[pve-devel] Kernel crash with 4.1 kernel

2015-09-09 Thread Stefan Priebe - Profihost AG 
Hi,

today i had a strange crash of ALL nodes in parallel while running the
4.1 kernel.

I was rebooting one node and all other nodes crashed with the following
output:
dlm: closing connection to node 5
BUG: unable to handle kernel NULL pointer dereference at 08e

As there were no more at the monitor nor in the netconsole i've no idea
where to search or look for this.

Anybody an idea?

Stefan
___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] disabling transparent hugepage by default like kernel 2.6.32 ?

2015-09-09 Thread Daniel Hunsaker 
What about setting CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS = N?  As I understand
it, "not set" means "use the default setting the kernel devs think is the
most sane for normal use", which in this case would be Y...

On Wed, Sep 9, 2015, 06:58 Alexandre DERUMIER  wrote:

> >>I could set it in kernel config, for example:
> >>
> >>CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y
> >>
> >>would that help?
>
> Well, this is like  "echo madvise >
> /sys/kernel/mm/transparent_hugepage/enabled"
>
> But I don't see how to set never.
>
> maybe
>
> # CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS is not set
> # CONFIG_TRANSPARENT_HUGEPAGE_MADVISE is not set
>
>
> But I don't known how to unset CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS in the
> new pve-kernel package makefile ?
>
>
> I have tried to add
>
> PVE_CONFIG_OPTS= \
> ..
> -d CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS \
> ..
>
> but it don't seem to work, I still have
> CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS=Y in .config
>
> Any idea ?
>
>
>
> - Mail original -
> De: "dietmar" 
> À: "aderumier" , "pve-devel" <
> pve-devel@pve.proxmox.com>
> Envoyé: Mercredi 9 Septembre 2015 06:33:05
> Objet: Re: [pve-devel] disabling transparent hugepage by default like
> kernel 2.6.32 ?
>
> > I think it's better to disable it by default.
> >
> > echo never > /sys/kernel/mm/transparent_hugepage/enabled.
> >
> > (I don't known what is the best way to disable it by default)
>
> I could set it in kernel config, for example:
>
> CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y
>
> would that help?
> ___
> pve-devel mailing list
> pve-devel@pve.proxmox.com
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
>
___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] Kernel crash with 4.1 kernel

2015-09-09 Thread Stefan Priebe - Profihost AG 

Am 09.09.2015 um 14:50 schrieb Martin Maurer:
>> Stefan Priebe - Profihost AG  hat am 9. September 2015
>> um 14:12 geschrieben:
>>
>>
>> Hi,
>>
>> today i had a strange crash of ALL nodes in parallel while running the
>> 4.1 kernel.
> 
> Not an answer to your question but we are already on kernel 4.2 (for Proxmox 
> VE
> 4beta2) - we will update the pvetest repository within the next 24 hours.

Thanks - but any reason? I mean 4.1 is a long term kernel greg kohan
will support until 2017.

Stefan

> Martin
> 
___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] [PAST pve-manager] use correct checked value for ha resource edit dialog

2015-09-09 Thread Thomas Lamprecht 

s/PAST/PATCH/
Sorry but shouldn't have an effect.

On 09/09/2015 03:21 PM, Thomas Lamprecht wrote:

Signed-off-by: Thomas Lamprecht 
---
  www/manager/ha/ResourceEdit.js | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/www/manager/ha/ResourceEdit.js b/www/manager/ha/ResourceEdit.js
index 962cdda..bc48c70 100644
--- a/www/manager/ha/ResourceEdit.js
+++ b/www/manager/ha/ResourceEdit.js
@@ -54,7 +54,7 @@ Ext.define('PVE.ha.VMResourceInputPanel', {
{
xtype: 'pvecheckbox',
name: 'enable',
-   checked: true,
+   checked: me.enable,
uncheckedValue: 0,
fieldLabel: gettext('enable')
}



___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

[pve-devel] [PAST pve-manager] use correct checked value for ha resource edit dialog

2015-09-09 Thread Thomas Lamprecht 
Signed-off-by: Thomas Lamprecht 
---
 www/manager/ha/ResourceEdit.js | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/www/manager/ha/ResourceEdit.js b/www/manager/ha/ResourceEdit.js
index 962cdda..bc48c70 100644
--- a/www/manager/ha/ResourceEdit.js
+++ b/www/manager/ha/ResourceEdit.js
@@ -54,7 +54,7 @@ Ext.define('PVE.ha.VMResourceInputPanel', {
{
xtype: 'pvecheckbox',
name: 'enable',
-   checked: true,
+   checked: me.enable,
uncheckedValue: 0,
fieldLabel: gettext('enable')
}
-- 
2.1.4

___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] disabling transparent hugepage by default like kernel 2.6.32 ?

2015-09-09 Thread Andreas Steinel 
Hi,

In my benchmarks, THP increase the memory transaction speed by 5-10% in the
KVM guest if THP is activated on the host side, not in the VM side. I know
that THP for databases is bad, but we are not using THP in the guest where
the database lives. Can someone provide some benchmarks for that case or
further information? Any suggestions?

