astlinux-devel Mailing List for AstLinux

Oh wow, now I am embarrassed for my ignorance. Well we all started somewhere!

uname -m was added to the web interface which I did not revert.

Sorry I wasted your time.

Regards
Michael Knill






On 26/03/2016, 11:30 PM, "Lonnie Abelbeck" <li...@lo...> wrote:


On Mar 26, 2016, at 2:05 AM, Michael Knill <mic...@ip...> wrote:

> Sorry all but I have broken my build system (my fault) but I am having problems fixing it.
> 
> Basically I have my own repository but I accidentally put a 64 bit image in the 32bit directory and upgraded my build system (1.2.4.1 to 1.2.6).
> It seemed to upgrade fine but I decided I wanted to revert but it still shows a 64bit Kernel (astlinux-1.2.4.1 i686 - Asterisk 11.19.0 as the Astlinux Release).

"i686" is 32-bit

"x86_64" is 64-bit

from the "uname -m" command.

Looks like your revert worked as expected.

Lonnie


------------------------------------------------------------------------------
Transform Data into Opportunity.
Accelerate data analysis in your applications with
Intel Data Analytics Acceleration Library.
Click to learn more.
http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140
_______________________________________________
Astlinux-devel mailing list
Ast...@li...
https://lists.sourceforge.net/lists/listinfo/astlinux-devel

Donations to support AstLinux are graciously accepted via PayPal to pa...@kr....

On Mar 26, 2016, at 2:05 AM, Michael Knill <mic...@ip...> wrote:

> Sorry all but I have broken my build system (my fault) but I am having problems fixing it.
> 
> Basically I have my own repository but I accidentally put a 64 bit image in the 32bit directory and upgraded my build system (1.2.4.1 to 1.2.6).
> It seemed to upgrade fine but I decided I wanted to revert but it still shows a 64bit Kernel (astlinux-1.2.4.1 i686 - Asterisk 11.19.0 as the Astlinux Release).

"i686" is 32-bit

"x86_64" is 64-bit

from the "uname -m" command.

Looks like your revert worked as expected.

Lonnie

Sorry all but I have broken my build system (my fault) but I am having problems fixing it.

Basically I have my own repository but I accidentally put a 64 bit image in the 32bit directory and upgraded my build system (1.2.4.1 to 1.2.6).
It seemed to upgrade fine but I decided I wanted to revert but it still shows a 64bit Kernel (astlinux-1.2.4.1 i686 - Asterisk 11.19.0 as the Astlinux Release).

Yes really stupid but how would I go about fixing it? Do I need to rebuild?

Regards
Michael Knill

On Mar 21, 2016, at 1:31 PM, Michael Knill <mic...@ip...> wrote:

> Yes thanks Lonnie.
> So I assume that the Build Engine works because it is not using Centos 7?

Yes.  Just to be clear the real problem is the rp-pppoe package and cross-compiling.

