Preparing the 2.6.2 point release

[Iustin Pop]

Hi all,

We're preparing the 2.6.2 release, and I'd like your feedback to ensure
we don't miss any easy fixes that could go in it.

Currently we have queued:

- Fix disk adoption interaction with ipolicy checks (Issue 255)
- Force tap's MAC prefix to "fe" (Issue 217, backported from master)
- Update instance modify message (per the recent thread)
- Add option to force master-failover without voting (Issue 282)

Any other _small_ fixes that people think could go into this?

We're aiming for a 1st November release, fyi.

thanks,
iustin

[Marco Casavecchia M.]

What about the patch to specify the cpu model per instance?
I'm still stuck to my personally patched 2.5.1 because I need different cpu models in my instances.
In fact, I have nodes with different cpus (amd and intel) and instances that needs 32 and 64 bits cpus.

Can you also remember me if the "no cache" and the "native io" patches are already into the 2.6 branch?

Thanks in advance Iustin.

[Lucas, Sascha]

Hi,

From: Iustin Pop
Date: Mon, 29. Oct 2012 17:26

> We're preparing the 2.6.2 release, and I'd like your feedback to ensure
> we don't miss any easy fixes that could go in it.

I don't know if following is easy to fix. But here is my wish list for what I think it could be easy (unordered):

* KVM: default to cache=none,aio=native parameter
* Issue 261: Possibility to set custom kvm parameters on an instance
* gnt-cluster verify: don't blame for non redundant instances in case of sharedfile...
* gnt-cluster modify --no-lvm-storage: validation bug? Failure: prerequisites not met for this operation: error type: wrong_input, error details: Parameter 'OP_CLUSTER_SET_PARAMS.vg_name' fails validation
* Issue 262: gnt-instance remove does not remove directory on shared file storage path
* gnt-instance modify --disk add:...: does not activate disks (in preparation for hotplug)
* Issue 225: kvm path is configurable only at build time
* similar to Issue 259: respect --prealloc-wipe-disks with grow-disk/modify --disk add

TIA Sascha.

Vorsitzender des Aufsichtsrates: Ralf Hiltenkamp
Geschäftsführung: Michael Krüger (Sprecher), Stephan Drescher
Sitz der Gesellschaft: Halle/Saale
Registergericht: Amtsgericht Stendal | Handelsregister-Nr. HRB 208414
UST-ID-Nr. DE 158253683
Diese E-Mail enthält vertrauliche und/oder rechtlich geschützte Informationen. Wenn Sie nicht der richtige Empfänger sind oder diese E-Mail irrtümlich erhalten haben, informieren Sie bitte sofort den Absender und vernichten Sie diese Mail. Das unerlaubte Kopieren sowie die unbefugte Weitergabe dieser Mail oder des Inhalts dieser Mail sind nicht gestattet. Diese Kommunikation per E-Mail ist nicht gegen den Zugriff durch Dritte geschützt. Die GISA GmbH haftet ausdrücklich nicht für den Inhalt und die Vollständigkeit von E-Mails und den gegebenenfalls daraus entstehenden Schaden. Sollte trotz der bestehenden Viren-Schutzprogramme durch diese E-Mail ein Virus in Ihr System gelangen, so haftet die GISA GmbH - soweit gesetzlich zulässig - nicht für die hieraus entstehenden Schäden.

[Iustin Pop]

On Tue, Oct 30, 2012 at 07:39:35AM +0100, Marco Casavecchia M. wrote:
> What about the patch to specify the cpu model per instance?
> I'm still stuck to my personally patched 2.5.1 because I need different cpu models in my instances.
> In fact, I have nodes with different cpus (amd and intel) and instances that needs 32 and 64 bits cpus.

That patch modifies the configuration (adding the cpu_type key), so we
can't backport it to a stable release.

> Can you also remember me if the "no cache" and the "native io" patches are already into the 2.6 branch?

Hmm, I can't find these at all. Do you remember more details?