On Wed, Sep 9, 2015 at 3:45 PM, Daniel Hunsaker 
wrote:

> What about setting CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS = N?  As I
> understand it, "not set" means "use the default setting the kernel devs
> think is the most sane for normal use", which in this case would be Y...
>
> On Wed, Sep 9, 2015, 06:58 Alexandre DERUMIER  wrote:
>
>> >>I could set it in kernel config, for example:
>> >>
>> >>CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y
>> >>
>> >>would that help?
>>
>> Well, this is like  "echo madvise >
>> /sys/kernel/mm/transparent_hugepage/enabled"
>>
>> But I don't see how to set never.
>>
>> maybe
>>
>> # CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS is not set
>> # CONFIG_TRANSPARENT_HUGEPAGE_MADVISE is not set
>>
>>
>> But I don't known how to unset CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS in the
>> new pve-kernel package makefile ?
>>
>>
>> I have tried to add
>>
>> PVE_CONFIG_OPTS= \
>> ..
>> -d CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS \
>> ..
>>
>> but it don't seem to work, I still have
>> CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS=Y in .config
>>
>> Any idea ?
>>
>>
>>
>> - Mail original -
>> De: "dietmar" 
>> À: "aderumier" , "pve-devel" <
>> pve-devel@pve.proxmox.com>
>> Envoyé: Mercredi 9 Septembre 2015 06:33:05
>> Objet: Re: [pve-devel] disabling transparent hugepage by default like
>> kernel 2.6.32 ?
>>
>> > I think it's better to disable it by default.
>> >
>> > echo never > /sys/kernel/mm/transparent_hugepage/enabled.
>> >
>> > (I don't known what is the best way to disable it by default)
>>
>> I could set it in kernel config, for example:
>>
>> CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y
>>
>> would that help?
>> ___
>> pve-devel mailing list
>> pve-devel@pve.proxmox.com
>> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
>>
>
> ___
> pve-devel mailing list
> pve-devel@pve.proxmox.com
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
>
>
___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] disabling transparent hugepage by default like kernel 2.6.32 ?

2015-09-09 Thread Alexandre DERUMIER 
>>I could set it in kernel config, for example: 
>>
>>CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y 
>>
>>would that help? 

Well, this is like  "echo madvise > /sys/kernel/mm/transparent_hugepage/enabled"

But I don't see how to set never.

maybe 

# CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS is not set
# CONFIG_TRANSPARENT_HUGEPAGE_MADVISE is not set


But I don't known how to unset CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS in the new 
pve-kernel package makefile ?


I have tried to add

PVE_CONFIG_OPTS= \
..
-d CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS \
..

but it don't seem to work, I still have CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS=Y in 
.config

Any idea ?



- Mail original -
De: "dietmar" 
À: "aderumier" , "pve-devel" 
Envoyé: Mercredi 9 Septembre 2015 06:33:05
Objet: Re: [pve-devel] disabling transparent hugepage by default like kernel 
2.6.32 ?

> I think it's better to disable it by default. 
> 
> echo never > /sys/kernel/mm/transparent_hugepage/enabled. 
> 
> (I don't known what is the best way to disable it by default) 

I could set it in kernel config, for example: 

CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y 

would that help?
___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] Kernel crash with 4.1 kernel

2015-09-09 Thread Alexandre DERUMIER 
>>Thanks - but any reason? I mean 4.1 is a long term kernel greg kohan
>>will support until 2017.

As proxmox use now ubuntu kernel (for lxc features), and current ubuntu is not 
lts,

ubuntu don't care to run on LTS kernel.

I think we need wait for the next lts ubuntu to have a lts kernel.



- Mail original -
De: "Stefan Priebe" 
À: "Martin Maurer" , "pve-devel" 
Envoyé: Mercredi 9 Septembre 2015 14:51:53
Objet: Re: [pve-devel] Kernel crash with 4.1 kernel

Am 09.09.2015 um 14:50 schrieb Martin Maurer: 
>> Stefan Priebe - Profihost AG  hat am 9. September 
>> 2015 
>> um 14:12 geschrieben: 
>> 
>> 
>> Hi, 
>> 
>> today i had a strange crash of ALL nodes in parallel while running the 
>> 4.1 kernel. 
> 
> Not an answer to your question but we are already on kernel 4.2 (for Proxmox 
> VE 
> 4beta2) - we will update the pvetest repository within the next 24 hours. 

Thanks - but any reason? I mean 4.1 is a long term kernel greg kohan 
will support until 2017. 

Stefan 

> Martin 
> 
___ 
pve-devel mailing list 
pve-devel@pve.proxmox.com 
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel 
___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

[pve-devel] [PATCH pve-container] Don't die in vm_shutdown when we have no errors

2015-09-09 Thread Emmanuel Kasper 
This fixes bugzilla entry: https://bugzilla.proxmox.com/show_bug.cgi?id=707
---
 src/PVE/API2/LXC/Status.pm | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/PVE/API2/LXC/Status.pm b/src/PVE/API2/LXC/Status.pm
index 761a3c8..5e5e116 100644
--- a/src/PVE/API2/LXC/Status.pm
+++ b/src/PVE/API2/LXC/Status.pm
@@ -316,7 +316,7 @@ __PACKAGE__->register_method({
run_command($cmd);
}
 
-   die $err if !$err;
+   die $err if $err;
 
return;
};
-- 
2.1.4


___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

[pve-devel] DRBD pending action

2015-09-09 Thread Gilberto Nunes 
Hi...

I am testing PVE 4 with drbd and I remove a volume but is not delete any
way from the DRBD layer. Instead I get this message:

drbdmanage list-volumes
++

| Name  | Vol ID |  Size | Minor |
|   State |
++

| vm-100-disk-1 |  0 | 32768 |10 |
| pending actions: remove |

Is there any other command like commit or something???

Thanks
-- 

Gilberto Ferreira
+55 (47) 9676-7530
Skype: gilberto.nunes36
___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] disabling transparent hugepage by default like kernel 2.6.32 ?

2015-09-09 Thread Alexandre DERUMIER 
>>Do you really want 'never'? I thought 'madvise' would be better, because we 
>>can
>>enable it
>>for some application (KVM)?