Lonnie



> I am thinking that 1.2.6 should be removed or at least patched from mirror.astlinux.org as this is really bad if it is used for a remote upgrade as I do.
> 
> Regards
> Michael Knill
> 
> 
> On 22/03/2016, 2:17 AM, "Lonnie Abelbeck" <li...@lo...> wrote:
> 
> Should be fixed in the SVN with r7604:
> https://sourceforge.net/p/astlinux/code/7604
> 
> Lonnie
> 
> 
> On Mar 21, 2016, at 6:50 AM, Lonnie Abelbeck <li...@lo...> wrote:
> 
>> Michael,
>> 
>> I see the problem, it is caused when building on CentOS 7 as the official builds are now.
>> 
>> Quick answer, if you do the following commands from the CLI, that should fix it (I have no way to test PPPoE)...
>> --
>> ln -s /usr/bin/id /bin/id
>> ln -s /usr/bin/setsid /bin/setsid
>> ln -s /usr/sbin/pppd /sbin/pppd
>> --
>> Note: Also remove /oldroot/mnt/asturw/usr/sbin/pppoe-start if you edited that file from another email.
>> 
>> Though I would only do that in a pinch, not a solution for all your PPPoE customers, as cleaning-up later is a pain.
>> 
>> Another solution is to use the build engine for 1.2.6 : http://build.astlinux.org
>> The standard 1.2.6 builds are already built.
>> 
>> We will have a solution for 1.2.7 .
>> 
>> Lonnie
>> 
>> 
>> On Mar 21, 2016, at 5:30 AM, Michael Knill <mic...@ip...> wrote:
>> 
>>> Im so glad I tested this as it would be REAL BAD to do a remote upgrade e.g. No more connectivity :(
>>> 
>>> My PPPoE is not coming up.
>>> 
>>> 3999-IPCBuild-CM1 kd # pppoe-start
>>> /usr/sbin/pppoe-start: line 44: /bin/id: No such file or directory
>>> pppoe-start: You must be root to run this script
>>> 
>>> Why is it /bin/id when it should be /usr/bin/id ?????
>>> Why has this script changed?
>>> 
>>> Regards
>>> Michael Knill
>>> ------------------------------------------------------------------------------
>>> Transform Data into Opportunity.
>>> Accelerate data analysis in your applications with
>>> Intel Data Analytics Acceleration Library.
>>> Click to learn more.
>>> http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140_______________________________________________
>>> Astlinux-devel mailing list
>>> Ast...@li...
>>> https://lists.sourceforge.net/lists/listinfo/astlinux-devel
>>> 
>>> Donations to support AstLinux are graciously accepted via PayPal to pa...@kr....
>> 
>> 
>> ------------------------------------------------------------------------------
>> Transform Data into Opportunity.
>> Accelerate data analysis in your applications with
>> Intel Data Analytics Acceleration Library.
>> Click to learn more.
>> http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140
>> _______________________________________________
>> Astlinux-devel mailing list
>> Ast...@li...
>> https://lists.sourceforge.net/lists/listinfo/astlinux-devel
>> 
>> Donations to support AstLinux are graciously accepted via PayPal to pa...@kr....
>> 
>> 
> 
> 
> ------------------------------------------------------------------------------
> Transform Data into Opportunity.
> Accelerate data analysis in your applications with
> Intel Data Analytics Acceleration Library.
> Click to learn more.
> http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140
> _______________________________________________
> Astlinux-devel mailing list
> Ast...@li...
> https://lists.sourceforge.net/lists/listinfo/astlinux-devel
> 
> Donations to support AstLinux are graciously accepted via PayPal to pa...@kr....
> 
> ------------------------------------------------------------------------------
> Transform Data into Opportunity.
> Accelerate data analysis in your applications with
> Intel Data Analytics Acceleration Library.
> Click to learn more.
> http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140
> _______________________________________________
> Astlinux-devel mailing list
> Ast...@li...
> https://lists.sourceforge.net/lists/listinfo/astlinux-devel
> 
> Donations to support AstLinux are graciously accepted via PayPal to pa...@kr....
> 
> 

Yes thanks Lonnie.
So I assume that the Build Engine works because it is not using Centos 7?
I am thinking that 1.2.6 should be removed or at least patched from mirror.astlinux.org as this is really bad if it is used for a remote upgrade as I do.

Regards
Michael Knill







On 22/03/2016, 2:17 AM, "Lonnie Abelbeck" <li...@lo...> wrote:

Should be fixed in the SVN with r7604:
https://sourceforge.net/p/astlinux/code/7604

Lonnie


On Mar 21, 2016, at 6:50 AM, Lonnie Abelbeck <li...@lo...> wrote:

> Michael,
> 
> I see the problem, it is caused when building on CentOS 7 as the official builds are now.
> 
> Quick answer, if you do the following commands from the CLI, that should fix it (I have no way to test PPPoE)...
> --
> ln -s /usr/bin/id /bin/id
> ln -s /usr/bin/setsid /bin/setsid
> ln -s /usr/sbin/pppd /sbin/pppd
> --
> Note: Also remove /oldroot/mnt/asturw/usr/sbin/pppoe-start if you edited that file from another email.
> 
> Though I would only do that in a pinch, not a solution for all your PPPoE customers, as cleaning-up later is a pain.
> 
> Another solution is to use the build engine for 1.2.6 : http://build.astlinux.org
> The standard 1.2.6 builds are already built.
> 
> We will have a solution for 1.2.7 .
> 
> Lonnie
> 
> 
> On Mar 21, 2016, at 5:30 AM, Michael Knill <mic...@ip...> wrote:
> 
>> Im so glad I tested this as it would be REAL BAD to do a remote upgrade e.g. No more connectivity :(
>> 
>> My PPPoE is not coming up.
>> 
>> 3999-IPCBuild-CM1 kd # pppoe-start
>> /usr/sbin/pppoe-start: line 44: /bin/id: No such file or directory
>> pppoe-start: You must be root to run this script
>> 
>> Why is it /bin/id when it should be /usr/bin/id ?????
>> Why has this script changed?
>> 
>> Regards
>> Michael Knill
>> ------------------------------------------------------------------------------
>> Transform Data into Opportunity.
>> Accelerate data analysis in your applications with
>> Intel Data Analytics Acceleration Library.
>> Click to learn more.
>> http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140_______________________________________________
>> Astlinux-devel mailing list
>> Ast...@li...
>> https://lists.sourceforge.net/lists/listinfo/astlinux-devel
>> 
>> Donations to support AstLinux are graciously accepted via PayPal to pa...@kr....
> 
> 
> ------------------------------------------------------------------------------
> Transform Data into Opportunity.
> Accelerate data analysis in your applications with
> Intel Data Analytics Acceleration Library.
> Click to learn more.
> http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140
> _______________________________________________
> Astlinux-devel mailing list
> Ast...@li...
> https://lists.sourceforge.net/lists/listinfo/astlinux-devel
> 
> Donations to support AstLinux are graciously accepted via PayPal to pa...@kr....
> 
> 