(but in any case, they would again change the configuration, so I expect
not)

thanks,
iustin

[Iustin Pop]

On Tue, Oct 30, 2012 at 08:50:07AM +0100, Lucas, Sascha wrote:
> Hi,
>
> From: Iustin Pop
> Date: Mon, 29. Oct 2012 17:26
>
> > We're preparing the 2.6.2 release, and I'd like your feedback to ensure
> > we don't miss any easy fixes that could go in it.
>
> I don't know if following is easy to fix. But here is my wish list for what I think it could be easy (unordered):

Thanks for the list!

>
> * KVM: default to cache=none,aio=native parameter

Sorry, more information about this? Was there a thread on it?

> * Issue 261: Possibility to set custom kvm parameters on an instance

Ack, that would be useful, but it would change the config, so it has to
go onto the next major release (2.7)

> * gnt-cluster verify: don't blame for non redundant instances in case of sharedfile...

Thanks, good point, I'll try to get this in; patch sent.

> * gnt-cluster modify --no-lvm-storage: validation bug? Failure: prerequisites not met for this operation: error type: wrong_input, error details: Parameter 'OP_CLUSTER_SET_PARAMS.vg_name' fails validation

Oh. I didn't know about this, and I can't find an issue. I filled issue
285 for it, thanks!

> * Issue 262: gnt-instance remove does not remove directory on shared file storage path

Thanks, this should be an easy fix; patch sent.

> * gnt-instance modify --disk add:...: does not activate disks (in preparation for hotplug)

Hmm… I suppose this is part of/before issue 263, right? I'll have to
take a look, not sure if it's easy or not.

> * Issue 225: kvm path is configurable only at build time

This would require a new hypervisor parameter, so can't go in 2.6.

> * similar to Issue 259: respect --prealloc-wipe-disks with grow-disk/modify --disk add

This has been implemented in master, but since it's not a trivial fix
we'll not backport it now. I will update the issue status though, thanks
for the reminder.

thanks!
iustin

[marcolinuz]

I have sent you the patches for both the modification on top of the 2.5.1 some time ago but you told me that someone else was already working on the same modifications.

The patch is here and I'm using it in my production environment from a while without any problem.

https://groups.google.com/forum/?fromgroups#!searchin/ganeti-devel/aio$20native/ganeti-devel/pgqdvnb1p_g/a3lwKd1lhJEJ

PS: Indeed the patch changes the configuration.. :(

--
By MCM.

«Un computer è come il dio del vecchio testamento: ha un casino di regole e nessuna pietà.»

[Iustin Pop]

On Tue, Oct 30, 2012 at 02:48:23PM +0100, marcolinuz wrote:
> I have sent you the patches for both the modification on top of the 2.5.1
> some time ago but you told me that someone else was already working on the
> same modifications.
>
> The patch is here and I'm using it in my production environment from a
> while without any problem.
>
> https://groups.google.com/forum/?fromgroups#!searchin/ganeti-devel/aio$20native/ganeti-devel/pgqdvnb1p_g/a3lwKd1lhJEJ
>
> PS: Indeed the patch changes the configuration.. :(

Yes. So the cpu type patch has went it, but not the aio/native/threads
one. I'll see about that (but again not for 2.6).

thanks,
iustin

[marcolinuz]

On Tue, Oct 30, 2012 at 3:06 PM, Iustin Pop <ius...@google.com> wrote:

On Tue, Oct 30, 2012 at 02:48:23PM +0100, marcolinuz wrote:
> I have sent you the patches for both the modification on top of the 2.5.1
> some time ago but you told me that someone else was already working on the
> same modifications.
>
> The patch is here and I'm using it in my production environment from a
> while without any problem.
>
> https://groups.google.com/forum/?fromgroups#!searchin/ganeti-devel/aio$20native/ganeti-devel/pgqdvnb1p_g/a3lwKd1lhJEJ
>
> PS: Indeed the patch changes the configuration.. :(

Yes. So the cpu type patch has went it, but not the aio/native/threads
one. I'll see about that (but again not for 2.6).

Okay,

I will wait for the first release that will have the aio/native/threads patch before attempting to upgrade my production environment.

 

--
By MCM.