AFAIK, even jemalloc for example are not perfect with madvise , they are still 
latency impact.

They are a lot of bug report with transparent hudgepage with software like 
redis,varnish,mongodb,mysql,hadoop,oracle, event with madvise.


So I think it's better to setup to safe default value, "never".

User can still enable it after with "echo madvise|always > 
/sys/kernel/mm/transparent_hugepage/enabled"


- Mail original -
De: "dietmar" 
À: "aderumier" 
Cc: "pve-devel" 
Envoyé: Mercredi 9 Septembre 2015 16:34:32
Objet: Re: [pve-devel] disabling transparent hugepage by default like kernel 
2.6.32 ?

> >>I could set it in kernel config, for example: 
> >> 
> >>CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y 
> >> 
> >>would that help? 
> 
> Well, this is like "echo madvise > 
> /sys/kernel/mm/transparent_hugepage/enabled" 
> 
> But I don't see how to set never. 

Do you really want 'never'? I thought 'madvise' would be better, because we can 
enable it 
for some application (KVM)? 
___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] DRBD pending action

2015-09-09 Thread Dietmar Maurer 

> I am testing PVE 4 with drbd and I remove a volume but is not delete any

AFAIK this is already fixed with newest drbdmanage packages.

___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] DRBD pending action

2015-09-09 Thread Gilberto Nunes 
Do you mean, from Proxmox repos?? I tried but no new version available...

2015-09-09 11:59 GMT-03:00 Dietmar Maurer :

>
> > I am testing PVE 4 with drbd and I remove a volume but is not delete any
>
> AFAIK this is already fixed with newest drbdmanage packages.
>
>


-- 

Gilberto Ferreira
+55 (47) 9676-7530
Skype: gilberto.nunes36
___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] disabling transparent hugepage by default like kernel 2.6.32 ?

2015-09-09 Thread Alexandre DERUMIER 
>>In my benchmarks, THP increase the memory transaction speed by 5-10% in the 
>>KVM guest if THP is activated on the host side, not in the VM side. I know 
>>that THP for databases is bad, but we are not using THP in the guest >>where 
>>the database lives. Can someone provide some benchmarks for that case or 
>>further information? Any suggestions?

Oh, good to known.

How much memory do you use it your vm ? what is the database software ?

- Mail original -
De: "Andreas Steinel" 
À: "Daniel Hunsaker" 
Cc: "aderumier" , "dietmar" , 
"pve-devel" 
Envoyé: Mercredi 9 Septembre 2015 16:32:23
Objet: Re: [pve-devel] disabling transparent hugepage by default like kernel 
2.6.32 ?

Hi, 
In my benchmarks, THP increase the memory transaction speed by 5-10% in the KVM 
guest if THP is activated on the host side, not in the VM side. I know that THP 
for databases is bad, but we are not using THP in the guest where the database 
lives. Can someone provide some benchmarks for that case or further 
information? Any suggestions? 

On Wed, Sep 9, 2015 at 3:45 PM, Daniel Hunsaker < danhunsa...@gmail.com > 
wrote: 




What about setting CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS = N? As I understand it, 
"not set" means "use the default setting the kernel devs think is the most sane 
for normal use", which in this case would be Y... 

On Wed, Sep 9, 2015, 06:58 Alexandre DERUMIER < aderum...@odiso.com > wrote: 

BQ_BEGIN
>>I could set it in kernel config, for example: 
>> 
>>CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y 
>> 
>>would that help? 

Well, this is like "echo madvise > /sys/kernel/mm/transparent_hugepage/enabled" 

But I don't see how to set never. 

maybe 

# CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS is not set 
# CONFIG_TRANSPARENT_HUGEPAGE_MADVISE is not set 


But I don't known how to unset CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS in the new 
pve-kernel package makefile ? 


I have tried to add 

PVE_CONFIG_OPTS= \ 
.. 
-d CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS \ 
.. 

but it don't seem to work, I still have CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS=Y in 
.config 

Any idea ? 



- Mail original - 
De: "dietmar" < diet...@proxmox.com > 
À: "aderumier" < aderum...@odiso.com >, "pve-devel" < pve-devel@pve.proxmox.com 
> 
Envoyé: Mercredi 9 Septembre 2015 06:33:05 
Objet: Re: [pve-devel] disabling transparent hugepage by default like kernel 
2.6.32 ? 

> I think it's better to disable it by default. 
> 
> echo never > /sys/kernel/mm/transparent_hugepage/enabled. 
> 
> (I don't known what is the best way to disable it by default) 

I could set it in kernel config, for example: 

CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y 

would that help? 
___ 
pve-devel mailing list 
pve-devel@pve.proxmox.com 
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel 




___ 
pve-devel mailing list 
pve-devel@pve.proxmox.com 
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel 


BQ_END

___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] [PAST pve-manager] use correct checked value for ha resource edit dialog

2015-09-09 Thread Dietmar Maurer 
that looks strange to me - is that really required?
The current GUI works for me, so how can I reproduce the bug?

> On September 9, 2015 at 3:21 PM Thomas Lamprecht 
> wrote:
> 
> 
> Signed-off-by: Thomas Lamprecht 
> ---
>  www/manager/ha/ResourceEdit.js | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/www/manager/ha/ResourceEdit.js b/www/manager/ha/ResourceEdit.js
> index 962cdda..bc48c70 100644
> --- a/www/manager/ha/ResourceEdit.js
> +++ b/www/manager/ha/ResourceEdit.js
> @@ -54,7 +54,7 @@ Ext.define('PVE.ha.VMResourceInputPanel', {
>   {
>   xtype: 'pvecheckbox',
>   name: 'enable',
> - checked: true,
> + checked: me.enable,
>   uncheckedValue: 0,
>   fieldLabel: gettext('enable')
>   }
> -- 
> 2.1.4
> 
> ___
> pve-devel mailing list
> pve-devel@pve.proxmox.com
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
> 

___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] [PATCH pve-container] Don't die in vm_shutdown when we have no errors

2015-09-09 Thread Dietmar Maurer 
> 
> This fixes bugzilla entry: https://bugzilla.proxmox.com/show_bug.cgi?id=707
>

applied, thanks.