------------------------------------------------------------------------------
Transform Data into Opportunity.
Accelerate data analysis in your applications with
Intel Data Analytics Acceleration Library.
Click to learn more.
http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140
_______________________________________________
Astlinux-devel mailing list
Ast...@li...
https://lists.sourceforge.net/lists/listinfo/astlinux-devel

Donations to support AstLinux are graciously accepted via PayPal to pa...@kr....

Should be fixed in the SVN with r7604:
https://sourceforge.net/p/astlinux/code/7604

Lonnie


On Mar 21, 2016, at 6:50 AM, Lonnie Abelbeck <li...@lo...> wrote:

> Michael,
> 
> I see the problem, it is caused when building on CentOS 7 as the official builds are now.
> 
> Quick answer, if you do the following commands from the CLI, that should fix it (I have no way to test PPPoE)...
> --
> ln -s /usr/bin/id /bin/id
> ln -s /usr/bin/setsid /bin/setsid
> ln -s /usr/sbin/pppd /sbin/pppd
> --
> Note: Also remove /oldroot/mnt/asturw/usr/sbin/pppoe-start if you edited that file from another email.
> 
> Though I would only do that in a pinch, not a solution for all your PPPoE customers, as cleaning-up later is a pain.
> 
> Another solution is to use the build engine for 1.2.6 : http://build.astlinux.org
> The standard 1.2.6 builds are already built.
> 
> We will have a solution for 1.2.7 .
> 
> Lonnie
> 
> 
> On Mar 21, 2016, at 5:30 AM, Michael Knill <mic...@ip...> wrote:
> 
>> Im so glad I tested this as it would be REAL BAD to do a remote upgrade e.g. No more connectivity :(
>> 
>> My PPPoE is not coming up.
>> 
>> 3999-IPCBuild-CM1 kd # pppoe-start
>> /usr/sbin/pppoe-start: line 44: /bin/id: No such file or directory
>> pppoe-start: You must be root to run this script
>> 
>> Why is it /bin/id when it should be /usr/bin/id ?????
>> Why has this script changed?
>> 
>> Regards
>> Michael Knill
>> ------------------------------------------------------------------------------
>> Transform Data into Opportunity.
>> Accelerate data analysis in your applications with
>> Intel Data Analytics Acceleration Library.
>> Click to learn more.
>> http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140_______________________________________________
>> Astlinux-devel mailing list
>> Ast...@li...
>> https://lists.sourceforge.net/lists/listinfo/astlinux-devel
>> 
>> Donations to support AstLinux are graciously accepted via PayPal to pa...@kr....
> 
> 
> ------------------------------------------------------------------------------
> Transform Data into Opportunity.
> Accelerate data analysis in your applications with
> Intel Data Analytics Acceleration Library.
> Click to learn more.
> http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140
> _______________________________________________
> Astlinux-devel mailing list
> Ast...@li...
> https://lists.sourceforge.net/lists/listinfo/astlinux-devel
> 
> Donations to support AstLinux are graciously accepted via PayPal to pa...@kr....
> 
> 

Am 21.03.2016 um 12:50 schrieb Lonnie Abelbeck <li...@lo...>:

> Michael,
> 
> I see the problem, it is caused when building on CentOS 7 as the official builds are now.
> 
> Quick answer, if you do the following commands from the CLI, that should fix it (I have no way to test PPPoE)...
> --
> ln -s /usr/bin/id /bin/id
> ln -s /usr/bin/setsid /bin/setsid
> ln -s /usr/sbin/pppd /sbin/pppd
> --
> Note: Also remove /oldroot/mnt/asturw/usr/sbin/pppoe-start if you edited that file from another email.
> 
> Though I would only do that in a pinch, not a solution for all your PPPoE customers, as cleaning-up later is a pain.
> 
> Another solution is to use the build engine for 1.2.6 : http://build.astlinux.org
> The standard 1.2.6 builds are already built.
> 
> We will have a solution for 1.2.7 .
> 
> Lonnie

Well then it makes sense, as I installed nearly all boxes from our build page.

