[DRBD-user] Best practice: drbd+lvm+gfs2+dm-crypt on dual primary

Patrick Prilisauer prilisauer at googlemail.com
Sat Feb 21 20:25:52 CET 2015

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

Thanks to Digimer ;-)

It works now. C2000 Chipset under CentOS 6.6 has some "minor" issues on a
peer to peer connection, even forcing a special transmittion rate (e.g.
100Mbit half) hasn't worked on my Supermicro boards.

Putting a Switch between solved it.

Now everything works fine, except a fencing issue I haven't figured out
yet. The fence_ipmilan works. the IPMI seems so stop responding after a
while after boottime. so corosync fills up the
memory till dead.

Thanks again

2015-02-12 16:13 GMT+01:00 Patrick Prilisauer <prilisauer at googlemail.com>:

> What should I report?
> The Cluster peer to peer worked fine under centos 7 under 6.6 after about
> 5 minutes the devices reports "desitnation unreachable"... and over night
> one node rebooted und could not boot any more,.. the other one booted and
> run fine, but now both don't come up.
> /dev/mapper/vg_at01.....-user is mounted
> e2fsck: cannot continue, aborting
> could this occur by clvmd?
> 2015-02-09 19:12 GMT+01:00 Digimer <lists at alteeve.ca>:
>> On 09/02/15 01:03 PM, Patrick Prilisauer wrote:
>> > Hey,.. well I know many shortcuts ;-P SLES, RHEL aka EL,etc... so
>> > XENserver "must" be fine too.
>> >
>> > My goal is to have:
>> >
>> > a Server for KVM hosting... Mail server (Open change + Cyrus +
>> > postfix,...) Fileserver, AD DC using Samba 4,
>> >
>> > and some more services...
>> >
>> > If you Ask yourself why I try to do this using an Atom 2750 8 Core ad 16
>> > GB on Standard Sata disks I can answer it easily. I have to serve this
>> > services for 5 User and I do not have to care about High IO loads etc,
>> > to separate sequential from random IO to push the performance higher.
>> >
>> > My second Problem is that this server have their place in very small
>> > room without any possibility of cooling. So Atom with 20W TDP each must
>> > do this.
>> >
>> > another reason is the physical size,..
>> >
>> > If I would try to put one of my HP DL 380 I have at home, I would not be
>> > able to close the door..
>> >
>> > So better don't ask why I choose this components. It even took me 2
>> > hours to disassemble and assemble the rack again in fact the door frame
>> > was 3mm smaller :-P
>> >
>> > BUT I must see the good things: SCSI Cluster no longer exists…. My first
>> > and my worst shared storage.
>> I have no thoughts or concerns about performance, that's a question for
>> you and your users. I only care about properly built HA. :)
>> If your goal is to build an HA VM platform, and setup your HA
>> applications in VMs, you could try following the tutorial as it is (cman
>> + rgmanager + drbd). It may not be the most up-to-date set of
>> applications, but it's very well field tested.
>> --
>> Digimer
>> Papers and Projects: https://alteeve.ca/w/
>> What if the cure for cancer is trapped in the mind of a person without
>> access to education?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20150221/7c7d069a/attachment.htm>

More information about the drbd-user mailing list