___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] Kernel crash with 4.1 kernel

2015-09-09 Thread Dietmar Maurer 
> I was rebooting one node and all other nodes crashed with the following
> output:
> dlm: closing connection to node 5

May I ask why do you use dlm at all?

___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] [PAST pve-manager] use correct checked value for ha resource edit dialog

2015-09-09 Thread Thomas Lamprecht 


Didn't replied to the list, sorry.

 Weitergeleitete Nachricht 
Betreff: 	Re: [pve-devel] [PAST pve-manager] use correct checked value 
for ha resource edit dialog

Datum:  Wed, 9 Sep 2015 18:10:50 +0200
Von:Thomas Lamprecht 
An: Dietmar Maurer 



Reproduction:
Disable an service and then try to edit it. You cannot enable it
directly as the checkbox is already checked.
So you need to save it as disabled (although it is already disabled)
reopen the edit box and only then you can enable it.

Am 09.09.2015 um 17:31 schrieb Dietmar Maurer:

that looks strange to me - is that really required?
The current GUI works for me, so how can I reproduce the bug?


On September 9, 2015 at 3:21 PM Thomas Lamprecht 
wrote:


Signed-off-by: Thomas Lamprecht 
---
  www/manager/ha/ResourceEdit.js | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/www/manager/ha/ResourceEdit.js b/www/manager/ha/ResourceEdit.js
index 962cdda..bc48c70 100644
--- a/www/manager/ha/ResourceEdit.js
+++ b/www/manager/ha/ResourceEdit.js
@@ -54,7 +54,7 @@ Ext.define('PVE.ha.VMResourceInputPanel', {
{
xtype: 'pvecheckbox',
name: 'enable',
-   checked: true,
+   checked: me.enable,
uncheckedValue: 0,
fieldLabel: gettext('enable')
}
--
2.1.4

___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel





___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] disabling transparent hugepage by default like kernel 2.6.32 ?

2015-09-09 Thread Alexandre DERUMIER 
here an interesting recent lklm discussion about THP

http://www.spinics.net/lists/linux-mm/msg84776.html

- Mail original -
De: "aderumier" 
À: "Andreas Steinel" 
Cc: "pve-devel" 
Envoyé: Mercredi 9 Septembre 2015 17:43:58
Objet: Re: [pve-devel] disabling transparent hugepage by default like kernel 
2.6.32 ?

>>In my benchmarks, THP increase the memory transaction speed by 5-10% in the 
>>KVM guest if THP is activated on the host side, not in the VM side. I know 
>>that THP for databases is bad, but we are not using THP in the guest >>where 
>>the database lives. Can someone provide some benchmarks for that case or 
>>further information? Any suggestions? 

Oh, good to known. 

How much memory do you use it your vm ? what is the database software ? 

- Mail original - 
De: "Andreas Steinel"  
À: "Daniel Hunsaker"  
Cc: "aderumier" , "dietmar" , 
"pve-devel"  
Envoyé: Mercredi 9 Septembre 2015 16:32:23 
Objet: Re: [pve-devel] disabling transparent hugepage by default like kernel 
2.6.32 ? 

Hi, 
In my benchmarks, THP increase the memory transaction speed by 5-10% in the KVM 
guest if THP is activated on the host side, not in the VM side. I know that THP 
for databases is bad, but we are not using THP in the guest where the database 
lives. Can someone provide some benchmarks for that case or further 
information? Any suggestions? 

On Wed, Sep 9, 2015 at 3:45 PM, Daniel Hunsaker < danhunsa...@gmail.com > 
wrote: 




What about setting CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS = N? As I understand it, 
"not set" means "use the default setting the kernel devs think is the most sane 
for normal use", which in this case would be Y... 

On Wed, Sep 9, 2015, 06:58 Alexandre DERUMIER < aderum...@odiso.com > wrote: 

 
>>I could set it in kernel config, for example: 
>> 
>>CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y 
>> 
>>would that help? 

Well, this is like "echo madvise > /sys/kernel/mm/transparent_hugepage/enabled" 

But I don't see how to set never. 

maybe 

# CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS is not set 
# CONFIG_TRANSPARENT_HUGEPAGE_MADVISE is not set 


But I don't known how to unset CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS in the new 
pve-kernel package makefile ? 


I have tried to add 

PVE_CONFIG_OPTS= \ 
.. 
-d CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS \ 
.. 

but it don't seem to work, I still have CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS=Y in 
.config 

Any idea ? 



- Mail original - 
De: "dietmar" < diet...@proxmox.com > 
À: "aderumier" < aderum...@odiso.com >, "pve-devel" < pve-devel@pve.proxmox.com 
> 
Envoyé: Mercredi 9 Septembre 2015 06:33:05 
Objet: Re: [pve-devel] disabling transparent hugepage by default like kernel 
2.6.32 ? 