> On Mar 21, 2016, at 5:30 AM, Michael Knill <mic...@ip...> wrote:
> 
>> Im so glad I tested this as it would be REAL BAD to do a remote upgrade e.g. No more connectivity :(
>> 
>> My PPPoE is not coming up.
>> 
>> 3999-IPCBuild-CM1 kd # pppoe-start
>> /usr/sbin/pppoe-start: line 44: /bin/id: No such file or directory
>> pppoe-start: You must be root to run this script
>> 
>> Why is it /bin/id when it should be /usr/bin/id ?????
>> Why has this script changed?
>> 
>> Regards
>> Michael Knill

Michael

http://www.mksolutions.info

Michael,

I see the problem, it is caused when building on CentOS 7 as the official builds are now.

Quick answer, if you do the following commands from the CLI, that should fix it (I have no way to test PPPoE)...
--
ln -s /usr/bin/id /bin/id
ln -s /usr/bin/setsid /bin/setsid
ln -s /usr/sbin/pppd /sbin/pppd
--
Note: Also remove /oldroot/mnt/asturw/usr/sbin/pppoe-start if you edited that file from another email.

Though I would only do that in a pinch, not a solution for all your PPPoE customers, as cleaning-up later is a pain.

Another solution is to use the build engine for 1.2.6 : http://build.astlinux.org
The standard 1.2.6 builds are already built.

We will have a solution for 1.2.7 .

Lonnie


On Mar 21, 2016, at 5:30 AM, Michael Knill <mic...@ip...> wrote:

> Im so glad I tested this as it would be REAL BAD to do a remote upgrade e.g. No more connectivity :(
> 
> My PPPoE is not coming up.
> 
> 3999-IPCBuild-CM1 kd # pppoe-start
> /usr/sbin/pppoe-start: line 44: /bin/id: No such file or directory
> pppoe-start: You must be root to run this script
> 
> Why is it /bin/id when it should be /usr/bin/id ?????
> Why has this script changed?
> 
> Regards
> Michael Knill
> ------------------------------------------------------------------------------
> Transform Data into Opportunity.
> Accelerate data analysis in your applications with
> Intel Data Analytics Acceleration Library.
> Click to learn more.
> http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140_______________________________________________
> Astlinux-devel mailing list
> Ast...@li...
> https://lists.sourceforge.net/lists/listinfo/astlinux-devel
> 
> Donations to support AstLinux are graciously accepted via PayPal to pa...@kr....

Am 21.03.2016 um 11:57 schrieb Michael Knill <mic...@ip...>:

> Hmm I can’t see an upgrade in the release notes.
> Even though I changed the script, it still didn’t work (TIMEOUT) but I didn’t get the error any more.
>  
> Regards
> Michael Knill
> 
> From: Michael Knill <mic...@ip...>
> Reply-To: AstLinux Developers Mailing List <ast...@li...>
> Date: Monday, 21 March 2016 at 9:30 PM
> To: AstLinux Developers Mailing List <ast...@li...>
> Subject: [Astlinux-devel] URGENT. PPPoE is broken in 1.2.6
> 
> Im so glad I tested this as it would be REAL BAD to do a remote upgrade e.g. No more connectivity :(
> 
> My PPPoE is not coming up.
> 
> 3999-IPCBuild-CM1 kd # pppoe-start
> /usr/sbin/pppoe-start: line 44: /bin/id: No such file or directory
> pppoe-start: You must be root to run this script
> 
> Why is it /bin/id when it should be /usr/bin/id ?????
> Why has this script changed?
> 
> Regards
> Michael Knill

Hi Michael,

I have several boxes with PPPoE running fine with AstLinux 1.2.6.
I cannot remember, that we had changes to PPPoE in the last time.

Also my "/usr/sbin/pppoe-start" reads in line 44 (as the original in "output/build/rp-pppoe-3.11/scripts/"):

if [ "`/usr/bin/id -u`" != 0 ] ; then

Run "show-union" to find out, if you maybe had changed pppoe-start yourself before.

Michael

http://www.mksolutions.info

Hmm I can’t see an upgrade in the release notes.
Even though I changed the script, it still didn’t work (TIMEOUT) but I didn’t get the error any more.

Regards
Michael Knill

From: Michael Knill <mic...@ip...<mailto:mic...@ip...>>
Reply-To: AstLinux Developers Mailing List <ast...@li...<mailto:ast...@li...>>
Date: Monday, 21 March 2016 at 9:30 PM
To: AstLinux Developers Mailing List <ast...@li...<mailto:ast...@li...>>
Subject: [Astlinux-devel] URGENT. PPPoE is broken in 1.2.6

Im so glad I tested this as it would be REAL BAD to do a remote upgrade e.g. No more connectivity :(

My PPPoE is not coming up.

3999-IPCBuild-CM1 kd # pppoe-start
/usr/sbin/pppoe-start: line 44: /bin/id: No such file or directory
pppoe-start: You must be root to run this script

Why is it /bin/id when it should be /usr/bin/id ?????
Why has this script changed?

