J

Anytime, have fun! :)

E-Mail: digimer at alteeve.com
AN!Whitepapers: http://alteeve.com
Node Assassin:  http://nodeassassin.org

More information about the drbd-user mailing list