> I think it's better to disable it by default. 
> 
> echo never > /sys/kernel/mm/transparent_hugepage/enabled. 
> 
> (I don't known what is the best way to disable it by default) 

I could set it in kernel config, for example: 

CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y 

would that help? 
___ 
pve-devel mailing list 
pve-devel@pve.proxmox.com 
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel 




___ 
pve-devel mailing list 
pve-devel@pve.proxmox.com 
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel 


 

___ 
pve-devel mailing list 
pve-devel@pve.proxmox.com 
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel 
___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] Kernel crash with 4.1 kernel

2015-09-09 Thread Dietmar Maurer 
> >>Thanks - but any reason? I mean 4.1 is a long term kernel greg kohan
> >>will support until 2017.
> 
> As proxmox use now ubuntu kernel (for lxc features), and current ubuntu is not
> lts,
> 
> ubuntu don't care to run on LTS kernel.
> 
> I think we need wait for the next lts ubuntu to have a lts kernel.

Yes, my current plan is to use the latest Ubuntu kernel, until 
the next Ubuntu LTS (16.04). After that we stick with 
the 16.04 LTS kernel.

___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

[pve-devel] pve-manager : add influxdb stats plugin V2

2015-09-09 Thread Alexandre Derumier 
changelog : rename graphiteserver option to generic server option


I'll add documentation in wiki for both graphite && influxdb stats plugin,
when this commit will be apply.


___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

[pve-devel] [PATCH] add influxdb stats plugin V2

2015-09-09 Thread Alexandre Derumier 
/etc/pve/status.cfg
---
influxdb:
  server influxdb3.odiso.net
  port 8089

This require influxdb >= 0.9 with udp enabled

influxdb.conf
-

[[udp]]
  enabled = true
  bind-address = "0.0.0.0:8089"
  database = "proxmox"
  batch-size = 1000
  batch-timeout = "1s"

Signed-off-by: Alexandre Derumier 
---
 PVE/Service/pvestatd.pm |   2 +
 PVE/Status/Graphite.pm  |  12 ++---
 PVE/Status/InfluxDB.pm  | 136 
 PVE/Status/Makefile |   1 +
 4 files changed, 145 insertions(+), 6 deletions(-)
 create mode 100644 PVE/Status/InfluxDB.pm

diff --git a/PVE/Service/pvestatd.pm b/PVE/Service/pvestatd.pm
index 486f24c..90add56 100755
--- a/PVE/Service/pvestatd.pm
+++ b/PVE/Service/pvestatd.pm
@@ -21,8 +21,10 @@ use PVE::AutoBalloon;
 
 use PVE::Status::Plugin;
 use PVE::Status::Graphite;
+use PVE::Status::InfluxDB;
 
 PVE::Status::Graphite->register();
+PVE::Status::InfluxDB->register();
 PVE::Status::Plugin->init();
 
 use base qw(PVE::Daemon);
diff --git a/PVE/Status/Graphite.pm b/PVE/Status/Graphite.pm
index 5f02754..b744968 100644
--- a/PVE/Status/Graphite.pm
+++ b/PVE/Status/Graphite.pm
@@ -6,7 +6,7 @@ use PVE::Status::Plugin;
 
 # example config (/etc/pve/status.cfg)
 #graphite:
-#  graphiteserver test
+#  server test
 #  port 2003
 #  path proxmox.mycluster
 #  disable 0
@@ -20,13 +20,13 @@ sub type {
 
 sub properties {
 return {
-   graphiteserver => {
+   server => {
type => 'string', format => 'dns-name',
-   description => "External graphite statistic server dns name",
+   description => "server dns name",
},
port => {
 type => 'integer',
-   description => "graphite server port",
+   description => "network port",
},
path => {
 type => 'string', format => 'graphite-path',
@@ -37,7 +37,7 @@ sub properties {
 
 sub options {
 return {
-   graphiteserver => {},
+   server => {},
port => { optional => 1 },
path => { optional => 1 },
disable => { optional => 1 },
@@ -72,7 +72,7 @@ sub update_storage_status {
 sub write_graphite_hash {
 my ($plugin_config, $d, $ctime, $object) = @_;
 
-my $host = $plugin_config->{graphiteserver};
+my $host = $plugin_config->{server};
 my $port = $plugin_config->{port} ? $plugin_config->{port} : 2003;
 my $path = $plugin_config->{path} ? $plugin_config->{path} : 'proxmox';
 
diff --git a/PVE/Status/InfluxDB.pm b/PVE/Status/InfluxDB.pm
new file mode 100644
index 000..507f9b2
--- /dev/null
+++ b/PVE/Status/InfluxDB.pm
@@ -0,0 +1,136 @@
+package PVE::Status::InfluxDB;
+
+use strict;
+use warnings;
+use PVE::Status::Plugin;
+use Data::Dumper;
+use PVE::SafeSyslog;
+
+# example config (/etc/pve/status.cfg)
+#influxdb:
+#  server test
+#  port 8089
+#  disable 0
+#
+
+use base('PVE::Status::Plugin');
+
+sub type {
+return 'influxdb';
+}
+
+sub options {
+return {
+   server => {},
+   port => {},
+   disable => { optional => 1 },
+   };
+}
+
+# Plugin implementation
+sub update_node_status {
+my ($class, $plugin_config, $node, $data, $ctime) = @_;
+
+$ctime *= 10;
+
+write_influxdb_hash($plugin_config, $data, $ctime, 
"object=nodes,host=$node");
+
+}
+
+sub update_qemu_status {
+my ($class, $plugin_config, $vmid, $data, $ctime) = @_;
+
+$ctime *= 10;
+
+my $object = "object=qemu,vmid=$vmid";
+if($data->{name} && $data->{name} ne '') {
+   $object .= ",host=$data->{name}";
+}
+$object =~ s/\s/\\ /g;
+write_influxdb_hash($plugin_config, $data, $ctime, $object);
+}
+
+sub update_lxc_status {
+my ($class, $plugin_config, $vmid, $data, $ctime) = @_;
+
+$ctime *= 10;
+
+my $object = "object=lxc,vmid=$vmid";
+if($data->{name} && $data->{name} ne '') {
+   $object .= ",host=$data->{name}";
+}
+$object =~ s/\s/\\ /g;
+
+write_influxdb_hash($plugin_config, $data, $ctime, $object);
+}
+
+sub update_storage_status {
+my ($class, $plugin_config, $nodename, $storeid, $data, $ctime) = @_;
+
+$ctime *= 10;
+
+my $object = "object=storages,nodename=$nodename,host=$storeid";
+if($data->{type} && $data->{type} ne '') {
+   $object .= ",type=$data->{type}";
+}
+$object =~ s/\s/\\ /g;
+
+write_influxdb_hash($plugin_config, $data, $ctime, $object);
+}
+
+sub write_influxdb_hash {
+my ($plugin_config, $d, $ctime, $tags) = @_;
+
+my $payload = {};
+
+build_influxdb_payload($payload, $d, $ctime, $tags);
+
+my $host = $plugin_config->{server};
+my $port = $plugin_config->{port};
+
+my $socket = IO::Socket::IP->new(
+PeerAddr=> $host,
+PeerPort=> $port,
+Proto   => 'udp',
+);
+
+$socket->send($payload->{string});
+$socket->close() if $socket;
+
+}
+
+sub build_influxdb_payload {
+