Regards
Michael Knill

Im so glad I tested this as it would be REAL BAD to do a remote upgrade e.g. No more connectivity :(

My PPPoE is not coming up.

3999-IPCBuild-CM1 kd # pppoe-start
/usr/sbin/pppoe-start: line 44: /bin/id: No such file or directory
pppoe-start: You must be root to run this script

Why is it /bin/id when it should be /usr/bin/id ?????
Why has this script changed?

Regards
Michael Knill

Hi Devs

Thanks for the new XMPP Staff Tab. Unfortunately it does not have the ability to edit the Shared Groups which customers want to do when they add a new user.
Its a bit hard because the staff user does not have access to the Edit Tab.

Any ideas on how to implement this for staff?

Regards
Michael Knill

Hi Devs,

FYI, with the advent of the "Let's Encrypt" project https://letsencrypt.org , the team thought it is worth adding some missing HTTPS support.

In particular we use the Busybox 'wget' for various downloads which does not support HTTPS, so we need to switch to 'curl' for those downloads.

It made me wonder how such a change would effect our current lowest-end supported board, the net5501/alix.

I just finished some testing worth sharing, downloading the same AstLinux run image...
--
(Maximum percentage CPU using 1 of 1 core)
net5501 curl HTTPS 26 %
net5501 curl HTTP  7 %
net5501 Busybox wget HTTP 3.5 %

Approximate same  650 KB/sec (downloaded to /dev/null) for each.
--

--
(Maximum percentage CPU using 1 of 4 cores)
Jetway NF9HG-2930 curl HTTPS 28 %
Jetway NF9HG-2930 curl HTTP  12 %
Jetway NF9HG-2930 Busybox wget HTTP 8 %

Approximate same 4050 KB/sec (downloaded to /dev/null) for each.
--

Good to know that the net5501 can still handle HTTPS with no increase in download time.


Lonnie

Devs,

With Revision 7576, a minor Linux version bump from linux-3.2.66 to linux-3.2.78 .  Many bug fixes over the last year or so.

No reason to rebuild the toolchain, the kernel headers is all the toolchain cares about.

BTW, the kernel URL is referenced in any custom config files you may have, so they will need to be tweaked.

Lonnie

Agreed.

As long as old build versions and upgrades are available for a little while.

Regards
Michael Knill







On 6/03/2016, 5:28 AM, "Michael Keuter" <li...@mk...> wrote:


Am 05.03.2016 um 19:21 schrieb Lonnie Abelbeck <li...@lo...>:

> On Mar 5, 2016, at 5:30 AM, Michael Keuter <li...@mk...> wrote:
> 
>> Am 02.03.2016 um 18:26 schrieb Lonnie Abelbeck <li...@lo...>:
>> 
>>> Devs,
>>> 
>>> Michael Keuter has been (indirectly) working with the new "beroNet Telephony Appliance 2.0" with the AstLinux ISO:
>>> http://www.beronet.com/products/telephony-appliance/
>>> 
>>> For some unknown reason beroNet is using Xen for their hypervisor, as such AstLinux genx86_64 runs but can't find an emulated network driver.
>>> 
>>> This may be due to issues in the beroNet hypervisor config, time will tell as it is very new:
>>> https://groups.google.com/forum/#!topic/berofix-users/9atxZ_rvIFE
>>> 
>>> 
>>> While VMware is excellent at emulating x86 systems without any guest changes, KVM and Xen hypervisors can benefit (or require) from Linux kernel tweaks such as "CONFIG_PARAVIRT=y", "CONFIG_KVM_GUEST=y" and a new set of Xen drivers "CONFIG_XEN=y" and possibly also "CONFIG_VIRTIO_NET=y".
>>> 
>>> While distros like CentOS 7 include such guest hypervisor tweaks in their minimal distribution...
>>> --
>>> # egrep '(CONFIG_PARAVIRT|CONFIG_KVM_GUEST|CONFIG_XEN|CONFIG_VIRTIO_NET)=' /boot/config-3.10.0-229.el7.x86_64 
>>> CONFIG_PARAVIRT=y
>>> CONFIG_XEN=y
>>> CONFIG_KVM_GUEST=y
>>> CONFIG_VIRTIO_NET=m
>>> --
>>> There may be some performance cost with "CONFIG_PARAVIRT=y" with bare metal installs without a hypervisor.
>>> 
>>> Also, with AstLinux's firmware based upgrade model, small footprint, built-in firewall, etc., it makes it a great choice for Virtual Machine installs, particularly now with x86_64 support and an ISO installer.
>>> 
>>> So, with that being said does it make sense to add a "genx86_64-vm" board type designed for VM's ?  That would entail adding a "project/astlinux/genx86_64/linux-smp-vm.config" kernel config customized for VM installs.
>>> 
>>> The thought of maintaining yet another kernel config is a drawback.
>>> 
>>> For simplicity, we can use the same "x86_64-configs/*.config" files and 'sed -i' the kernel config when building as we do for enabling SMP with geni586 configs.
>>> 
>>> Thoughts ?
>>> 
>>> Lonnie
>> 
>> I personally would like the idea of having a separate VM config.
>> Since we don't have so many kernel changes over the time, the extra "linux-smp-vm.config" kernel config would be more a one-time job.
>> The main question is: how can this be tested?
>> To do that we would need more interested users/devs with virtualization servers.
>> 
>> Update: Beronet just updated its hypervisor and included a E1000 emulation:
>> http://wiki.beronet.com/index.php/BeroNet_Hypervisor_Changelog
>> 
>> Michael
> 
> While CentOS 7 has CONFIG_PARAVIRT=y, CONFIG_PARAVIRT_SPINLOCKS=y, etc. set, these come at a performance penalty for native bare-metal, it does seem a board type specific for a guest virtual machine makes sense for us.
> 
> The timing is good as we can now focus on only 64-bit genx86_64 for VM guest applications.
> 
> Additionally, we could stop officially building alix and net5501 ISO installer images at any time, (though still provide firmware run images for upgrades for awhile longer).

