[DRBD-user] DRBD8, dedicated repository for PVE 4

Cesar Peschiera brain at click.com.py
Mon Dec 26 09:35:44 CET 2016

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

Hello Jasmin

Thanks for all your clarifications.
And let me to do a comments...

> The only thing which is currently not supported is life migration. The
> plugin checks if the other side is in secondary, otherwise it will not
> activate the DRBD8 volume.
> It might be possible to implement this (remove the check in the plugin)
> and use dual primary mode, but ... .
> In my opinion it is not necessary to life migrate, but others may see this
> different and extend the plugin to allow this and ofcourse test, if this
> is really working without getting a split brain.

In my experience of several years working with DRBD and PVE on mode
primary-primary, is that sometimes i have my DRBD resources desynchronized,
but never due to the "split brains", from that i configure the PVE Nodes in
bonding-rr in mode NIC-to-NIC between my PVE hosts only for the DRBD
resources (without a switch in the middle), and i think that this error is
due to a bug(s) in the DRBD code.

For me, this ugly behavior is avoided completely if i change the
configuration of the DRBD resources to primary-secondary mode, that is as i
have it configured now.
Then, as I compile, install and configure the DRBD software (in mode
primary-secondary), i have a perfect control of everything, and I can do:
- Live migration of VMs (with previous and subsequent manual steps of DRBD
commands(change status of primary  and secondary DRBD resources)).
- Apply HA manually (changing manually to primary the DRBD resource in the
host that is live, and after, also manually execute the command
"fence_ack_manual name_of_the_host_that_is_dead").

But if i want to use your plugin, i will lose the option of use live
migration, that it is important for me, and that's why i need to have and
second plugin (or something similar) that can mitigate with the primary and
secondary DRBD resources and do a live migrations of VMs successfully.

Moreover, I think that if the PVE developer team do a API (or something
similar) that can execute commands previous and subsequent to the live
migration, of this way, we will have the better of PVE for our DRBD
resources ("HA" and "live migration").

So, my question is if you can to talk with the PVE developer team for that
them can do the API (or something similar) so that can work the "Live
Migration" sucessfully without the need of execute manually some previous
and subsequent DRBD commands, ie, change the state of the primary and
secondary DRBD resources as part of the process of the live migration.
For example, vzdump has the option of execute commands previous and
subsequent of his own execution, and i guess that also will be possible have
it available for the live migration.

And finally, after of that (as i am not a developer nor know your plugin
inwardly), maybe also be need to change something in the code of your plugin
for that works well (due to that for a short time, in the process of the
live migration, we will need that both of them DRBD resources are in primary

So I'd like to know what you think about all that.

> I had a similar configuration before, but now I am free with the number of
> disks and number of VMs which I have.
Please, excuse me for my error, I meant to say that i use one or more
partitions of disk for each VM (without limit), of this way, if my DRBD
resources are desynchronized, quickly and in online mode, i can
resynchronize them without some risk (generally some few seconds for
resynchronize them).

> To be honest, I don't trust the PVE cluster manager here. Or better, I
> don't know if it would be possible to write really from both sides the
> same disk blocks. If someone likes to test it, I wrote above how it might
> work with the existing plugin by simply removing the check for the other
> side being secondary.
If write really from both sides to the same disk at the same time, we will
get a serious and real disaster.

> If someone likes to test it, I wrote above how it might work with the
> existing plugin by simply removing the check for the other side being
> secondary.
I would like to use your plugin in primary-secondary mode, and never of
other way, but as also i need the live migration, then, i need that both
features work well.
For this reason i use my own DRBD compiled (as i explained above), but as
the manual execution of DRBD commands isn't elegant for apply a "HA" or a
"Live Migration" (as also i explained above), then, i would like to have
these features availables without have that manually execute some command.

> > I did several tests about that, and i can say with certainty that in
> > DRBD8
> > the online resize only is possible if the VM is not running and the DRBD
> > resource is unmounted in the host (very tested in my lab with his
> > possible
> > alternatives, always with the condition that resource DRBD be as small
> > as
> > possible, ie, according to the size of the Virtual disk)
> Thanks for sharing this information!
> So you need to shutdown the VM first. Then the plugin will change to
> secondary on both sides and then you can do the LVM/DRBD8 shrink/extend.
> And
> ofcourse shrinking requires more attention and intelligence to do it.
I don't know if your schema is as the mine...
LVM-Partition / DRBD  / LVM-PV / LVM-VG / LVM-LV(The virtual disk)
Or this...
LVM-Partition / LVM-PV / LVM-VG / LVM-LV /DRBD(The virtual disk)
or what ???
But i can to say that my process of extend the vitual disk is very complex,
due to that i do several things, as delete the partition and recreate it
more big, up and down the DRBD resources as part of the process, as well as
some other things, but very tested in my lab, and inclusive realized in
production environments.

If you are interested in knowing the complete procedure, i can send you by
email my own manual for this procedure, but as i speak spanish, my manual is
done in my own language.

Also i think that this procedure not will be util for create a development
for PVE due to the complexity of the same, as also i have the primordial
condition that each DRBD resource not be more big that the virtual disk.
But if you find a easy way for extend a virtual disk with such condition,
please, let me know it !!!.

> BR
>    Jasmin


More information about the drbd-user mailing list