Re: [pve-devel] DRBD pending action

2015-09-09 Thread Gilberto Nunes 
Nice... I'll wait for it...
Thanks

2015-09-09 14:46 GMT-03:00 Dietmar Maurer :

>
> > Do you mean, from Proxmox repos?? I tried but no new version available...
>
> we will upload new packages tomorrow ...
>
> >
> > 2015-09-09 11:59 GMT-03:00 Dietmar Maurer :
> >
> > >
> > > > I am testing PVE 4 with drbd and I remove a volume but is not delete
> any
> > >
> > > AFAIK this is already fixed with newest drbdmanage packages.
> > >
> > >
> >
> >
> > --
> >
> > Gilberto Ferreira
> > +55 (47) 9676-7530
> > Skype: gilberto.nunes36
>
>


-- 

Gilberto Ferreira
+55 (47) 9676-7530
Skype: gilberto.nunes36
___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] [PAST pve-manager] use correct checked value for ha resource edit dialog

2015-09-09 Thread Dietmar Maurer 
> Reproduction:
> Disable an service and then try to edit it. You cannot enable it 
> directly as the checkbox is already checked.
> So you need to save it as disabled (although it is already disabled) 
> reopen the edit box and only then you can enable it.

I cannot reproduce that.

> 
> Am 09.09.2015 um 17:31 schrieb Dietmar Maurer:
> > that looks strange to me - is that really required?
> > The current GUI works for me, so how can I reproduce the bug?

___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] DRBD pending action

2015-09-09 Thread Dietmar Maurer 

> Do you mean, from Proxmox repos?? I tried but no new version available...

we will upload new packages tomorrow ...

> 
> 2015-09-09 11:59 GMT-03:00 Dietmar Maurer :
> 
> >
> > > I am testing PVE 4 with drbd and I remove a volume but is not delete any
> >
> > AFAIK this is already fixed with newest drbdmanage packages.
> >
> >
> 
> 
> -- 
> 
> Gilberto Ferreira
> +55 (47) 9676-7530
> Skype: gilberto.nunes36

___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] Kernel crash with 4.1 kernel

2015-09-09 Thread Dietmar Maurer 
> this is PVE 3.4 on wheezy with the current 4.1 kernel. I've not changed 
> explicitly anything regarding dlm. Isn't it the default on 3.4?

Well, it is confusing if you talk about kernel 4.1 without mentioning
that you use Proxmox 3.4.

AFAIK nobody else use that combination, so nobody else tests that.

___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] Kernel crash with 4.1 kernel

2015-09-09 Thread Stefan Priebe 


Am 09.09.2015 um 19:59 schrieb Dietmar Maurer:

this is PVE 3.4 on wheezy with the current 4.1 kernel. I've not changed
explicitly anything regarding dlm. Isn't it the default on 3.4?


Well, it is confusing if you talk about kernel 4.1 without mentioning
that you use Proxmox 3.4.

AFAIK nobody else use that combination, so nobody else tests that.



Yes sorry. I ignored PVE itself as i thought this must be a kernel only 
problem. But you're right as nobody uses DLM in that case ...


Greets,
Stefan
___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] [PAST pve-manager] use correct checked value for ha resource edit dialog

2015-09-09 Thread Thomas Lamprecht 



Am 09.09.2015 um 19:42 schrieb Dietmar Maurer:

Reproduction:
Disable an service and then try to edit it. You cannot enable it
directly as the checkbox is already checked.
So you need to save it as disabled (although it is already disabled)
reopen the edit box and only then you can enable it.

I cannot reproduce that.
Ok, better reproduction: disable an HA service resource via web 
interface and then try to disable another one, then the check-box state 
doesn't represent the current state.





Am 09.09.2015 um 17:31 schrieb Dietmar Maurer:

that looks strange to me - is that really required?
The current GUI works for me, so how can I reproduce the bug?