Yes, ISO installers for Alix and net5501 are definitely not needed.

> As Michael stated, testing is needed to tune a "genx86_64-vm" board type for the various hypervisors out there.  Possibly some cloud testing like with linode.com would be in order if anyone is interested.
> 
> Lonnie

Michael

http://www.mksolutions.info




------------------------------------------------------------------------------
_______________________________________________
Astlinux-devel mailing list
Ast...@li...
https://lists.sourceforge.net/lists/listinfo/astlinux-devel

Donations to support AstLinux are graciously accepted via PayPal to pa...@kr....

Am 05.03.2016 um 19:21 schrieb Lonnie Abelbeck <li...@lo...>:

> On Mar 5, 2016, at 5:30 AM, Michael Keuter <li...@mk...> wrote:
> 
>> Am 02.03.2016 um 18:26 schrieb Lonnie Abelbeck <li...@lo...>:
>> 
>>> Devs,
>>> 
>>> Michael Keuter has been (indirectly) working with the new "beroNet Telephony Appliance 2.0" with the AstLinux ISO:
>>> http://www.beronet.com/products/telephony-appliance/
>>> 
>>> For some unknown reason beroNet is using Xen for their hypervisor, as such AstLinux genx86_64 runs but can't find an emulated network driver.
>>> 
>>> This may be due to issues in the beroNet hypervisor config, time will tell as it is very new:
>>> https://groups.google.com/forum/#!topic/berofix-users/9atxZ_rvIFE
>>> 
>>> 
>>> While VMware is excellent at emulating x86 systems without any guest changes, KVM and Xen hypervisors can benefit (or require) from Linux kernel tweaks such as "CONFIG_PARAVIRT=y", "CONFIG_KVM_GUEST=y" and a new set of Xen drivers "CONFIG_XEN=y" and possibly also "CONFIG_VIRTIO_NET=y".
>>> 
>>> While distros like CentOS 7 include such guest hypervisor tweaks in their minimal distribution...
>>> --
>>> # egrep '(CONFIG_PARAVIRT|CONFIG_KVM_GUEST|CONFIG_XEN|CONFIG_VIRTIO_NET)=' /boot/config-3.10.0-229.el7.x86_64 
>>> CONFIG_PARAVIRT=y
>>> CONFIG_XEN=y
>>> CONFIG_KVM_GUEST=y
>>> CONFIG_VIRTIO_NET=m
>>> --
>>> There may be some performance cost with "CONFIG_PARAVIRT=y" with bare metal installs without a hypervisor.
>>> 
>>> Also, with AstLinux's firmware based upgrade model, small footprint, built-in firewall, etc., it makes it a great choice for Virtual Machine installs, particularly now with x86_64 support and an ISO installer.
>>> 
>>> So, with that being said does it make sense to add a "genx86_64-vm" board type designed for VM's ?  That would entail adding a "project/astlinux/genx86_64/linux-smp-vm.config" kernel config customized for VM installs.
>>> 
>>> The thought of maintaining yet another kernel config is a drawback.
>>> 
>>> For simplicity, we can use the same "x86_64-configs/*.config" files and 'sed -i' the kernel config when building as we do for enabling SMP with geni586 configs.
>>> 
>>> Thoughts ?
>>> 
>>> Lonnie
>> 
>> I personally would like the idea of having a separate VM config.
>> Since we don't have so many kernel changes over the time, the extra "linux-smp-vm.config" kernel config would be more a one-time job.
>> The main question is: how can this be tested?
>> To do that we would need more interested users/devs with virtualization servers.
>> 
>> Update: Beronet just updated its hypervisor and included a E1000 emulation:
>> http://wiki.beronet.com/index.php/BeroNet_Hypervisor_Changelog
>> 
>> Michael
> 
> While CentOS 7 has CONFIG_PARAVIRT=y, CONFIG_PARAVIRT_SPINLOCKS=y, etc. set, these come at a performance penalty for native bare-metal, it does seem a board type specific for a guest virtual machine makes sense for us.
> 
> The timing is good as we can now focus on only 64-bit genx86_64 for VM guest applications.
> 
> Additionally, we could stop officially building alix and net5501 ISO installer images at any time, (though still provide firmware run images for upgrades for awhile longer).

Yes, ISO installers for Alix and net5501 are definitely not needed.

> As Michael stated, testing is needed to tune a "genx86_64-vm" board type for the various hypervisors out there.  Possibly some cloud testing like with linode.com would be in order if anyone is interested.
> 
> Lonnie

Michael

http://www.mksolutions.info

On Mar 5, 2016, at 5:30 AM, Michael Keuter <li...@mk...> wrote:

> 
> Am 02.03.2016 um 18:26 schrieb Lonnie Abelbeck <li...@lo...>:
> 
>> Devs,
>> 
>> Michael Keuter has been (indirectly) working with the new "beroNet Telephony Appliance 2.0" with the AstLinux ISO:
>> http://www.beronet.com/products/telephony-appliance/
>> 
>> For some unknown reason beroNet is using Xen for their hypervisor, as such AstLinux genx86_64 runs but can't find an emulated network driver.
>> 
>> This may be due to issues in the beroNet hypervisor config, time will tell as it is very new:
>> https://groups.google.com/forum/#!topic/berofix-users/9atxZ_rvIFE
>> 
>> 
>> While VMware is excellent at emulating x86 systems without any guest changes, KVM and Xen hypervisors can benefit (or require) from Linux kernel tweaks such as "CONFIG_PARAVIRT=y", "CONFIG_KVM_GUEST=y" and a new set of Xen drivers "CONFIG_XEN=y" and possibly also "CONFIG_VIRTIO_NET=y".
>> 
>> While distros like CentOS 7 include such guest hypervisor tweaks in their minimal distribution...
>> --
>> # egrep '(CONFIG_PARAVIRT|CONFIG_KVM_GUEST|CONFIG_XEN|CONFIG_VIRTIO_NET)=' /boot/config-3.10.0-229.el7.x86_64 
>> CONFIG_PARAVIRT=y
>> CONFIG_XEN=y
>> CONFIG_KVM_GUEST=y
>> CONFIG_VIRTIO_NET=m
>> --
>> There may be some performance cost with "CONFIG_PARAVIRT=y" with bare metal installs without a hypervisor.
>> 
>> Also, with AstLinux's firmware based upgrade model, small footprint, built-in firewall, etc., it makes it a great choice for Virtual Machine installs, particularly now with x86_64 support and an ISO installer.
>> 
>> So, with that being said does it make sense to add a "genx86_64-vm" board type designed for VM's ?  That would entail adding a "project/astlinux/genx86_64/linux-smp-vm.config" kernel config customized for VM installs.
>> 
>> The thought of maintaining yet another kernel config is a drawback.
>> 
>> For simplicity, we can use the same "x86_64-configs/*.config" files and 'sed -i' the kernel config when building as we do for enabling SMP with geni586 configs.
>> 
>> Thoughts ?
>> 
>> Lonnie
> 
> I personally would like the idea of having a separate VM config.
> Since we don't have so many kernel changes over the time, the extra "linux-smp-vm.config" kernel config would be more a one-time job.
> The main question is: how can this be tested?
> To do that we would need more interested users/devs with virtualization servers.
> 
> Update: Beronet just updated its hypervisor and included a E1000 emulation:
> http://wiki.beronet.com/index.php/BeroNet_Hypervisor_Changelog
> 
> Michael

While CentOS 7 has CONFIG_PARAVIRT=y, CONFIG_PARAVIRT_SPINLOCKS=y, etc. set, these come at a performance penalty for native bare-metal, it does seem a board type specific for a guest virtual machine makes sense for us.

The timing is good as we can now focus on only 64-bit genx86_64 for VM guest applications.

Additionally, we could stop officially building alix and net5501 ISO installer images at any time, (though still provide firmware run images for upgrades for awhile longer).

As Michael stated, testing is needed to tune a "genx86_64-vm" board type for the various hypervisors out there.  Possibly some cloud testing like with linode.com would be in order if anyone is interested.