___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] disabling transparent hugepage by default like kernel 2.6.32 ?

2015-09-09 Thread Andreas Steinel 
I tried with a simple 128 MB Jessie Box and 'mbw' program and got these results:

mbw -n 100 50 | grep AVG > hthp_1_gthp_1



HostTransparentHugePages_(1|0)_GuestTransparentHugePages_(1|0)


Test hthp_0_gthp_0
AVG Method: MEMCPY  Elapsed: 0.01728MiB: 50.0   Copy:
2893.123 MiB/s
AVG Method: DUMBElapsed: 0.01770MiB: 50.0   Copy:
2824.581 MiB/s
AVG Method: MCBLOCK Elapsed: 0.0MiB: 50.0   Copy:
4499.575 MiB/s
Test hthp_0_gthp_1
AVG Method: MEMCPY  Elapsed: 0.01927MiB: 50.0   Copy:
2594.132 MiB/s
AVG Method: DUMBElapsed: 0.01869MiB: 50.0   Copy:
2674.970 MiB/s
AVG Method: MCBLOCK Elapsed: 0.01157MiB: 50.0   Copy:
4322.956 MiB/s
Test hthp_1_gthp_0
AVG Method: MEMCPY  Elapsed: 0.01645MiB: 50.0   Copy:
3039.115 MiB/s
AVG Method: DUMBElapsed: 0.01572MiB: 50.0   Copy:
3180.148 MiB/s
AVG Method: MCBLOCK Elapsed: 0.00987MiB: 50.0   Copy:
5066.072 MiB/s
Test hthp_1_gthp_1
AVG Method: MEMCPY  Elapsed: 0.02151MiB: 50.0   Copy:
2324.526 MiB/s
AVG Method: DUMBElapsed: 0.01553MiB: 50.0   Copy:
3220.203 MiB/s
AVG Method: MCBLOCK Elapsed: 0.00998MiB: 50.0   Copy:
5009.729 MiB/s

On Wed, Sep 9, 2015 at 5:43 PM, Alexandre DERUMIER  wrote:
>>>In my benchmarks, THP increase the memory transaction speed by 5-10% in the 
>>>KVM guest if THP is activated on the host side, not in the VM side. I know 
>>>that THP for databases is bad, but we are not using THP in the guest >>where 
>>>the database lives. Can someone provide some benchmarks for that case or 
>>>further information? Any suggestions?
>
> Oh, good to known.
>
> How much memory do you use it your vm ? what is the database software ?
>
> - Mail original -
> De: "Andreas Steinel" 
> À: "Daniel Hunsaker" 
> Cc: "aderumier" , "dietmar" , 
> "pve-devel" 
> Envoyé: Mercredi 9 Septembre 2015 16:32:23
> Objet: Re: [pve-devel] disabling transparent hugepage by default like kernel 
> 2.6.32 ?
>
> Hi,
> In my benchmarks, THP increase the memory transaction speed by 5-10% in the 
> KVM guest if THP is activated on the host side, not in the VM side. I know 
> that THP for databases is bad, but we are not using THP in the guest where 
> the database lives. Can someone provide some benchmarks for that case or 
> further information? Any suggestions?
>
> On Wed, Sep 9, 2015 at 3:45 PM, Daniel Hunsaker < danhunsa...@gmail.com > 
> wrote:
>
>
>
>
> What about setting CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS = N? As I understand 
> it, "not set" means "use the default setting the kernel devs think is the 
> most sane for normal use", which in this case would be Y...
>
> On Wed, Sep 9, 2015, 06:58 Alexandre DERUMIER < aderum...@odiso.com > wrote:
>
> BQ_BEGIN
>>>I could set it in kernel config, for example:
>>>
>>>CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y
>>>
>>>would that help?
>
> Well, this is like "echo madvise > 
> /sys/kernel/mm/transparent_hugepage/enabled"
>
> But I don't see how to set never.
>
> maybe
>
> # CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS is not set
> # CONFIG_TRANSPARENT_HUGEPAGE_MADVISE is not set
>
>
> But I don't known how to unset CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS in the new 
> pve-kernel package makefile ?
>
>
> I have tried to add
>
> PVE_CONFIG_OPTS= \
> ..
> -d CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS \
> ..
>
> but it don't seem to work, I still have CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS=Y 
> in .config
>
> Any idea ?
>
>
>
> - Mail original -
> De: "dietmar" < diet...@proxmox.com >
> À: "aderumier" < aderum...@odiso.com >, "pve-devel" < 
> pve-devel@pve.proxmox.com >
> Envoyé: Mercredi 9 Septembre 2015 06:33:05
> Objet: Re: [pve-devel] disabling transparent hugepage by default like kernel 
> 2.6.32 ?
>
>> I think it's better to disable it by default.
>>
>> echo never > /sys/kernel/mm/transparent_hugepage/enabled.
>>
>> (I don't known what is the best way to disable it by default)
>
> I could set it in kernel config, for example:
>
> CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y
>
> would that help?
> ___
> pve-devel mailing list
> pve-devel@pve.proxmox.com
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
>
>
>
>
> ___
> pve-devel mailing list
> pve-devel@pve.proxmox.com
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
>
>
> BQ_END
>
___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] disabling transparent hugepage by default like kernel 2.6.32 ?

2015-09-09 Thread Alexandre DERUMIER 
Do you have tested it with a real workload and big memory ? (100-200GB for 
example).

I'm curious to see what happen with a lot of alloc/desalloc and fragmentation.


- Mail original -
De: "Andreas Steinel" 
À: "aderumier" 
Cc: "Daniel Hunsaker" , "dietmar" , 
"pve-devel" 
Envoyé: Mercredi 9 Septembre 2015 18:41:37
Objet: Re: [pve-devel] disabling transparent hugepage by default like kernel 
2.6.32 ?