Lonnie

Am 02.03.2016 um 18:26 schrieb Lonnie Abelbeck <li...@lo...>:

> Devs,
> 
> Michael Keuter has been (indirectly) working with the new "beroNet Telephony Appliance 2.0" with the AstLinux ISO:
> http://www.beronet.com/products/telephony-appliance/
> 
> For some unknown reason beroNet is using Xen for their hypervisor, as such AstLinux genx86_64 runs but can't find an emulated network driver.
> 
> This may be due to issues in the beroNet hypervisor config, time will tell as it is very new:
> https://groups.google.com/forum/#!topic/berofix-users/9atxZ_rvIFE
> 
> 
> While VMware is excellent at emulating x86 systems without any guest changes, KVM and Xen hypervisors can benefit (or require) from Linux kernel tweaks such as "CONFIG_PARAVIRT=y", "CONFIG_KVM_GUEST=y" and a new set of Xen drivers "CONFIG_XEN=y" and possibly also "CONFIG_VIRTIO_NET=y".
> 
> While distros like CentOS 7 include such guest hypervisor tweaks in their minimal distribution...
> --
> # egrep '(CONFIG_PARAVIRT|CONFIG_KVM_GUEST|CONFIG_XEN|CONFIG_VIRTIO_NET)=' /boot/config-3.10.0-229.el7.x86_64 
> CONFIG_PARAVIRT=y
> CONFIG_XEN=y
> CONFIG_KVM_GUEST=y
> CONFIG_VIRTIO_NET=m
> --
> There may be some performance cost with "CONFIG_PARAVIRT=y" with bare metal installs without a hypervisor.
> 
> Also, with AstLinux's firmware based upgrade model, small footprint, built-in firewall, etc., it makes it a great choice for Virtual Machine installs, particularly now with x86_64 support and an ISO installer.
> 
> So, with that being said does it make sense to add a "genx86_64-vm" board type designed for VM's ?  That would entail adding a "project/astlinux/genx86_64/linux-smp-vm.config" kernel config customized for VM installs.
> 
> The thought of maintaining yet another kernel config is a drawback.
> 
> For simplicity, we can use the same "x86_64-configs/*.config" files and 'sed -i' the kernel config when building as we do for enabling SMP with geni586 configs.
> 
> Thoughts ?
> 
> Lonnie

I personally would like the idea of having a separate VM config.
Since we don't have so many kernel changes over the time, the extra "linux-smp-vm.config" kernel config would be more a one-time job.
The main question is: how can this be tested?
To do that we would need more interested users/devs with virtualization servers.

Update: Beronet just updated its hypervisor and included a E1000 emulation:
http://wiki.beronet.com/index.php/BeroNet_Hypervisor_Changelog

Michael

http://www.mksolutions.info

Devs,

Michael Keuter has been (indirectly) working with the new "beroNet Telephony Appliance 2.0" with the AstLinux ISO:
http://www.beronet.com/products/telephony-appliance/

For some unknown reason beroNet is using Xen for their hypervisor, as such AstLinux genx86_64 runs but can't find an emulated network driver.

This may be due to issues in the beroNet hypervisor config, time will tell as it is very new:
https://groups.google.com/forum/#!topic/berofix-users/9atxZ_rvIFE


While VMware is excellent at emulating x86 systems without any guest changes, KVM and Xen hypervisors can benefit (or require) from Linux kernel tweaks such as "CONFIG_PARAVIRT=y", "CONFIG_KVM_GUEST=y" and a new set of Xen drivers "CONFIG_XEN=y" and possibly also "CONFIG_VIRTIO_NET=y".

While distros like CentOS 7 include such guest hypervisor tweaks in their minimal distribution...
--
# egrep '(CONFIG_PARAVIRT|CONFIG_KVM_GUEST|CONFIG_XEN|CONFIG_VIRTIO_NET)=' /boot/config-3.10.0-229.el7.x86_64 
CONFIG_PARAVIRT=y
CONFIG_XEN=y
CONFIG_KVM_GUEST=y
CONFIG_VIRTIO_NET=m
--
There may be some performance cost with "CONFIG_PARAVIRT=y" with bare metal installs without a hypervisor.

Also, with AstLinux's firmware based upgrade model, small footprint, built-in firewall, etc., it makes it a great choice for Virtual Machine installs, particularly now with x86_64 support and an ISO installer.

So, with that being said does it make sense to add a "genx86_64-vm" board type designed for VM's ?  That would entail adding a "project/astlinux/genx86_64/linux-smp-vm.config" kernel config customized for VM installs.

The thought of maintaining yet another kernel config is a drawback.

For simplicity, we can use the same "x86_64-configs/*.config" files and 'sed -i' the kernel config when building as we do for enabling SMP with geni586 configs.

Thoughts ?

Lonnie