I tried with a simple 128 MB Jessie Box and 'mbw' program and got these 
results: 

mbw -n 100 50 | grep AVG > hthp_1_gthp_1 



HostTransparentHugePages_(1|0)_GuestTransparentHugePages_(1|0) 


Test hthp_0_gthp_0 
AVG Method: MEMCPY Elapsed: 0.01728 MiB: 50.0 Copy: 
2893.123 MiB/s 
AVG Method: DUMB Elapsed: 0.01770 MiB: 50.0 Copy: 
2824.581 MiB/s 
AVG Method: MCBLOCK Elapsed: 0.0 MiB: 50.0 Copy: 
4499.575 MiB/s 
Test hthp_0_gthp_1 
AVG Method: MEMCPY Elapsed: 0.01927 MiB: 50.0 Copy: 
2594.132 MiB/s 
AVG Method: DUMB Elapsed: 0.01869 MiB: 50.0 Copy: 
2674.970 MiB/s 
AVG Method: MCBLOCK Elapsed: 0.01157 MiB: 50.0 Copy: 
4322.956 MiB/s 
Test hthp_1_gthp_0 
AVG Method: MEMCPY Elapsed: 0.01645 MiB: 50.0 Copy: 
3039.115 MiB/s 
AVG Method: DUMB Elapsed: 0.01572 MiB: 50.0 Copy: 
3180.148 MiB/s 
AVG Method: MCBLOCK Elapsed: 0.00987 MiB: 50.0 Copy: 
5066.072 MiB/s 
Test hthp_1_gthp_1 
AVG Method: MEMCPY Elapsed: 0.02151 MiB: 50.0 Copy: 
2324.526 MiB/s 
AVG Method: DUMB Elapsed: 0.01553 MiB: 50.0 Copy: 
3220.203 MiB/s 
AVG Method: MCBLOCK Elapsed: 0.00998 MiB: 50.0 Copy: 
5009.729 MiB/s 

On Wed, Sep 9, 2015 at 5:43 PM, Alexandre DERUMIER  wrote: 
>>>In my benchmarks, THP increase the memory transaction speed by 5-10% in the 
>>>KVM guest if THP is activated on the host side, not in the VM side. I know 
>>>that THP for databases is bad, but we are not using THP in the guest >>where 
>>>the database lives. Can someone provide some benchmarks for that case or 
>>>further information? Any suggestions? 
> 
> Oh, good to known. 
> 
> How much memory do you use it your vm ? what is the database software ? 
> 
> - Mail original - 
> De: "Andreas Steinel"  
> À: "Daniel Hunsaker"  
> Cc: "aderumier" , "dietmar" , 
> "pve-devel"  
> Envoyé: Mercredi 9 Septembre 2015 16:32:23 
> Objet: Re: [pve-devel] disabling transparent hugepage by default like kernel 
> 2.6.32 ? 
> 
> Hi, 
> In my benchmarks, THP increase the memory transaction speed by 5-10% in the 
> KVM guest if THP is activated on the host side, not in the VM side. I know 
> that THP for databases is bad, but we are not using THP in the guest where 
> the database lives. Can someone provide some benchmarks for that case or 
> further information? Any suggestions? 
> 
> On Wed, Sep 9, 2015 at 3:45 PM, Daniel Hunsaker < danhunsa...@gmail.com > 
> wrote: 
> 
> 
> 
> 
> What about setting CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS = N? As I understand 
> it, "not set" means "use the default setting the kernel devs think is the 
> most sane for normal use", which in this case would be Y... 
> 
> On Wed, Sep 9, 2015, 06:58 Alexandre DERUMIER < aderum...@odiso.com > wrote: 
> 
>  
>>>I could set it in kernel config, for example: 
>>> 
>>>CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y 
>>> 
>>>would that help? 
> 
> Well, this is like "echo madvise > 
> /sys/kernel/mm/transparent_hugepage/enabled" 
> 
> But I don't see how to set never. 
> 
> maybe 
> 
> # CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS is not set 
> # CONFIG_TRANSPARENT_HUGEPAGE_MADVISE is not set 
> 
> 
> But I don't known how to unset CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS in the new 
> pve-kernel package makefile ? 
> 
> 
> I have tried to add 
> 
> PVE_CONFIG_OPTS= \ 
> .. 
> -d CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS \ 
> .. 
> 
> but it don't seem to work, I still have CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS=Y 
> in .config 
> 
> Any idea ? 
> 
> 
> 
> - Mail original - 
> De: "dietmar" < diet...@proxmox.com > 
> À: "aderumier" < aderum...@odiso.com >, "pve-devel" < 
> pve-devel@pve.proxmox.com > 
> Envoyé: Mercredi 9 Septembre 2015 06:33:05 
> Objet: Re: [pve-devel] disabling transparent hugepage by default like kernel 
> 2.6.32 ? 
> 
>> I think it's better to disable it by default. 
>> 
>> echo never > /sys/kernel/mm/transparent_hugepage/enabled. 
>> 
>> (I don't known what is the best way to disable it by default) 
> 
> I could set it in kernel config, for example: 
> 
> CONFIG_TRANSPARENT_HUGEPAGE_MADVISE=y 
> 
> would that help? 
> ___ 
> pve-devel mailing list 
> pve-devel@pve.proxmox.com 
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel 
> 
> 
> 
> 
> ___ 
> pve-devel mailing list 
> pve-devel@pve.proxmox.com 
>

Re: [pve-devel] Kernel crash with 4.1 kernel

2015-09-09 Thread Stefan Priebe 


Am 09.09.2015 um 17:32 schrieb Dietmar Maurer:

I was rebooting one node and all other nodes crashed with the following
output:
dlm: closing connection to node 5


May I ask why do you use dlm at all?



Hi,

this is PVE 3.4 on wheezy with the current 4.1 kernel. I've not changed 
explicitly anything regarding dlm. Isn't it the default on 3.4?


Stefan

___
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel